Scott Hanselman

2025-05-02 · open the original ↗

• DotnetSnes compiles C# to SNES ROMs by transpiling .NET IL to C. Shapiro, a staff video architect at GameChanger, built DNTC, a transpiler that walks a dependency graph from a root function and turns the IL stack machine into C. Inspired by Nim compiling to C, the aim is to run C# anywhere a C compiler exists, without the .NET team building ahead-of-time compilation for every platform. The C output then goes through an existing SNES C toolchain to the console's 65816 CPU.
• Hardware without dynamic memory allocation suits the transpiler for now. It does not yet support reference types because there is no garbage collector, which he named as his next task. That fitted eBPF, which he said Windows is also moving toward after the CrowdStrike outage, and he has proof-of-concept C# eBPF programs running in the Linux kernel. It also fitted the SNES, which has no dynamic allocation either.
• Attributes let C# and native C interoperate at zero cost. Calls work in both directions. Attributes for custom function names and native functions and globals map C# onto the C SDK's expectations, such as a lowercase main and assembly labels at fixed memory locations. Because everything links into one binary, native calls carry no overhead, which Hanselman compared to P/Invoke.
• The project stands on community libraries. It builds on Mono.Cecil for reading IL and on PVSnesLib, a C SDK for the SNES that grew from an older one. Because PVSnesLib targets non-C99-compliant C, the transpiler takes plugins to adjust its output. The SDK's support for Tiled maps gives collision detection, and the Mario sample is a near line-by-line port while the Breakout sample is more idiomatic C#.
• It runs easily today and has already run on real hardware. Hanselman reported building the Mario ROM in WSL in about 10 minutes, after updating submodules recursively and using SSH keys. Shapiro said someone ran the Breakout game on a real Super Nintendo from a flash cart. His next aim is a basic PlayStation 1 C# engine, after a Pikuma course.

2025-04-30 · open the original ↗

• The one-shot fractal shader was partly luck. Hanselman puts the hand-written version at four to six hours, against under 15 minutes vibe coding; Mark says he gave it about an 80% chance with the example file supplied. The first attempt on Claude 3.7 thinking failed, and Mark argues that because output is non-deterministic, the success might have been a 10% draw, which is why some vibe coders run a prompt several times and pick the best result.
• Bigger context windows don't fix work on large codebases. Mark cites needle-in-a-haystack tests in Microsoft's Eureka benchmarks, where frontier models fail to connect two related facts placed far apart at 32k or even 8k tokens, far below the million- or 128,000-token limits. He also cites the "Lost in the Middle" paper, which found that models tend to ignore information in the middle of a long context.
• Vibe-coded v1 becomes an inherited codebase you must maintain. Mark says models struggle with brownfield changes: they lose track of state, forget to update functions, drop unrelated functionality while refactoring, and claim fixes they did not make. Hanselman calls the result "a crappy code base by a mediocre programmer", and at that point you may still have to spend the four to six hours learning HLSL.
• Enterprises shouldn't vibe code, because humans stay accountable. Mark's examples are unvetted packages that may carry vulnerabilities or bloat, models defaulting to a deprecated version of the OpenAI API because of training cutoffs, and specs that are only about 80% written up front. He concludes that, given transformer context limits, AI will not take over and experts must own the code, though agents that can check their own output will help.
• Code citations, licensing and indemnity matter in practice. Hanselman credits GitHub Copilot with flagging the Stack Overflow and GitHub sources it drew on, along with their licenses, which let them credit the authors in a README; Mark adds that Microsoft indemnifies Copilot users in copyright disputes. Separately, Hanselman dismisses Google's "25% of our code is AI generated" figure, since lines of code is "a lie" and the number may count autocompletes or unaccepted code.

2025-04-24 · open the original ↗

• Never-ending support is a drop-in package feed, not a migration. Barnes, open source partner manager at HeroDevs, says customers get a token from the customer portal and point their package sources (package.json, Maven, Gradle) at the HeroDevs repository. From then on, high and critical CVEs in end-of-life versions are patched and pulled in on normal builds under the company's SLA. He describes HeroDevs as about 120 employees and mostly bootstrapped, helped by Google effectively blessing it as AngularJS's maintainer when that was deprecated.
• Paid support for old versions funds and relieves upstream maintainers. Barnes says HeroDevs hires key project staff, sponsors bodies such as OpenJS, funds community efforts and shares revenue upstream. In his account, projects can then point to HeroDevs in their deprecation notices and keep innovating on new versions. Hanselman frames it as a balance between open source and capitalism, though he says he will keep urging people to upgrade.
• Running unpatched software carries real regulatory and legal exposure. Barnes, a former practicing attorney, cites CISA rules requiring government contractors to patch vulnerabilities within set times, extra requirements in insurance, financial services and healthcare, and EU regulation. He adds that a breach traced to an unpatched, remotely exploitable high-severity CVE helps show negligence in a class action.
• Demand for end-of-life versions grows the longer they're dead. Barnes calls the growth "exponential", because companies and agencies rolling out mandatory CVE scanning keep finding forgotten apps; large .NET customers run tens to hundreds of applications. The current .NET focus is LTS releases that have gone end of life, starting with .NET 6, which ended support in November 2024. An STS version would be considered on demand, since the catalog is customer-request driven.
• Long-tail libraries are triaged per customer. At onboarding, HeroDevs reviews each dependency and decides whether to fork it internally or pay its maintainer to keep a branch, drawing on research into the support life cycles of the top 200 community NuGet packages. Builds must match distribution packaging bit for bit for apt and RPM repos, with a WinGet repo planned.

2025-04-17 · open the original ↗

• The book aims to keep middle schoolers by casting them as builders. Looper's Computer Science for Kids: A Storytelling Approach (Wiley) targets US grades 6 through 8, ages 11 to 13, the point where she says her research shows girls in particular drop out of computer science. Readers follow a guide through a mystical forest solving problems such as securing a gate against a moss thief, with a hidden curriculum on sustainability.
• Every chapter is mapped to a formal curriculum standard. The standards are the Computer Science Teachers Association's. Looper says they are demanding, down to compliance for sixth to eighth graders and training machine learning models from scratch by high school, and expects them to be revised toward AI this year and next. Hanselman singles out the mapping as what separates the book from whimsical kids' titles, along with its recurring "think like a computer scientist" sections on problem solving.
• Everything runs in a browser, so cost isn't a barrier. The progression runs from low-code to Microsoft MakeCode and micro:bit, which has an emulator, then Twine and finally 3D worlds in A-Frame. Looper's reasoning is that "there is ability everywhere, but access is quite limited". Teachers get a password-protected site with answers and a Common Cartridge package for learning management systems such as Canvas. Hanselman says he works with the MakeCode team.
• Local testing showed the textbook format is the weak point. Looper took the idea to the computer science teacher at her local school in Wellesley, Massachusetts, where she says computer science is often staffed from the math department, and then to a focus group of sixth graders. The students asked for video and quests, which she has begun adding on the companion website.
• Looper hasn't settled how AI belongs in teaching. She has paused CS 601, her front-end course at Boston University's Metropolitan College, while she works it out. She favours writing HTML by hand with Copilot off, and for advanced students, code-reviewing a site an agent scaffolds. In her own work she found Copilot fast for a Vue 2 to Vue 3 upgrade and Bolt.new clean, but found a fully agentic tool "terrifying" and its code sloppy. She wants Copilot as a junior engineer for toil, with the creative work kept human.

2025-04-16 · open the original ↗

• ShaderGlass overlays retro shaders on any window, invisible to capture. Hanselman demos the app, written by a self-described retrocoder whose name the captions garble, applying effects such as a Game Boy, NTSC VCR or CRT-geometry look over VS Code. Neither the podcast screen share nor a screenshot could see it, so Mark watched from his iPhone. Mark explains that it writes directly to the frame buffer, so Windows is unaware of it. Hanselman calls it "100% useless" and delightful: people should build things "for the heck of it".
• Windows Terminal compiles HLSL pixel shaders dynamically. Hanselman points the experimental pixel shader path in Terminal's default profile settings at readable HLSL files from GitHub, editing parameters such as tint colour, bloom and scan lines. He says he uses a CRT shader when talking to his PDP-11.
• Vibe-coding a fractal shader failed first, then worked. Hanselman gave Claude 3.7 thinking the Dawn shader as a base. The result failed to compile ("incorrect number of arguments to numeric type") and the session broke, and he concedes he is only at the hello-world level of writing shaders. A restart produced a working dynamic-zoom fractal shader, which he published by dragging the file into a GitHub repo and having Claude write the README, which they call shovelware.
• Copilot's code citations surface licensing questions AI doesn't weigh. The output was flagged as matching code under four license types. Mark notes that many hits were boilerplate every Terminal shader contains, and that copied "magic numbers" trace back to whoever first conceived them. One source was a 9-year-old GLSL file with no license; a function traced to a 2013 Stack Overflow answer under a "do what you want" licence, which Hanselman notes is not OSI-approved but is compatible.
• Hanselman claims AI-generated code can't be copyrighted. He says the US Copyright Office ruled that AI-generated content cannot be copyrighted, so the shader is free by definition. Only lines they modify themselves would be copyrighted, which he calls "interesting territory" for AI-assisted coding. Mark first asks whether the ruling covered only generated art.

2025-04-10 · open the original ↗

• Missing data, not models, is the Global South's AI barrier. Kallott, founder and CEO of Amini in Nairobi, says 2.6 billion people are unconnected and so "invisible" to training data, and Africa has more than 2,500 languages, many only spoken. The continent has only 1/8 of the minimum density of meteorological stations. Her team set out to build an AI company for Africa, found there was no data, and pivoted to data infrastructure, starting with environmental data.
• Global earth-science models lose accuracy on African ground. Kallott contrasts large US monocrop farms with African farms of around one square kilometre growing several crops, so open satellite imagery at 10 metres has to be super-resolved toward one metre. Readings are calibrated against ground truth, such as soil samples and yields collected by data partners. Amini worked bottom-up, starting with one region of Kenya, because she says East Africa has little cloud cover and the West a lot.
• Amini built its own geoprocessing engine to beat cloud cover. Over about a year and a half it built a Google Earth Engine–style system that layers multispectral, hyperspectral and synthetic aperture radar satellites and applies cloud masking. It also takes in drone imagery, tabular data and IoT sensors, including weather stations built into a startup's e-bike chargers.
• Revenue comes from insights; models and community access stay free. Kallott says customers are global exporters of agricultural and mineral commodities wanting first-mile data, and governments mapping their resources, while developers never pay. Version 0 of the foundation model went on Hugging Face the month before, with the next version in training and a first ICLR poster accepted. Amini runs hackathons on Zindi, such as telling cocoa from coconut trees in Côte d'Ivoire. Expansion to the Caribbean and Southeast Asia rests on her finding that about 70% of work by agro-ecological zone carried over.
• Africa must become "AI-ready" before it can leapfrog. Kallott claims Amini is the only company doing this and wants to build "an ecosystem, not a product". She compares it to Palantir digitising government data more than 10 years ago, though not in defence. She argues the work is best done by teams "born from within", who build with a UPS and generator through power cuts and train foundation models on office workstations.

2025-04-03 · open the original ↗

• Swap linear goals for experiments wherever life allows. Le Cunff, a neuroscientist, founder of Ness Labs and author of Tiny Experiments, says linear goals suit narrow, predefined outcomes such as saving for a mortgage. Elsewhere she would start from a question or hypothesis you are curious enough to test, rather than a fixed outcome you tie your happiness to.
• The obsession with finding one purpose backfires. In her research for the book she found that mentions of "finding your purpose" in books rose 700% over two decades. She argues it puts all of a person's self-worth in one basket, citing founders who suffer chronic anxiety and depression after a startup fails, and that people who have "found their thing" describe a messy process, not a plan. Hanselman blames social media, and she calls it "a giant artificial leaderboard". She doesn't tell people to stop scrolling, since she still does it herself. Instead she suggests replacing some of it with sharing something useful once a week.
• A tiny experiment is an action plus a duration, reviewed honestly. Her own example: during a book launch with recordings from 8 a.m. to 8 p.m., she took a 20-minute walk daily for 20 days, then judged by felt signals such as mood and sleep whether to continue, tweak or stop. A weekly YouTube video for about three months drew good subscribers and comments, but she dreaded making them and quit. She says this kind of trial should come before committing to a habit, and that the popular 66-day habit figure oversimplifies the research.
• Learning in public is a spectrum, starting with one person. Le Cunff says it can be one friend or a temporary WhatsApp group exploring a new tool for three weeks, not a large following. It is optional, but she says a support system, like a marathon runner's team, is how tiny experiments grow into ambitious ones.
• Procrastination is a signal to diagnose, not beat with willpower. Her "triple check" asks whether the block is in the head (should this be done at all?), the heart (is it enjoyable?) or the hand (are the tools and skills there?). Applied to Hanselman's TED talk outline, stuck at 60% for three months, she diagnoses heart and hand. Her remedies are intentional imperfection, working alongside a fun friend, and asking someone who has given a TED talk for the two things that matter.

2025-04-02 · open the original ↗

• Internal audiences allow a candour external talks can't. At Microsoft's internal TechReady conference, Hanselman gave a talk arguing "we're selling Copilot wrong" and found the openness so freeing that he is now saying more of it publicly. Mark describes giving an Azure strategy talk on the hyperscaler landscape there that he could never give externally because it is business strategy and opinion.
• Mark publicly called for treating C and C++ as deprecated. In a 2022 tweet he argued Rust was ready and the industry should shift to memory safety, after reviewing trends showing 70% of vulnerabilities were memory-safety problems. He says he posted it as a technical opinion without press clearance and got a Teams call the next morning asking whether he really believed it. The stance built the reputation that led to his recent keynote at Rust Nation UK.
• Microsoft tolerates employee posting, but discretion still applies. Hanselman says Microsoft is far more relaxed than Google or Apple, where friends were told to stop blogging or YouTube. Mark says he has told team members that social media is the wrong forum for some points. Hanselman's team rule, "don't dunk on other teams", comes from a tweet about Windows 10 printers that brought three VPs down on him; the bug got fixed, but he says he should have asked internally.
• Being inside lets them get Windows bugs fixed directly. Mark recounts a printer failure that took a driver reinstall to clear and says the Windows printing team took it up; they point to the universal printer driver as the fix for new printers. Hanselman says a regression in how USB device INF files were set broke Zune on Windows 11, which he found while preparing Zunes for Marvel's Guardians of the Galaxy 3, and that the fix has rolled out.

2025-03-28 · open the original ↗

• Technology adoption in medicine "flips" expectations. Lee's example is ultrasound: proposed as a diagnostic tool in the late 1950s, feared, then standard of care about a decade later, to the point that refusing it now would look like malpractice. He concedes skills are lost along the way, but says the depth of debate among leading thinkers, which he compares to the one after the human genome was mapped, makes him optimistic.
• Reaching a capability first costs roughly ten times more per tier. Lee puts a GPT-2-level model at about $5–10 million of compute, GPT-3 level at $50–100 million, and GPT-4 level near $1 billion, in any domain. Once someone shows it can be done, followers get there "smarter and a hell of a lot cheaper", and he asks whether the premium on being first still holds.
• AI will soon build things that work but no human understands. Lee predicts that in "much less than 5 years" an AI will produce a mathematical proof in a proof language such as Lean that type-checks as valid but defies even the smartest humans' comprehension. He says that would be his first mark of AGI or superintelligence.
• Patients rated GPT-4's replies far more empathetic than doctors'. Lee cites a UC San Diego and Stanford paper, published a month after his book, in which GPT-4's answers to patient emails were equally accurate and judged more empathetic by a factor of 9 to 1 in a blind test. His reading is that a frazzled doctor writes two or three sentences while the AI writes paragraphs and recalls personal details, which he calls both interesting and disturbing.
• AI for science is stalled at GPT-2 level for lack of data. Lee says Microsoft Research's models for weather, proteins, molecules and materials are published in journals and released on Azure Foundry simultaneously, but there is "no internet of molecular dynamics simulations". He predicts the first practical scalable quantum machines in 2025, and wants them first used to generate large sets of perfectly labelled training data.

2025-03-28 · open the original ↗

• Image-based, A/B-updated Linux desktops give Chromebook stability without the restrictions. Garrison runs Bluefin, built on Fedora Silverblue, where rpm-ostree works like "git for your file system": a full distribution upgrade is just a reboot, and he can bookmark an OS version and fall back to it. He installs command-line tools with Brew and GUI apps as Flatpaks, and sees the read-only system as a security practice moving from servers onto the desktop.
• Linux already runs the media industries' workstations. Garrison managed Linux at Disney Animation, where all the artist workstations ran Red Hat Linux and the company ran on Google's web apps rather than Microsoft Office, and he says DreamWorks and ILM run Linux too. Having also managed thousands of Windows machines at a university, he says Windows "gets broken just the same".
• Wayland's strictness hurt, but its sandboxing is deliberate. Garrison says the move from X11 was rough because Wayland refused broad compatibility and much software was never updated. In his view the isolation is there for a reason, though it still costs him, for instance when he cannot drag a file from an NFS share into a sandboxed Slack.
• Owning more of your stack is worth the friction. Garrison has switched to Firefox and is "de-Googling", and finds many web apps are Chrome-specific rather than web-standard. He rips DVDs into Plex and runs local home automation that keeps working when the internet is down. Hanselman calls pushing back against a browser monoculture noble rather than self-inflicted.
• Give non-technical users a Chromebook, or an appliance-style distro. Garrison's first answer for family is a Chromebook. For Linux he now recommends Bazzite or Bluefin, because updates arrive automatically on reboot, and he says Linux Mint, which he once promoted, lacks a good update story.

2025-03-19 · open the original ↗

• You can message anyone if the question is precise and theirs to answer. Hanselman recounts messaging a distinguished engineer he had never met about how the ONNX runtime does quantization for an NPU, which led to a half-hour conversation. The senior PM watching said they had never felt empowered to do that. Hanselman concedes his VP title and name probably played a part, not only the quality of the question.
• Reputation is built by never bothering people with randomness. Mark says his title, seniority and reputation let him contact anyone. He leaves requests that could have been Googled, such as "where's the deck", read and unanswered, but answers thoughtful ones; someone at the last Ignite told him he had replied to their email 15 years earlier. Hanselman compares this to Donald Knuth's reward checks for finding errors in his books: people do their homework before approaching him.
• Answer a good question once, publicly, and send the link. Hanselman says he turns a thoughtful question, such as an Indiana University student's email about impostor syndrome, into a blog post or other public piece and replies with the URL, so others benefit too.
• Admitting confusion is fine only after you've done the homework. Mark says he forces himself to say "I don't understand" in meetings, but doesn't want to stop a meeting over something he should have prepared. Hanselman draws the line between not having prepped and having done the research while some points have not yet clicked, which he says is appropriate at every level.
• At senior levels, the right point can land in the wrong venue. Mark weighs whether an opinion is deep enough to raise in a meeting with the senior leadership team and Satya. Hanselman says his VP replied "not helpful" when he made a point to 83 people that only three needed. He keeps his emails to Satya to two or three a year, and warns that asking basic questions flips the "bozo bit", which is hard to flip back.

2025-03-19 · open the original ↗

• VTM is a text-mode desktop that deserves far more attention. Hanselman, who says he doesn't know its developer, notes the open-source project had about 2,000 GitHub stars and argues it deserves 20,000. He likens it to DESQview, the old text-mode multitasking environment, and places it among TUIs such as Midnight Commander and Malcolm Smith's YEdit. He says the whole thing is about five megabytes.
• Its windows behave like a desktop, with tmux-style splits. In the graphical mode he shows resizable windows with z-order that can be flung around with physics, running DOS, bash, PowerShell and Vim side by side. There is a split window manager with "empty slots" to drop terminals into, hotkeys such as Shift+Alt+W, and an XML file configuring the five launcher apps. He notes it "gets a little angry" with his three-monitor layout.
• Run as a text app, it reattaches to detached sessions on an infinite canvas. After closing the graphical version, Hanselman starts VTM inside Windows Terminal and reconnects to the still-running processes, as screen and tmux do. Viewport markers show how far away off-screen windows are, which he likens to an infinite Canva-style terminal manager. He calls it "pretty obscure and confusing", with distinct left, right, middle and Ctrl-click behaviours.
• Windows has long been "missing the text mode boat". Hanselman points back to his blog post of that name from 2004, 21 years earlier, and to Gui.cs for building console apps with UI. He credits both the Windows Terminal team and VTM's developer for this working inside Windows Terminal.

2025-03-13 · open the original ↗

• Regulatory economics, not science, keep novel devices out of medicine. Jepsen cites a paper averaging $658 million and 13 years to approve something novel and therapeutic, built on "quality builds" of about 10 units that rule out Moore's law and contract manufacturing. She says a million papers in 20 years on infrared, ultrasound or electromagnetic treatments have produced a rounding error to zero in the healthcare system. A single rare disease such as glioblastoma, about 5,000 US patients a year, ends up as a treatment costing around $1 million after roughly $1 billion spent.
• Openwater sells an open-source, general-purpose device as its business model. Jepsen likens it to Android, where the "apps" are regulatory approvals her customers obtain. She says the system costs $10,000 now, against $1 million earlier, and could reach the cost of a smartphone at volume. It is released under AGPL and Creative Commons 4.0 ShareAlike; the company has raised $54 million last year and $100 million in total, and expects to ship a few thousand units this year and go profitable. A $50 million non-dilutive check from Ethereum's founder came with a veto over ever closing the source.
• Smartphone camera chips and a pulsed laser can spot strokes. Jepsen says recording holograms with the micron-sized pixels of phone cameras gives blood-flow readings in phantom tissue accurate to 0.1%, which she calls 20 times better than CT, MRI or Doppler. She describes it as an "ECG for your forehead" to route large-vessel-occlusion stroke patients, a type she says kills 6.5 million a year, to the 5% of hospitals that can treat it. Treated within two hours, she says, there is a 90% chance of no neural deficit; the laser starts production next quarter.
• Low-intensity ultrasound tuned to cancer's resonant frequency kills it selectively. Jepsen reports sweeping frequencies on 16 glioblastoma lines grown as organoids, then moving 30 mice into remission with no damaged healthy cells found at autopsy. She says the treatment releases proteins that "vaccinate" the body against its own cancer. In a severe-depression study almost half the patients went into remission on 10-minute doses over three weeks, and 150 kHz pulses at a 10% duty cycle break microclots below five microns in preclinical work. None of it is FDA-approved yet.
• Open safety data and hacker access are the next bottleneck. Jepsen wants all safety and outcome data shared with regulators, doctors and patients. She notes that ultrasound use on people is regulated as Class 2, that experiments on anything with a backbone need ethics review boards, and that the open-source community needs hardware and lab space. Hanselman, whose team runs Microsoft's open source programs office, offers Microsoft Reactor spaces.

2025-03-11 · open the original ↗

• Distilled DeepSeek R1 models run locally on a Copilot+ PC's NPU. Hanselman uses the free AI Toolkit extension for Visual Studio Code on a 12-core Snapdragon Elite laptop with 32 GB of RAM. It offers NPU-optimized distillations of DeepSeek's large mixture-of-experts model at 1.5, 7 and 14 billion parameters, which he calls the "squished", "JPEGy" student versions.
• The 14B model works fully offline, but slowly and inaccurately. He switches the laptop to airplane mode mid-answer to show nothing is going to the web, and reports about 40 tokens a second. The model "yaps" through its thinking and gets details about him wrong, so he warns against using it as a fact machine. The 1.5B model runs much faster and loads the machine far less, at a cost in accuracy and breadth.
• Loading the big model took trial and error. The 14B model first failed to load, which he blamed on having too many models in memory, so he killed the toolkit's inference service agent; he later found the toolkit's own "Unload" option. He notes it also offers a playground, a prompt builder, local fine-tuning and free cloud models through GitHub Models, none of which he tried.

2025-03-10 · open the original ↗

• Colour-code your calendar by how each meeting will feel. Hanselman uses Outlook categories: green for positive personal time such as lunches, red for meetings with bosses or VPs, orange for presentations and podcasts as "approaching stress", pinks and purples for one-on-ones, and blue for general business. He argues a single-colour calendar is like playing Tetris with one colour.
• Grey focus blocks protect time without looking like appointments. He colours focus blocks grey, the same as the calendar background, so the time is held against bookings but doesn't read as a meeting.

2025-03-07 · open the original ↗

• A migration is far harder than copying files. Beaudoin-Allaire, a principal software engineer on ShareGate at Workleap, says you don't own the data or know what you'll find: a thousand documents or a million, odd settings, and sources and destinations with different capabilities. Large files must be streamed from source to destination rather than held in memory, and Microsoft 365 now has chunked upload sessions for them. He recalls the file-size limit growing from 2 GB to perhaps 50 GB or more.
• Microsoft Graph simplified things, but legacy APIs are still needed. Graph gives one way to authenticate and consent, documented calls, and a dependable SDK, unlike older SOAP and CSOM calls that were barely documented. It still has blind spots, so ShareGate keeps an ordered fallback list that includes SOAP calls and exposed SharePoint DLLs, while on-premises versions from 2003 onward need their own APIs. Hanselman calls Graph one of the smartest things Microsoft 365 ever did.
• Every extra call is multiplied a millionfold. Adding one call to the ten or so needed per file version adds up across a million files and can trigger throttling, which he likens to a car braking for no reason and slowing all the traffic behind it. Limits differ by service, shared in SharePoint and per mailbox in Exchange, so work is parallelized to match, and the software is built around fault tolerance in the spirit of chaos engineering.
• ShareGate exposes a performance slider rather than adapting automatically. Beaudoin-Allaire says they cannot tell how close a tenant is to its limits, and must handle both graceful 429 "too many requests" responses and 503s, the service saying it is "on fire". The slider predates cloud rate limiting: on-premises servers could crash under load before signalling anything. Most of the time the network, not the customer's machine, is the bottleneck.
• Metadata that doesn't map gets an approximation and a warning. Permission levels differ between Google and Microsoft, so ShareGate maps them to read and write equivalents and lets customers remap them. Where something cannot be supported, it says so in documentation or warns during the copy. Google Drive files appearing in several places through links were copied to one place, with a warning.

2025-03-07 · open the original ↗

• A blog platform must make leaving easy and make content last. Reece, founder and CEO of Micro.blog, says the service archives pages you link to, keeps a paid blog online unless you ask otherwise, can ping the Internet Archive on every post, and mirrors posts and photos to GitHub about once a week. He says the platform would be "hypocrites" to preach the open web while trapping content, which Hanselman ties to the link rot he finds on working actors' expired sites.
• Micro.blog deliberately stays small. Reece argues that very large, ad-based platforms develop bad incentives and moderation problems, while small fediverse communities can manage themselves. He says tens of thousands of users is fine, and that at 100,000 they would probably have to stop taking new users. He took Hanselman's word "boutique" for it.
• Discovery is hand-curated and shows no follower counts. Reece says trending hashtags let communities clash and invite hateful piggybacking, so a human curates the Discover page from recent posts. An early "people to follow" list was dropped because it snowballed popularity, as Hanselman recalls Twitter's suggested-user list doing. Reece concedes the balance may lean too far toward making people hard to find, and is open to linking distributed starter packs.
• POSSE concedes that not everyone reads RSS. Publishing on your own site and syndicating elsewhere lets you keep control while still reaching people on social networks. Reece says he added cross-posting first for Twitter, because existing tools added unwanted links and he wanted short posts to read like tweets. It now supports eight or nine services, each coded to its API, though platforms such as Instagram lock down posting.
• A $1-a-month plan is an experiment made possible by static hosting. Micro.one launched a week before the recording with domain, photo and podcast hosting, alongside the $5 standard plan, which Reece says will always be $5, and a $10 premium plan. A credit card is required, mainly to deter spammers and AI-generated content farms, and unlimited photo storage is his main cost worry. Blogs are published with Hugo as static HTML served separately from the platform, so they stay up during an outage.

2025-03-07 · open the original ↗

• Visual Studio on Windows has a built-in remote Linux file explorer. Hanselman shows that adding the "Remote File Explorer for Linux" component, about 30 MB under the Linux and embedded development workload in the Visual Studio Installer, puts a Remote File Explorer on the View menu. With it he browses his PiDP-11, a PDP-11 replica kit running on a Raspberry Pi, opens its start-up shell script, has Copilot explain it line by line, edits it with Copilot completions, and saves it back to the Linux machine.
• Visual Studio's terminal can SSH straight into the remote machine. The terminal's connection list includes his WSL distributions and the remote connection, so he opens a shell on the PiDP-11 without any setup and boots Unix V7 on it. In the same terminal he also runs YEdit and htop, which he takes as a sign of how much the team cares.

2025-03-05 · open the original ↗

• Systems thinking matters more as AI does the coding. Hanselman relays a systems architect's argument, from what he calls his most popular Hanselminutes episode, that the learn-to-code movement taught for-loops but not the systems code lives in. Her example is a cheap Amazon package that arrives next day with nobody thinking about the oil, freight, labour and macroeconomics behind it. Mark agrees and equates systems thinking with architectural thinking, saying that with AI writing so much code, the value lies in the thinking on top of it.
• Sketch the rough shape, then sculpt iteratively. Mark says he starts a Sysinternals feature from a broad structure and reworks it as he goes, likening coding to sculpture where the chisel strikes can't be predicted. Hanselman says his own systems thinking leaves him stuck among nine ways to do something. In brownfield code, both say you should sculpt in the previous sculptor's style rather than bolt things on.
• Nobody holds today's systems whole, so learn just enough to decide. Mark relays Scott Guthrie being asked who still has Azure in their head; the answer was nobody, and Mark says he knew a far larger share of Azure 10 to 15 years ago. As a senior executive reviewing experts' work, he asks how much he needs to know to make a call, looks for risk areas from experience, and asks questions where there are gaps.
• Mark sees "no line of sight" to AI replacing programmers. His first reason is specification: you never fully specify what you want, the AI's fixes break other things, and you end up owning code you didn't write, which leaves a non-coder "dead in the water". Hanselman's example is a site built with his 17-year-old, where the AI made red ones-and-zeros "Matrix rain" from 256 hard-coded spans that overheated his laptop. Mark says over 95% of his own research code is AI-generated, which he can manage only because he understands it.
• AI is stuck on old APIs and thin documentation. Mark says a model generated code against an older OpenAI SDK because it had seen more examples of it, and that generated reference docs lack the context a human or an AI would need to use a new API. His conclusion is that AI accelerates experts but limits beginners to simple projects, so young people still have to learn systems and how to program.

2025-02-28 · open the original ↗

• Warp breaks terminal conventions to behave like a notebook. Lloyd says the mouse and text selection work as in a normal editor, and each command is married to its output as a "block" that can be shared, filtered or attached as AI context. He accepts some compatibility cost for a native GUI over "terminal purity": "rules are made up."
• Agent mode drives the terminal, asking approval by default. An on-device classifier decides whether input is a command or English; agent mode can run commands such as git status to gather its own context, as in his merge-conflict example. Users can opt into LLM-judged read-only commands or allow and deny lists; models come from OpenAI, Anthropic and DeepSeek (R1), with Claude 3.5 Sonnet the default "as of today".
• AI, not collaboration, is what now pays for a terminal. Asked about a reported $50 million raise, Lloyd says free usage made the early years stressful, and Copilot and Cursor changed developers' expectation to paying for AI's marginal cost; he says revenue is "growing pretty well". His stated aim is "the world's best place where developers can build software", with the terminal as the starting point, and Warp Drive extends shared team commands into context for its AI.
• Windows came last because Warp went deep on Mac first. Founded June 2020 and written in Rust, Warp shipped on Mac, then Linux; Windows was the top feature request, serious work began about a year before recording, and the Windows build launched the day before. Lloyd says he would now do Windows before Linux and concedes Windows Terminal is more polished in places; the episode is sponsored by Warp, and Hanselman says he spent hours with the team on its Windows experience.
• Offline-first, with AI and telemetry under user control. Everything except AI and real-time collaboration works offline, one toggle removes all AI, and local models via Ollama are being looked at but not built. Smart completions with flag explanations come from a local database built on open-source completion specs.

2025-02-22 · open the original ↗

• Recognizing only your own commands makes local voice fast and accurate. Hansen built Rhasspy around asking "which of these things did the user say" rather than "what did the user say", using Kaldi with a bespoke language model of a few hundred kilobytes in place of a gigabyte one. He reports sub-second responses on a Raspberry Pi 4 and under a tenth of a second on a desktop, but close options such as 15 and 50 need clearer speech, combinatorics can blow up to millions of sentences, and rejecting out-of-scope speech only recently became robust.
• Local-first means choice, not rejecting the cloud. Hansen disputes Hanselman's "vote of no confidence" framing: components are modular, Nabu Casa's cloud routes speech through Microsoft Azure, and requests the local system cannot handle can fall back to ChatGPT, Claude, Gemini or a local model via Ollama. He cites quadriplegic users who ran Alexa, Google and Siri together for redundancy and wanted something more robust.
• ESPHome hardware is how voice reached Home Assistant. Hansen, who works for Nabu Casa, says its 2023 "year of voice" push extended ESPHome firmware, and the company sells the $59 Home Assistant Voice Preview Edition with an XMOS audio chip, "preview" because the software is still a work in progress. The device keeps only enough audio to detect the wake word, then streams to the user's Home Assistant server, where Whisper suits beefy hardware and an add-on he is building trains itself on the home's device names for a Pi.
• Training data, not models, limits accents and languages. He says wake-word data lacks accented English and female or high-register voices; his Kaldi models use public-domain Mozilla Common Voice data, and custom pronunciations can accommodate accents. For his Piper text-to-speech voices about a dozen volunteers record two to three hours each, Nabu Casa pays for studio time in some languages and releases the data to the public domain, and LibriVox audiobooks train poorly because narration lacks short phrases.
• Cloud assistants are declining as local LLM hardware lags. Hansen says Alexa and Google are getting worse as their makers fail to profit, while local LLM boxes cost two to five thousand dollars and are not plug-and-play with Home Assistant. He wants an in-home AI box, which would allow uses such as a vision model describing a camera snapshot every five seconds that a paid API would make too costly.

2025-02-20 · open the original ↗

• Explorer's inconsistencies are the price of decades of compatibility. Hanselman calls some of the hate toward Windows 11 File Explorer unfair but shows its rough edges: a flash of light mode in dark mode, and modern menus that open older-looking dialogs. He says Explorer carries COM objects, XAML islands and plug-ins that a from-scratch app can ignore, and that search improvements are underway.
• Files is a polished open-source alternative aimed at aesthetics. He believes files.community is largely one main developer with many contributors; it lists WSL and cloud drives such as Dropbox separately, gives removable drives an eject button, and adds a Git column showing status and the latest commit message. It also enumerates the shell extensions plugged into Explorer under "show more options", and supports tear-off tabs; it is on the Windows Store.
• File Pilot is a tiny from-scratch beta built for speed. The whole app is 2 megabytes, it is free during the beta with a license mentioned for the future, and he guesses it is written in C or possibly Rust. He shows instant Ctrl+T tab spam, a fast filter and an Everything-style global filter, asynchronous folder sizes on hover, middle-click split views, and an open-console command, but warns its default font is "shockingly small" and notes there is no cloud-drive call-out yet.
• The competition is good for Explorer. Hanselman calls the two apps "effectively a vote of no confidence against Explorer" but welcomes them as unapologetically Windows-native apps that give the Explorer team something to aspire to. He also points to Midnight Commander, by Miguel de Icaza, as a Norton Commander-style option.

2025-02-19 · open the original ↗

• Mark wrote ZoomIt for his own Windows internals demos. He says he built it because a Toshiba laptop zoom utility was clunky for the Sysinternals demos he gave with Dave Solomon, and it has evolved for 20 to 25 years. Most of it sits in one C file around a 7,000-line function; it now ships in PowerToys and is open source in the Sysinternals GitHub repo, where Mark says a refactoring PR would be considered.
• ZoomIt is an everyday tool, not just a presenting aid. Mark says he zooms almost daily on fonts he cannot resize, and Hanselman calls it the first thing he installs. Against Windows Magnifier, which they describe as an accessibility tool with a floating toolbar, they stress ZoomIt's keystroke-driven annotation, which Hanselman likens to John Madden drawing on screen; he says people with eyesight impairments and teachers use it too.
• The feature set runs from zoom-and-draw to live zoom and recording. Mark demos Ctrl+1 to freeze and zoom, drawing with shapes, letter-keyed colors, highlighting and blur, typing, Ctrl+2 to annotate unzoomed, and the newer live zoom that annotates a moving screen. Ctrl+6 copies a screenshot or crops one to file, and Ctrl+5 records the screen or a region; there are also break timers and black or white whiteboard screens.
• Fluency in ZoomIt comes from drilled practice, like a kata. Hanselman argues that pointing, talking and drawing at once only looks effortless because it is practiced, and grades it in improvised belts from simple zoom-and-click up to typing while talking. His tips are to point at the target before Ctrl+1 so the zoom lands there, and to avoid nauseating the audience, though he could not demo live drawing because his build lacked it.
• DemoType replays prepared code as if typed live. Hanselman says he pitched it to Mark after the joke site hackertyper, and Mark built it with others' contributions: snippets in a text file, split by start and end markers, are typed out automatically or driven by random keystrokes, and can include Enter and arrow keys to walk through IntelliSense. Hanselman concedes it is not yet in his muscle memory, though he says it improved his South Africa keynote.

2025-02-19 · open the original ↗

• "Squishy" human experiences can be measured and predict real outcomes. Hicks, who runs Pluralsight's Developer Success Lab, says psychology builds scales tested across thousands of people, likening its measurement error to atmospheric dust clouding starlight. Sense of belonging predicts persistence and grades, though she stresses rules have limits: belonging will not keep someone in a field that does not pay for groceries.
• Belonging on a team tracks with reported productivity. In her study of 1,282 developers, those on teams with a high sense of belonging and time to learn reported being more productive over the last month and rated their teams more effective. She says the lone-genius, 10x-developer stereotype holds grains of truth but, over-activated as a team norm, leads teams in the wrong direction.
• Code review anxiety competes with code for working memory. Hicks says anxious reviewers avoid or mentally check out of reviews, because worrying about colleagues uses the same cognitive resources as problem solving. Self-efficacy, the belief you will get past the friction, predicted productivity in her research, and she favors brief, repeated exercises such as her lab's workbook that build a "virtuous cycle" rather than a one-off intervention.
• Hackathons let people safely try on new roles. Pluralsight engineers asked her lab to measure their hackathon's value, and the lab found that people on supportive teams were more willing to try new skills afterward, whether or not product features resulted. She argues the temporary team lowers the stakes of experimenting with a new identity in front of a manager.
• AI skill threat is common, and team culture halves it. Surveying over 2,000 developers and 795 engineering managers, she found roughly 43% to 45% agreed they would not keep up or be judged fairly in an AI-coding future, and being a manager did not protect against it. Those on teams with belonging and learning cultures had about half the threat; she says cognitive offloading can be good or bad, the answer is "never going to be in the tool", and she worries tool builders are not bringing in learning scientists.

2025-02-05 · open the original ↗

• Prep time scales with the stakes and how new the material is. Mark says after thousands of talks he preps little because his content is incremental, yet their joint prepared demos take 10 to 15 hours. Hanselman lists six talks in one week prepped from 10 minutes to a couple of hours, against a 45-minute AI Tour slot in Johannesburg before 5,000 people with backup machines, and a TEDx Portland talk of 15 to 18 minutes that is getting in his head.
• Design the talk for one defined target listener. Mark says to decide who the audience is and what they know, aiming at that person and briefly bringing others up with "for those of you not familiar"; not knowing the audience is when he has been flummoxed. Hanselman says to start from what people should leave with, such as something they can implement that day, and to remember the room wants you to win; both pick a friendly face on each side to play to.
• Time-box sections and keep the plan on paper. Mark divides his minutes across sections, rebalances them, writes clock times to check against, and memorizes slide order because unrehearsed transitions look amateur. Hanselman splits 45 minutes into 5-minute intro and outro bumpers plus seven 5-minute sections in three acts, keeps notebooks from every talk, and advises beginners to keep two or three talks ready, since in his experience one or two conferences in ten lose a speaker.
• Pre-stage nerves are physiology to manage, not a warning. Mark uses a power pose and biofeedback, watching his heart rate and breathing to get it under 100 before going on. Hanselman says his heart rate runs 105 to 110 before a talk and he reframes it as his body sending blood to his brain to help him succeed.
• Keep forward momentum instead of second-guessing words. Mark primes himself to speak slowly and accept the words he picks, likening stumbling to watching your feet while running downhill; one host says he has replaced "um" with silence over roughly the last 10 years, which they agree audiences do not notice. Hanselman credits Billy Hollis with staging tricks such as assigning client and server sides of the stage, and recommends Toastmasters as proof that speaking is learned technique.

2025-01-30 · open the original ↗

• Children should be creators of technology, not only consumers. Answering Hanselman's story of a family who found a 3D printer too expensive while sitting in front of a PlayStation 5 and its games, Berry says gaming is fine but should come with building: Minecraft worlds, Roblox, and the maker spaces and 3D printers now in public libraries.
• Art belongs in STEM, and may be the hook for girls. Berry argues engineers need history, ethics and a sense of how people meet their designs (the uncanny valley in robotics), and that ugly buildings and bridges don't get funded. She says sewing, cross-stitching or dolls do not disqualify anyone, and objects to parents at her robot exhibits going to fetch a girl's brother.
• High school homework builds the executive function engineering needs. She reports that since the pandemic, incoming engineering majors cannot self-organize or plan long-term, and wants homework for reflection whether or not it is graded. Because students are used to scrolling, she teaches her 50-minute lectures (phones banned) through active learning, not "chalk talks".
• AI is a tool to check, not a crutch. Unlike English professors who oppose ChatGPT on assignments, she wants engineers judging whether an answer is correct and efficient; she has seen ChatGPT give wrong answers to her exam questions. She compares it to a calculator: in AC circuits a student should be able to tell whether 350 volts makes sense.
• Students learn debugging by struggling through it themselves. She will not point out a forgotten power-supply switch or missing semicolon, because a student who spends five hours finding it never forgets; her own 17-hour session ended in a carriage return. She frames the skill as breaking a system into subsystems, and teaches multidisciplinary robotics because no design avoids another discipline; the broken-toaster exercise ("are the neighbours' lights on?") is Hanselman's illustration.

2025-01-23 · open the original ↗

• Blogging matters more now, because you own your words. Downie, a principal product manager at Microsoft who maintains dasBlog, says social networks drained blogs through sheer ease, but the risk of losing everything, which he learned when Windows Live Spaces shut down, made owning his words worth the setup. Hanselman calls posting on platforms "digital sharecropping", adding that people now feed AI models unpaid.
• Friction is the barrier, so keep an exit strategy. Hanselman argues that domains and certificates are too much for most people, much as the one extra step of picking a server cost Mastodon users. Downie, who prefers "stick shift" and keeps his blog as plain XML files, says anyone on an easier hosted service needs a way to take their words and images with them.
• The blog should be the origin, not a federated node. Downie considered adding ActivityPub support to dasBlog but gave it up, finding the spec too complex to keep up with even after 20 years of blogging, and prefers to publish on his own site and push elsewhere. Both say putting short posts on a curated long-form blog, as IndieWeb practitioners do, would need a separate notion of short posts.
• Write for your forgetful future self, with full context. Downie's first test is whether a post will help him later, assuming his "goldfish" memory will have lost it; Hanselman's criticism of other bloggers is that they always leave out a paragraph of context. Downie blogs answers to questions he gets repeatedly, including from mentees, as a force multiplier.
• Don't chase audience or cadence. Downie's most popular post, a throwaway on how he writes his yearly review, draws tens of thousands of views, and he blogs about monthly or when the feeling hits. Hanselman says 100 readers is a real audience, urges people to publish their drafts, and says a long-running blog shows more about a technologist than a follower count.

2025-01-22 · open the original ↗

• Public clouds grew out of platforms built for companies' own services. Russinovich traces Azure to the mobile shift, Ray Ozzie's 2005 services memo, and Project Red Dog, formed as he joined Microsoft in 2006, which Dave Cutler joined; AWS launched EC2 and S3 a year after the memo for Amazon's own needs. He says opening a platform to strangers adds most of the work, because you cannot trust the public as you trust your own staff.
• Cloud economics come from pooling and self-service. Enterprises provision for peak load and leave 30% to 60% headroom idle, while a multi-tenant cloud oversubscribes statistically because customers peak at different times, running close to 100% utilization. The other gain is agility, such as Scott Guthrie's early goal of a website in 90 seconds.
• Azure assumes "hostile multi-tenancy", isolated by virtualization. Russinovich coined the term against friendly enterprise multi-tenancy, treating the hypervisor's small trusted computing base as the security boundary, with networks virtualized too. Placement is handled by an allocator he co-authored a paper on (rendered "protein"), which packs VMs to avoid fragmenting servers and uses ML to predict arrivals; he puts the fleet's out-for-repair share at about 3% or lower.
• Azure launched as Windows-only PaaS; sales forced IaaS. Announced at PDC 2008 and commercially launched in February 2010, it required apps rewritten as cloud services; in 2011 Microsoft's head of sales told Amitabh Srivastava he would stop selling it without IaaS. Russinovich and Corey Sanders wrote the plan, IaaS shipped about a year later with Linux VMs, and he says Azure now likely runs more Linux than Windows VMs.
• Most developers need not learn the layers beneath them. Russinovich cites falling sales of the Inside Windows books, from about 250,000 copies for the 1996 edition to 75,000, 50,000 and 40,000 for later ones, as abstraction making internals specialist knowledge. Hanselman's advice is to learn one layer below your comfort zone.

2025-01-16 · open the original ↗

• Quantum computing will be hybrid with classical compute and AI. Svore, a technical fellow and VP of advanced quantum development for Azure Quantum, describes sending a subroutine to a quantum processor and feeding its results back, including as training data no classical machine can produce. She says Microsoft has demonstrated an end-to-end workflow using HPC pre-processing, logical qubits and an AI model.
• The best targets are low-input, low-output, compute-heavy problems. She points to chemistry and materials science, where accurate modelling of strongly correlated electrons matters, with aims such as carbon capture and cheaper, cleaner fertilizer catalysts. Hanselman compares the processor to GPUs, FPGAs and NPUs as a complement rather than a replacement.
• Physical qubits are too noisy, so logical qubits are built over them. Today's qubits fail about once per 1,000 operations, against the one in 10 million or 100 million needed; Microsoft's "qubit virtualization" uses error correction (Hanselman likens it to ECC memory). With Atom Computing it made 20 logical qubits from 80 physical ones, then 28 logical qubits that beat 28 physical ones, shown at Ignite.
• "Quantum advantage" means solving useful problems, not academic benchmarks. Svore prefers the term to "quantum supremacy": just over 50 qubits can beat classical machines on an academic problem, but useful science such as quantum magnet models needs at least 100 highly reliable qubits. The roadmap runs to "level two" resilient computing, then about 1,000 logical qubits needing upward of a million physical ones.
• Scaling is limited by control, which topological qubits should ease. Qubits cannot simply be stamped out by the million because controlling them adds noise, including heat from the control lines. Alongside the neutral-atom machine co-built with Atom Computing and work with Quantinuum, Microsoft is building its own topological-qubit machine, which she says is fast, digitally controlled, and small enough for a million to fit on a credit-card chip.

2025-01-09 · open the original ↗

• Reporting has crowded out the actual work of managing people. Coleman Hynie, CEO of Counterpoint and formerly at Credit Karma, says managers should split time roughly 50/50 between their own contribution and growing people, but nowhere near that happens now, amid a thinning manager layer and managers moving back to individual-contributor roles. This tension is what she says motivated Counterpoint, her own product.
• Many managers never chose the job and treat directs as competition. She says highly effective engineers get assigned management rather than choosing it, resent it, and struggle to stop doing the work themselves; at a past employer she built a "do I want to be a manager" program. She adds that job-hopping every 18 months, which she says people cannot be faulted for given rents, produces managers skilled at their craft but not at people.
• Metrics serve the story organizations want to tell. From quality engineering, she recalls bug counts used as a quality metric though they say little about product quality, and argues numbers get found to justify a preferred narrative. Hanselman links this to eroding trust, citing devices that jiggle a mouse to fake activity.
• Treat employees as adults: give context, demand accountability. She says top-down communication has lost context, so no one feels ownership, and endorses the "we hire adults" line from a viral Spotify ad Hanselman raised. She argues accountability is largely lost in tech, and that people can hold big feelings and still be responsible.
• "Professionalism" has been used as a weapon. As a Black woman climbing the corporate ranks, she says she was pushed into executive-speak and lost her identity and relatability; she argues formality is part of why people are lonely at work. Hanselman notes senior people can be quirky where those lower down or less represented must stay buttoned up.

2025-01-07 · open the original ↗

• Modernizing pays off in security, cost, performance and hiring. Herceg, CEO of Riganti in Prague and author of the roughly 600-page "Modernizing .NET Web Applications" (Apress, about 500 hours of writing), says the case is strongest for apps still in active development. He cites security improvements, new regulations like GDPR, cheaper Linux hosting on Azure, and developers who started eight years ago never having used .NET Framework; Hanselman reports his own blog's hosting fell about $40 a month.
• Fear of legacy code comes from missing tests and lost intent. Herceg says teams with thousands of tests can make big changes, while legacy projects passed through several generations of developers hold code no one understands. His team has used ChatGPT to reconstruct the original intent and business logic, with manual review, and says Copilot helps after the hard first test.
• DotVVM became an accidental migration path off Web Forms. Riganti started it in 2014 to build rich client apps in C# and HTML; because it runs identically on .NET Framework and .NET Core, it can sit inside a Web Forms app while pages are rewritten one by one into testable view models. Once Web Forms dependencies are gone, the target framework can be switched to .NET 8 or 9.
• Migrate incrementally, because business won't pause features. Herceg says large modernizations take months to one or two years, and few companies will stop new features for that long. Several chapters use YARP to route migrated pieces to the new app and everything else to the old one, giving early benefits.
• Library swaps carry hidden runtime traps, but newer tooling repays them. Moving from LINQ to SQL to Entity Framework Core produces runtime errors because lazy loading is off by default; CoreWCF migrations were smooth for him. He cites unified logging and dependency injection replacing several logging libraries per project, lower memory use in .NET 9, and Aspire, which he uses even for single-app projects to provision databases as containers.

2024-12-26 · open the original ↗

• The "dark forest" of AI-generated content has already arrived. Appleton says she first gave her talk on the expanding dark forest in generative AI about two years earlier as a thought experiment with no evidence, and has since given it five or six times; now half the talk is AI-generated Facebook images with thousands of AI comments, and over half the Pinterest interior-design images she searched were AI-made.
• A chat box is the lazy default, not the right interface. She argues chatbots do an okay job across many generic tasks but are optimal for none, and that the year-long belief that user, domain and interface research could be skipped is proving untrue; complex expert work such as medicine or climate will not be served by a chat box. Hanselman adds that developers once taught "don't trust user input" and now build a single text box and trust it.
• Use language models as rubber ducks, not oracles. Appleton says thought-partner use is the most responsible use "at the moment", from an unpublished post she wrote three years earlier. Because models are an "eager improv partner" that won't challenge you, she is prototyping a critical essay advisor that runs narrow steps (list every claim, then check each for evidence) rather than one open prompt.
• Text-only models are unlikely to reach human-like intelligence. She credits her time at Elicit, whose founders expected AGI within five years, for her sense of the field, but says embodied cognition makes the lack of physical inputs a huge blind spot; she allows that a Transformer-scale breakthrough could change that, but sees none now.
• "Stochastic parrot" understates what models can do. Appleton calls the phrase apt but says it makes capabilities hard to reason about, since next-word prediction turns out to be very capable with good architecture and UX. She calls the state of things "capabilities gaslighting": models look like parrots yet pass the GRE at the 99th percentile.

2024-12-19 · open the original ↗

• OpenSauced was built as "GitHub Insights but better". Douglas says it told stories from raw GitHub Archive data, inspired by the CNCF's DevStats, with measures such as a "lottery factor" for how much of a project rests on one or two people and outside-contributor share. He corrects Hanselman: GitHub did try to build this a couple of times, but timing, infrastructure and a business case were not there, and he thinks GitHub is better placed to do it today.
• Its AI features were built expecting GitHub might copy them. A Chrome extension that let users chat with a repo's history via RAG came first; he expected Copilot to absorb it, and Copilot Chat arrived at the next GitHub Universe. StarSearch, for finding open source's "taste makers", shipped in April 2024, built on Azure with GitHub and Microsoft's AI innovation lab; he says he would have been fine with GitHub solving the problem first.
• Enterprise buyers saw it as nice-to-have, not urgent. OpenSauced targeted open source program offices at large companies; Douglas says blue-chip buyers would have needed three to four quarters of near-free collaboration just to get into next year's budget. That made him question whether they had built the right thing and seek distribution partners, which quickly became acquisition talks.
• The Linux Foundation deal was essentially an acqui-hire. From a 15-company list, the Linux Foundation bought the code, IP and trademark; three of the eight staff moved over (Douglas, the head of devrel and one engineer), because buyers wanted the team more than the product. OpenSauced stays up, free with billing removed, as the community top of the funnel for the LFX Insights enterprise product he now works on as head of developer experience.
• The ".pizza" name worked because it was memorable. Hanselman suspected opensauced.pizza was self-limiting; Douglas says companies such as Stripe and Boeing and VCs did not push back, and direct typed-in visits were the top referrer until they started SEO.

2024-12-11 · open the original ↗

• Reading crash dumps is a must-have skill most developers lack. Mark says you handicap yourself without it, especially in native code but in managed code too, since a dump is a static view of your app in a debugger. Hanselman guesses, as his own opinion, that under 10% of developers have the skill.
• NotMyFault and WinDbg let you practice kernel-crash analysis. Mark's NotMyFault installs a signed driver that deliberately crashes, hangs or leaks kernel memory. He opens an old 32-bit Windows XP IRQL dump in the new WinDbg from the Microsoft Store, which pulls Windows symbols from Microsoft's symbol server; its automated analysis skips Windows frames heuristically and names the MyFault driver as the culprit even without its symbols.
• ProcDump captures a dump when a chosen condition hits. Unlike Task Manager's on-demand dump, it monitors a process and triggers on CPU above or below a threshold for a set time, a hung window (added at the Outlook support team's request), exceptions or memory. Mark demos a 10% CPU trigger against the CPU Stress tool; ProcDump writes the trigger reason and hottest thread into the dump as a comment.
• Visual Studio opens the same dump, with a Copilot analysis. Its summary and Copilot analysis said the process was unresponsive, but it did not show ProcDump's comment, which Hanselman calls a missed opportunity. Mark says to use whichever debugger you are comfortable with; Hanselman points to Microsoft Learn's "getting started with WinDbg" tutorial, which debugs Notepad.
• Go one level below your comfort zone. Hanselman says the OS is not a black box: he traced the TikTok Live Studio virtual camera with Process Monitor and deleted its registry key without needing WinDbg. Mark's maxim is "when in doubt, run Procmon"; he adds that blue screens are now very rare in normal Windows use.

2024-12-05 · open the original ↗

• Their deepest difference is culture, not race. Mo, born in Zimbabwe and one of seven siblings, describes a "we culture" of ubuntu ("a person is a person through other people"), where the village and neighbourhood aunties take responsibility for children. Hanselman, with one sibling, contrasts American individualism and moving out at 19; in her upbringing you left home only for university or marriage.
• Presume you are on the same side. Hanselman says an early rule of their marriage, now nearly 25 years, was to assume positive intent when something looks wrong, later joined by "don't attribute to malice what is incompetence". Mo ties it to trust during his years of frequent travel.
• They split on respect versus patience with teenagers. With sons aged 17 and 19, Mo has little tolerance for disrespect and asks whether he would let them speak that way to his parents; Hanselman, citing his mother, treats a selfish teenager as developmentally normal and wants the root cause. Both cite their younger son saying he was hungry and needed space, then apologising an hour later, as a success for each of them.
• Unspoken expectations breed resentment. That is Hanselman's main advice for newlyweds: tiny unspoken assumptions, such as who makes dinner, compound over 25 years. Mo's is to communicate and make couple time a priority; they say chores have since shifted, with him cooking and her doing laundry, without feeling imbalanced.
• Mo wants to give back once the nest empties. After a decade as a nurse following a mid-life career change, she wants more schooling and medical volunteering in less fortunate countries, to "send the ladder back down". Hanselman would take classes, get a personal trainer and mentor people, half-joking about stand-up or a talk show.

2024-12-03 · open the original ↗

• Aspire's strongest case is cutting developer onboarding time. Simmon says a new hire can spend two weeks getting a project running; with Microsoft's eShop reference app (multiple .NET services, Blazor UI, Python, language models) he cloned the repo, found the Aspire app host, hit F5 and had every service running with AI-generated seed data within minutes.
• Aspire removes the networking chores developers dislike. Declaring services lets Aspire manage ports and inject references between them through environment variables or secrets. He says this ends cross-team port conflicts, such as every team defaulting its front end to port 3000.
• Local custom domains should use the reserved ".localhost" domain. At Workleap, with dozens of services, they use custom local domains because heavy localhost use bloats cookies. His Aspire extension edits the hosts file, creates matching TLS certificates with mkcert, and shows the URLs in the dashboard; Hanselman notes the IETF reserves .localhost so no one can buy it.
• Containers need extra wiring to trust local domains and certificates. Containers have their own hosts file and certificate store, so his extension bind-mounts the exported mkcert certificate authority into the distro-specific locations and registers the domains. He says it is a few lines of C#, which he sees as Aspire's main value as a framework rather than a tool, and hopes it lands in a future Aspire version.
• Cloud deployment needs different certificates or forwarded headers. Simmon has not used Aspire's deployment side and says self-signed certificates are not for the cloud; behind reverse proxies, forwarded headers let an app act as if served over HTTPS, and full end-to-end encryption depends on company requirements. Hanselman recalls his banking teams giving up and leaving traffic between containers inside the firewall unencrypted.

2024-11-27 · open the original ↗

• Mark's undocumented-API career began with reverse engineering. He studied a published disassembly of the Apple II ROM in about 1985 and added a computed GOSUB to Applesoft BASIC, which he wrote up in Compute! magazine. His Windows work grew out of his Carnegie Mellon PhD on "application transparent fault management": he tried to apply its checkpoint-and-restore idea to Windows 3.1 and 95, which took him deep into their internals.
• Regmon required reverse engineering the native API beneath Win32. Mark describes Win32 as a facade, one of the "personalities" Windows NT was designed to carry over an undocumented native API. He worked statically, with a disassembler he wrote that merged in the private symbols Microsoft accidentally shipped on an NT 4 beta CD for hardware developers. He also worked dynamically with the SoftICE kernel debugger, at whose maker, NuMega, he briefly worked.
• APIs stay undocumented so Microsoft keeps the freedom to change them. Mark says that once an important app depends on an API, it can no longer change without breaking that app. He accepted the risk because he maintained his commercial tools and simply fixed a diagnostic tool when it broke, and the core internals move slowly anyway. He says Microsoft must document any API its own products use, so there is no hidden trove of exciting APIs. Hanselman's answer was "document everything". Mark's was that internal functions, open source included, are internal for a reason.
• Compatibility shims exist to get enterprises onto new Windows. Hanselman argued that Microsoft gets too little credit for running old software, citing VisiCalc running on 32-bit Windows 10. Mark countered that the credit is the billion PCs Windows runs on: enterprises will not upgrade unless their old IT software comes along. He adds that apps lock themselves to a release in subtle ways, such as checking whether a function sits at a given address. Windows no longer has a 16-bit subsystem.
• Mark's NTCrash fuzzed Windows before "fuzzing" had a name. In 1996–97 he wrote it to barrage the NT system-call interface with garbage, inspired by the crashme tool he had used on BSD Unix at Carnegie Mellon. Microsoft fixed the faults he published. A later version that also hit the kernel-mode Win32 graphics and windowing interfaces found more. When he joined Microsoft he found NTCrash and NTCrash 2 checked into the Windows NT codebase as test tools.

2024-11-21 · open the original ↗

• Onboardbase sells itself as a simpler HashiCorp Vault. Its homepage reads "if HashiCorp Vault was simpler, priced right and just worked". Lex, its CEO, says he used Vault extensively and likes it, but it suits late-stage enterprises with security engineers to set it up. He says the comparison wins customers who already know Vault and need little explanation, and that HashiCorp accounts have signed up without ever contacting the company.
• Secrets leak because developers favour convenience over security. Lex says developers are wrongly assumed to be "secure by default", and they keep secrets in text files, Notion, Google Docs and Slack. He has dealt with a data breach at a former employer. His conclusion is that the secure tool has to be as convenient as pasting a secret into Slack, or it will not be used.
• Developers should work with secrets they never see. By Lex's account, Onboardbase injects secrets in real time into local machines and servers. Per-environment service tokens limit each process to what it needs, and device tokens can be revoked when someone leaves or a machine is compromised. Tokens can auto-expire or rotate after a set number of uses (100, say) or a set time (30 days). Deliberate sharing goes through an encrypted link that deletes itself a few minutes after viewing. His stated goal is to eliminate .env files entirely.
• SecureLog catches the secrets developers leak into logs anyway. Lex says it scans console.log and console.error calls and React server components, then masks or removes the secrets it finds and tells the developer. He says this keeps secrets out of production builds, including the browser's developer tools. Asked whether it would make developers lazy, he said he already assumes negligence and that mistakes are certain. Hanselman framed this as his "pit of success".
• Lex claims every feature gets a week of security testing. By his account, the product gets automated and manual testing, with only the marketing website exempt, and the team uses Onboardbase on all its own projects. He calls friction "the one thing you can always control". He says setup takes five minutes or less and every product he builds ships with a live demo, which he credits with bringing in customers.

2024-11-18 · open the original ↗

• Mark's skincare routine is deliberately minimal. He uses Head & Shoulders and bar soap, shaves with a cartridge razor because electric razors never shave close enough, and uses a moisturizer he shows on screen without naming. He is unsure whether it has SPF. He keeps a short stubble rather than a clean shave to avoid ingrown hairs.
• Hanselman argues men should moisturize with SPF daily. His routine is longer. He co-washes his hair with conditioner and uses Jack Black Double Duty face moisturizer with SPF 20 every day. He also uses Clinique All About Eyes serum, recommended by a makeup artist at Microsoft Build, a Neutrogena dark-spot corrector for sun spots, and retinol at night. If limited to two products he would keep the moisturizer and the shampoo. He says none of the products carry affiliate codes.
• Hanselman says clear aligners suit only light corrections. He got braces in his 40s and says they fixed his bite. He argues that aligners cannot apply enough torque to move a tooth more than a few degrees.

2024-11-14 · open the original ↗

• Islam argues Kotlin took the best of its predecessors. He calls it the newest major language: its first official release came in 2016, against roughly 30 years for Python, Java and C++ and 14–15 for Go, Julia and Rust. He credits it with clean syntax, type safety, elegant null handling, first-class functions and freedom to mix functional and object-oriented styles. It is fully interoperable with Java, including automatic translation of Java code and direct use of Java libraries. When Hanselman raised how much Kotlin borrowed from Scala, Islam said he does not know Scala.
• Google and JetBrains drove Kotlin's rapid adoption. Hanselman notes that the first commit was in 2010 and that Google made Kotlin its preferred Android language less than three years after the 2016 stable release. Islam cites figures he looked up: about 60% of professional Android developers use Kotlin, and it has been downloaded "more than 200,000" times. He adds that it comes built into JetBrains' IntelliJ IDEA. He says Compose Multiplatform is still evolving but is production-ready for Android and desktop.
• The book teaches problem solving, not Android development. Hanselman counts about seven mentions of Android in nearly 400 pages. Islam says the book began as one on computational thinking, and he chose Kotlin over Python for researchers and other non-professional coders. It has 37 projects using only high-school maths, such as calculating Earth's circumference. Its code comes in small chunks that mirror how the problem is broken down. He credits the in-code numbered callouts to his publisher, No Starch Press.
• Kotlin's academic future depends on visualization and ML libraries. Islam used JavaFX in the book but says it can look dated, and he found Julia better integrated with visual libraries. TensorFlow has been ported to Kotlin, but he says many data-mining and machine-learning tools remain in Java or Python. For casual research projects he says Kotlin already has enough, and he calls Python error-prone because it is dynamically typed. He plans to publish solutions to the book's exercises on Medium.

2024-11-13 · open the original ↗

• Mark says language choice divides first on garbage collection. Systems code cannot tolerate garbage-collection pauses, their unpredictability, or the memory that builds up between collections. He says C/C++ and Rust are the only mature non-GC languages, meaning they have the tooling and package ecosystems behind them.
• Azure allows no new C/C++ code, by Mark's own policy. He says he declared the policy a few years ago. Agents on Azure's Boost offload card, which runs Linux and were ported from Windows-based C++ on the host, are being rewritten in Rust, and new ones are written in Rust. For non-systems code he would use C#, Python, Go or maybe Rust. He credits Rust's ownership system: without the unsafe keyword you cannot create a memory-safety problem.
• The ecosystem, not the syntax, decides where a language dominates. Mark says AI research has pushed him into Python for the last couple of years because the ecosystem leaves no choice. Hanselman argues that languages are converging on the same keywords and collections. He offers a half-joke for why Python won ML: it is "more fun to say".
• Hanselman argues languages need not "win" to justify existing. His analogy is Finnish, which suits its speakers, while JavaScript, like English, is an undesigned mash that steals from everything. He also says languages built with hindsight are better, calling Kotlin "demonstrably better than Java". Mark has no favourite language and says languages serve different purposes and stylistic preferences. Hanselman says his own favourite is C#.
• CSnakes lets C# call Python in-process. Hanselman describes Anthony Shaw's tool, which uses source generators to parse a Python file and generate C# bindings, then marshals data across. The goal is to consume Python data-science libraries such as PyTorch from C#. Hanselman says they have that working, with edge cases likely. He contrasts it with earlier projects such as IronPython, which tried to reimplement Python.

2024-11-07 · open the original ↗

• Qiao left Meta because enterprises lacked the means to adopt AI. She spent five years taking PyTorch from a research project to production at Meta's scale. Working with companies such as Walmart, Disney and John Deere, she heard that they lacked the right software, hardware and teams. She judged that serving the whole industry was a bigger impact than staying at Meta.
• Fireworks chose inference over training as the larger problem. Qiao argues that inference demand scales with end users, bounded only by Earth's population, while training scales with the far smaller number of researchers. She says training is a hard business because models are superseded almost weekly and hardware generations now arrive several times a year instead of every two or three years. Fireworks bets on open-weights models plus lightweight fine-tuning. She names latency and serving cost as the two problems, and says startups that reach product-market fit can go bankrupt on cost.
• FireOptimizer tunes inference to each customer's workload. Qiao compares it to a database query optimizer. It shows each customer their own trade-off curve across quality, latency and cost, and lets them choose a point on it. Beneath it sit Fireworks' own PyTorch build, custom GPU kernels, a distributed execution engine and caching. She endorsed Hanselman's analogy of a reverse proxy in front of a model. Speculative decoding pairs a small and a large model, giving the small model's latency and the large model's quality only when the small one's hit rate is high. She says FireOptimizer aligns both models to the workload automatically.
• Qiao says safety starts with model providers, then customization. She cites Meta, for whom Fireworks was a Llama 3.2 launch partner, putting heavy post-training work into guardrails because it has no hosted API to control. She argues that safety means different things in different industries and regions, so customers must be able to adjust the model to their own guidelines.
• Qiao predicts business AI will move to ever smaller models. She argues that business tasks are constrained, not open-ended, so models of a billion parameters or fewer, plus customization, can match trillion-parameter models on narrow tasks. Hanselman added his own view: he is disappointed by the fascination with giant models and expects the interesting work in groups of small models running locally.

2024-10-31 · open the original ↗

• The app survived a failed first build and a full rewrite. Asante, a registered nurse who now manages a healthcare clinic, had the idea in 2018 and funded it by working 70-plus hours a week as a nurse; a pricey team of about 10 in India built version one, then she ran out of money as it developed problems. In 2023 she found a developer in Ghana through her brother, and they rebuilt from scratch after several developers who saw the old code said they could not follow it.
• A non-coder should pick the developer who cares, not just codes. She gave seven candidates access to the repository; she chose the one who was confident about what he saw, had a plan and seemed genuinely invested, where she felt the earlier team was "just slinging code". She lets him make the technical calls (the app is React Native on iOS and Android).
• App Store review rejects one problem at a time; ask for a call. Apple, stricter than Google Play, kept rejecting the app, the last time because a social app must let users report, block, hide or filter others. She says a phone call with a human at Apple explained everything and the app was accepted soon after.
• She wants a safe, structured space more than sheer scale. Conversation is organised around a specific book, film, show or club, so someone avoiding spoilers simply stays out of that room; she builds to her own vision rather than copying Letterboxd or Goodreads, and an admin panel for moderation is in progress. She wavers between a "boutique" community and expecting an entertainment media company to buy it out; Hanselman offered his "thousand true fans" framing (a thousand fans at $100 a year is a $100,000 business).
• Early traction comes from building in public on TikTok. At recording, two weeks after launch, the app had 440 users across iOS and Android, after about 200 in beta; she has close to 5,000 TikTok followers, about half from BookTok, and targets converting 10%. She is bootstrapping and waiting on a grant, has just hired a content marketing specialist in Nigeria, and courts independent authors; books arrive through an API refreshed every 24 hours, while TV and films are still added by hand.

2024-10-30 · open the original ↗

• Know a language only as well as efficiency requires. Russinovich says he can no longer code without AI and writes a lot of Python and PyTorch for AI research without knowing Python well, learning more only when relying on the model stops being the efficient route. Hanselman likens the worry to older fears that syntax highlighting, IntelliSense and calculators would "rot your brain"; Russinovich relays that university professors have settled on students learning fundamentals first, then leaning on tools such as GitHub Copilot.
• A model's tone comes from post-training, not the underlying technology. Russinovich attributes Claude's habit of saying "certainly", which the published Claude 3.5 Sonnet system prompt of July 12 tells it to avoid, to examples in its alignment training. He calls Grok the only GPT-4-class model without safety alignment, and says each model maker deciding what its model will and will not say risks a fractured landscape.
• Hallucination is inherent and cannot be driven to zero. He explains it from autoregressive next-token sampling: trolling in training data gives a wrong answer some probability, and a model that opens with "no" will invent a justification for it; omitting key facts from a summary counts too. Reliable triggers he gives are asking for "Russinovich's 10 immutable laws" (they are Scott Culp's laws of security) and asking for a paper reference, which fails even with web search; groundedness-checking APIs in Azure AI Studio remove only one class of error.
• Jailbreaks steer the model with its own earlier answers. His Crescendo technique asks what weapons the Finnish resistance made, then how they made them, getting Molotov cocktail instructions without ever naming the request. Hanselman compares it to a prosecutor leading a witness and to journalists who coerce a model and then report what it said.
• Guardrails should match the stakes of the deployment. Hanselman argues for barriers where traffic is fast, not in quiet neighbourhoods, and says no one would put a VM on the open internet without a firewall; Russinovich says anyone deploying an LLM must know which kind of risk they are running. His LLM fundamentals paper, covering hallucination, indirect prompt injection and jailbreaks, was forthcoming in Communications of the ACM at recording.

2024-10-24 · open the original ↗

• The Adaptive Controller is a hub, not a finished controller. Jones, a senior product manager for Xbox accessibility, describes 19 labelled 3.5 mm ports that externalise every standard button, plus two unmapped "wildcard" ports, X1 and X2, that can take a 10K potentiometer joystick. It appears to a machine as an ordinary controller with no drivers and stores up to three remap profiles on the device itself, so they travel to a friend's console; Hanselman, who used one after shoulder surgery, calls it "a controller construction kit".
• The new Adaptive Joystick reuses standards people already have. It is a one-thumbstick, ambidextrous device with six buttons that works alone on PC or console, with Controller Assist (formerly Copilot) or plugged into the Adaptive Controller. Its 1/4-20 tripod thread lets it be mounted for chin, foot or wheelchair use, the same logic as the headphone-jack ports; due early next year at recording, it is priced at $29.99 because users already spend heavily on medical and assistive devices.
• Adaptive Toppers let players 3D-print their own thumbstick grips. Announced in August through Xbox Design Lab, the tool offers about six topper shapes, adjustable height and width, and a free STL file for the Adaptive Joystick, the Elite or a standard Series X|S controller. Jones says Elite support mattered most because many players with disabilities use it; Hanselman notes prints take minutes and a library can make them.
• The product came from outside hackers brought into Microsoft. Jones was an occupational therapist who used games in treatment, and her father's nonprofit, Warfighter Engaged, an all-volunteer 501(c)(3), soldered adapted controllers for veterans with limb loss. A Microsoft engineer saw its Twitter feed, hackathons followed, and those partnerships fed into the Adaptive Controller; her father plans to stop custom designs at the end of the year, hoping commercial products like the joystick fill the gap.
• Accessibility work has no end point and is spreading to standard hardware. Jones wants inclusive features in every future Xbox device and hopes tools combining head tracking and voice can integrate with the platform; Hanselman discloses he is an investor in one such tool, Cephable, free for individual use. She cites Logitech's adaptive kit, with 12 buttons and boards for $100 against about $60 for a single assistive button before it, and both argue that ageing players need easier modes too.

2024-10-18 · open the original ↗

• Every unfamiliar tool is a fight, so he values stability. The recording started about 20 minutes late because St Pierre's mic and camera would not work with Hanselman's usual recording tool, and they never learned why. He puts blindness's daily cost at about 30% of his capacity, citing a replacement headset that changed how his screen reader pronounced things, and says this is why he stays long at companies and likes maintaining software.
• Adopting a screen reader late cost him; now it speeds him up. Legally blind from birth with a degenerative disease, he was told around age five he would be blind by 20, and lost his central vision in the last few years; he has used a screen reader for about three to three and a half years. He says he should have started five or six years earlier, took a year to accept it, and now runs it at 30% of "boost" speed, reads documentation faster than ever, and ends days far less exhausted.
• A pinhole field of view trained an unusually strong working memory. Even when he could see, he took in only two or three letters at a glance, so he learned to guess words and hold code in his head. He says he can answer questions away from the computer with "that's in that file near line 500", which Hanselman calls keeping the code's object model in his brain.
• Tooling, not syntax, makes a language accessible. He moved from .NET to Go after a year of Node, and credits Go's simplicity, helpful compiler and enforced formatting with tabs, because spaces must be heard. Since he cannot scan a stack trace or a block of JSON, he relies on linters; YAML whitespace has cost him hours. He left Linux because its screen reader did not work for him and now uses Windows with WSL, calling Windows Terminal and VS Code very accessible.
• He is bootstrapping a SaaS business after a small exit. He built the main system at a company that grew from 10 to about 35 people and was acquired by Equifax in 2023, and has been trying to self-fund a product since 2008. Kiwipolis is a homeschooling planner with a 2D game element; the game is a canvas and not accessible, so he debugs it with shortcut keys that skip collision detection. He says reluctance to travel, not blindness itself, has kept him from conferences.

2024-10-16 · open the original ↗

• Visible inconsistency makes customers assume the internals are worse. Hanselman walked out of an EV test drive after spotting three chained Android tablets, each in a different font, which to him exposed which group shipped the map, the climate controls and the speedometer. Mark agrees the visible experience is what people infer the rest from, and says that inconsistency lost the carmaker a sale.
• Consistency at scale comes from shared standards, not one tastemaker. Mark rejects relying on a Jobs- or Gates-style figure and points to Azure Resource Manager, whose plug-in resource providers must follow API guidelines published on GitHub, checked by an API review board and tooling. He concedes Azure is not perfect and its older API surfaces are less consistent.
• How much an inconsistency matters depends on whether it can be undone. Hanselman's example of Copilot shown as a bot, a sparkling pencil and a magic wand in Visual Studio can be fixed next release; Mark says inconsistent APIs are locked in by backward compatibility, cause developer mistakes, and force every improvement to be built several times. Azure moved fast at first and standardised once a better pattern emerged.
• Backward compatibility is the enterprise promise, at the cost of polish. Mark is the arbiter for breaking-change and service-deprecation approvals, under published policies and formal announcement channels, because breaking an enterprise app costs money and blocks upgrades. Hanselman, stressing this is his own opinion, argues Windows keeps compatibility and so sometimes ships the org chart, while Apple breaks things unapologetically; Mark calls Hanselman's roughly 20 years of running his blog on Windows Server 2003 "incredibly insecure".
• Enforce standards with guardrails, not documents. Mark lays out a spectrum from no rules to written guidelines, to auditing, to controls that make non-compliance impossible, such as TypeSpec, whose inherited top-level definitions make new resource providers compliant automatically. Controls are costly and must not add too much friction, so he reserves them for high-stakes areas; Hanselman sums it up as "Word documents don't break the build".

2024-10-16 · open the original ↗

• Even with authority, you must sell the vision to lead. Mark says he learned this as co-founder and chief software architect of Winternals, founded in 1996 and sold to Microsoft in 2006; as Azure's CTO he lacks direct authority over most of what he drives. His influence has grown since he joined Azure in 2010 through a track record of being right, and he still has to explain why each thing matters.
• Customer advocacy should occasionally get you in trouble. Hanselman says that if you do not get in trouble twice a year for advocating for the customer you are not pushing hard enough, and that he uses his seniority to shield team members who irritate others for the right reasons. Mark qualifies it: some advocates come across as constantly abrasive, so there is a line to find.
• Stalled priorities usually lack an owner, not agreement. Mark says that when he escalates to senior engineering leaders it is rarely because people ignore something; everyone agrees it matters but no one is accountable, or no one can reprioritise without someone who owns the resources reallocating them. He relays Satya Nadella's message to Microsoft's executive staff that your job is to fix your complaints, not voice them, and Hanselman recalls his boss Scott Hunter telling him that "someone should do something" now means him.
• Understand at least one level below where you operate. Mark says writing Windows Internals taught him to go a level deeper than his subject so he could be accurate and abstract it, which is why he digs deep in meetings. He offers Nadella forwarding Hacker News threads and technical blog posts, and Hanselman offers Hunter finding product gaps while preparing keynotes, as leaders doing the same.
• Engineers in silos lose the standing to influence the product. Teaching internals with Dave Solomon, Mark found the owner of one Windows component knew little about memory management or scheduling, and he says many Azure engineers have never seen their own service's portal or SDK experience. Knowing how a component fits the end-to-end customer experience, he argues, is what lets people propose improvements credibly; Hanselman calls this being a T-shaped developer.

2024-10-10 · open the original ↗

• Staying bootstrapped keeps the company focused on DNS alone. Eden says DNSimple has never taken outside money, whereas VC-funded or public rivals must keep adding hosting, email and marketing products to grow. He argues customers undervalue domains as a commodity, partly because big players commoditised them, and that focus is how a small firm keeps innovating; he is happy to keep running it as a boutique, while conceding "there's always a price".
• Domain takeovers mostly come through social engineering. Eden names social engineering as the most common takeover route and says DNSimple offers two-factor through hardware keys or one-time passwords rather than SMS because of SIM swapping. His advice is a registrar with strict policies against acting on a caller's behalf, automatic renewal, multi-year renewal for important names, and access for more than one person so a domain is not lost when someone leaves.
• Owning a top-level domain lets its owner set the rules. Asked why Microsoft is moving cloud services to its .microsoft TLD, Eden infers it removes a namespace outside Microsoft's control and lets it impose requirements from the top level, as Google did by enforcing HSTS across .app and .dev. He notes DNS was designed in the 1980s for a gentler internet and remains a regular DDoS target.
• Fast record changes come from custom sync and low TTLs. Hanselman, a customer since about 2011 who notes this is not a sponsored show, says changes on DNSimple appear almost at once. Eden credits an internal system instead of the standard DNS transfer protocols, pushing changes to eight points of presence with Australia as the benchmark, constant measurement and several redesigns, plus low default TTLs where registrars often default to a day; he wrote the core name server in Erlang, built to fail hard and restart cleanly.
• The domain control plane accepts that customers use several providers. It connects DNSimple to Route 53, Azure DNS or on-premises servers such as CoreDNS and manages zones bidirectionally, so they can be moved between providers. Eden says it was built within the last year and a half, is still early, and exists because DNSimple is "too small to force people to do anything".

2024-10-09 · open the original ↗

• The show aims to record nuanced disagreement, not loud opinions. Hanselman says his conversations with Mark sometimes leave him better off and should have been recorded, and that tech discourse lacks people who disagree and end at "I see why you think that", unlike streamers broadcasting a loud opinion about JavaScript. Mark says the format grew from the tangents in their joint conference presentations.
• It will ship as video and audio, starting a week later. The podcast was to be released on Hanselman's YouTube channel and wherever podcasts are available, with the first episode due a week after the trailer. They joke that most podcasts die around nine episodes after a run of apologies for irregular releases.

2024-10-03 · open the original ↗

• EGA was chosen because it was made for the screen. Minamata says the 16-colour palette of the late-1980s to early-1990s DOS era gives a crisp, unified world, where early VGA adventure games scanned painted backgrounds that clashed with their sprites. The game runs at 320x200, and she rules out moving a sequel to VGA because she sees EGA as the format she wants, not a stepping stone.
• Deliberate imperfection makes digital pixel art feel handmade. She avoids mirroring faces or objects because perfect symmetry looks stiff and uncanny, a rule she also gave artists as art director on Recommendation Dog. Dissecting Sierra screenshots showed her that their perspective is "a magic trick", since sprites never scale from front to back, and she copied it; she drew in Photoshop only out of habit and says beginners do not need it.
• The text parser is a deliberate choice despite costing audience. She says typing forces players to read and gives a sense of limitless possibility, with no dialogue trees, though it rules out a console port and made programming harder than point-and-click would have. Player feedback led to abbreviated commands for people with dexterity limits, plus an optional notebook, an online hint book and a choice of graphical or text inventory.
• An artist-first developer shipped it on community-built tools. A former magazine and newspaper illustrator with work in the New Yorker, she contrasts artists' acceptance of grind with programmers' shortcuts, citing pasting a link 71 times by hand for her newsletter. She built the game in Adventure Game Studio, a free, open-source, community-maintained engine with a built-in parser; it was part-time from its first demo in November 2018 until the last two years before its August launch, and getting builds onto Steam was the closest she came to giving up.
• A frontier mining-town setting suits a phone-free cozy mystery. The game is set in northern Ontario in 1914; she chose the golden age of Agatha Christie-style detective fiction, a single lodge setting and an era without pocket cameras, and tied it to Canada's push into resource-rich frontiers that left mining ghost towns. She has extensive notes for a sequel, hopes for a trilogy with the same EGA look, MT-32-style music and parser, and expects, warily, that it will take less time.

2024-09-29 · open the original ↗

• With scaling ending, gains must come from whole systems. Hwu, a senior director of research at Nvidia who taught at Illinois for 34 years, says Dennard scaling (lower voltage and power and faster switching as transistors shrink) mattered more than Moore's law: for about 15 years code got faster untouched. Dennard scaling ended in the early 2000s, and he sees "at least another two generations" of Moore's law. He expects the next advances in networking, moving from electrical to optical links that cost more than 10 times their copper equivalents, in primitives for synchronising huge numbers of machines, and in cooling, where data centres are "at the cooling limit".
• Research has to target the technology of ten years out. In the 1980s his Berkeley group under Yale Patt bet that abundant fast transistors could find parallelism no processor then detected. Intel's P6 (Pentium Pro), designed by 1993, embodied those mechanisms, and he draws the lesson that ideas take five to ten years to reach hardware.
• Python frameworks throw away parallelism that lower layers must recover. He cites the Python interpreter lock, and GPU work launched from Python frameworks such as Torch as separate, isolated API calls. He expects runtime techniques such as fusing those calls on the fly, long researched, to be deployed "in the next decade or so" because other optimisations are running out.
• He predicts a new processor unit built for retrieving data. People buy processors "reluctantly" to run applications, and he argues deep-learning models bypassed going straight to the data. His proposed unit would tolerate latency and pull, for example, one kilobyte out of four petabytes in real time to feed models. If the industry "does the right things", he expects within a decade a pocket device holding a person's lifetime of memory, and says he will retire when he has one.
• Enough value makes a missing software ecosystem appear. In 2011 an NSF review panel called his GPU-based Blue Waters supercomputer proposal "risky to say the least and probably irresponsible" for lacking math libraries; he answered that GPUs were "a small puppy with big paws". Thirteen years later those libraries exist. He describes the follow-on Delta project as filling a shortage of GPUs for academic researchers, not as pushing the frontier.

2024-09-19 · open the original ↗

• Staying a one-person shop trades polish for time. Finney, a Microsoft MVP and former mechanical engineer of 10 years who builds apps beside a full-time job, says time is his scarcest resource and finding the right hire could take months with nothing shipped. He accepts that design and architecture "suffer a little" as long as the result is good enough, and he picks whichever framework avoids fighting each app's one hard step.
• Data-loss bugs are the ones he will not tolerate. He reasons that large companies ship bugs too, so the fear he prioritises is Ink Calendar users losing their journals. In app messages and review replies he writes "I" rather than "the team" so users know it is just him.
• App store reviews give users no channel for dialogue. Text Grab users file GitHub issues with line numbers. Ink Calendar, found through the store, draws one-star reviews about Store download and update failures that have nothing to do with the app. Hanselman suggests developers review more kindly than consumers, and Finney agrees.
• Store packaging, not coding, is what eats an indie's time. For his React Native app he pays for Expo's hosted service, where "EAS build all" builds, signs and submits to both stores and generates icons. On Windows the toolchain fails midway or at certification, and the MS Store CLI has not worked for him. He defends the stores' cut (70 cents on a dollar) as covering hosting and distribution he could not run himself.
• Giving Text Extractor to PowerToys paid in exposure, not money. He contributed it when a GitHub issue proposed the feature, keeping a narrower scope while Text Grab stayed his more flexible app. Users still come to Text Grab from it, it led to his MVP award, and the team later built that functionality into Snipping Tool, which he calls "inevitable".

2024-09-12 · open the original ↗

• Battery life comes from an efficient Arm ISA plus dedicated accelerators. Van Doorn, a Qualcomm senior vice president, cites 19 hours for the Dell XPS and 26 hours for HP on Snapdragon X Elite. Running Recall on the NPU costs about 30 minutes of battery against 6 hours or more on a GPU. He credits Arm's later start with shedding x86's 40-year legacy, which he notes is also x86's strength.
• Qualcomm had to unlearn an IP company's caution to court developers. He says he joined two years ago and wrote an internal "open manifesto": a cell-phone business with a handful of customers does not scale to the thousands of developers in PC, XR, IoT and auto, so they need "no touch or low touch" enablement without licence terms. Legal and licensing had to change too. Engineers enabled Linux on X Elite outside the product plan and open-sourced it, and Qualcomm worked with Linaro and Microsoft on LLVM for Windows.
• The PC part is not a phone chip. It shares building blocks such as CPU cores with phone parts, but memory channels, accelerator counts and the NPU's processing elements are scaled per target. He says the PC chip is too big and too expensive for a phone.
• NPUs win on power but cost the most developer effort. NPUs compute in integers, which need far fewer transistors than GPU floating point, but models trained in floating point must be quantised, which he calls tricky and slow. Efficiency runs NPU over GPU over CPU, and ease of bring-up runs the other way, so teams pressed for time to market start on the CPU. He reports Llama 7B and a 1 to 1.5 billion parameter speech-to-text model running on the NPU.
• Local AI will run hybrid: on-device first, cloud as fallback. He gives privacy, response time and power as reasons, since cloud GPUs are expensive and power-hungry. Models of 7 billion parameters run fine on device, while 80 billion is "more tricky". He says the architecture could shift again if one large model gives way to many small ones under an orchestrator, and warns that hallucinations mean users still need expertise.

2024-09-05 · open the original ↗

• AI serves human desires and cannot supply them. Washington, a .NET programmer, Microsoft MVP and author of books on AI, says LLMs seem alive only because they manipulate language and humans think in language. His grand-niece at one year old, who wanted her block back, was "so much more advanced" than ChatGPT. AI will not take jobs, but "the people who use AI may", like a person with a sewing machine.
• Whole books from AI fail; single paragraphs succeed. Asked to write a whole book, the model produces work he grades F, against Hanselman's B-minus, and feeding in two books to get a third "will never ever... work". His free, open-source AI Story Builders writes one paragraph at a time that he then edits. He reports about 1,200 words in a 15-minute session, which he says was not possible before.
• Tight grounding in each prompt keeps paragraphs on the rails. Without it, a flashback to a five-year-old has "little Timmy" driving to an appointment. The app uses retrieval-augmented generation to put character attributes, age at that point, location, season and weather into the prompt. It also extracts new characters and attributes from paragraphs he writes, but he stresses a human still builds the database.
• Fine-tune for behaviour, never to add content. In what he calls his own opinion, which others dispute, new information belongs in RAG, while fine-tuning should only change behaviour such as writing style. After about three chapters, a button in the app produces a fine-tuning file. He uploads it to OpenAI and gets back a model in about an hour that writes closer to his style.
• Plain text files are enough to store embeddings. To keep the app free without paying for databases, he saves embeddings as text files. The Blazor MAUI app in the Windows store keeps them in app storage. The Blazor WebAssembly version zips them into browser local storage, and both run the same code.

2024-08-29 · open the original ↗

• The new Nature of Code is a full rewrite, not a refresh. Shiffman says the class began at NYU's ITP around 2008, and he self-published the book through an early Kickstarter, then left it largely unchanged for about 10 years. He brought in No Starch Press to "project manage" him and produce a better print object. The new edition is in JavaScript with p5.js instead of Java and runs to about 640 pages.
• Neuroevolution is the book's new centerpiece. The 2012 edition only covered a single perceptron. He rewrote that chapter around ml5, an open-source machine-learning library for p5 that he also works on, and added a chapter on evolving the weights of many random networks instead of using backpropagation. He calls it a lightweight way to play with machine learning in the browser with no dependencies; Hanselman notes the book builds from a dot moving at random to a Flappy Bird that learns to flap.
• Everyone should get the chance to code; not everyone must. Shiffman argues that understanding code is empowering when so much of life runs on software. He says teaching through art and creative exploration, with "no answer to this problem", brings in people who thought coding wasn't for them. He also says the book is not a learn-to-code book: it assumes variables, loops, conditionals, arrays, functions and objects.
• Giving the book away free sells it rather than cannibalizing it. The full text stays online under Creative Commons, and he calls the website marketing that widens the book's reach, while admitting he has only anecdotal evidence. He describes the free material as subsidized by his full-time NYU teaching job, since he counts open education materials as part of that work.
• p5.js reimagined Processing for the web, community first. Shiffman credits Casey Reas and Ben Fry with starting Processing in 2001, building on John Maeda's Design by Numbers, and calls its play button for running code revolutionary. As Java applets died, Lauren McCarthy rebuilt it for the browser and broke down the line between users and developers. The p5.js web editor hosts every example so code can be changed and shared from any machine, and he invites readers to port the examples to Processing, openFrameworks, Unity and other environments.

2024-08-29 · open the original ↗

• Treat Copilot as a rubber duck, not a one-shot oracle. While upgrading the Hanselminutes site to .NET 8, Hanselman says most people ask one question and stop if they dislike the answer. He builds up context instead with follow-ups ("tell me more about the setup method", "is this modern idiomatic .NET"), and keeps separate named chat threads for each piece of work.
• Scope what Copilot sees. He shows slash commands such as /explain working on selected code, and inline chat marking the exact scope it is reasoning over. He then widens that scope to a file plus its tests to ask whether the tests are affected.
• Generated commit messages are the standout feature. He expected the generated message to say only "added new tests". Having read the diff, it also listed earlier Dockerfile and .NET hosting changes he had forgotten. He calls this "the killer feature" and says he would have missed those changes himself.
• AI tools should show where they looked. In his demo, analyze-and-fix on a failing test immediately found the deliberately wrong assertion (690 expected, 694 actual), and ask-Copilot on a null-reference exception pointed to the line he had broken. He praises it for listing the references it used and argues no AI app should be a black box, so a user can see why it answered and correct it.
• The small "delighters" go unnoticed, and AI icons are inconsistent. He shows rename suggestions and conditional-breakpoint suggestions (it proposed "shows greater than 10" from context). He complains that the AI is marked with a sparkle in one place, a sparkle pen in another and a Copilot figure in a third, and says the industry hasn't settled on one icon. He recommends Help > What's New as the way to find such features.

2024-08-22 · open the original ↗

• A technical product chief reads customers better. Frederick, chief product officer at Stanza Systems, says engineering experience lets her tell whether a pain point is coming from product, platform engineering or SRE staff detached from business goals. She says a CPO without it is at best slowed down and more likely hindered.
• CI/CD maturity should fit the business, not be chased. She cites a team that pushed for faster releases while its heavy manual testing burden made each release more costly. She argues that an early-stage product still finding its business may not repay full test automation, and Hanselman concedes this challenges his instinct to automate anything done twice.
• Stanza moved its insights product ahead of schedule. Frederick says customers confirmed the problem behind Stanza's first product, rate limiting, but showed little urgency in a cost-focused market. They had also already paid heavily for observability and would not add another tool. So Stanza decided to draw insights from data customers already collect, a step it had planned for later.
• Tooling sped up building software but not running it. She argues that small teams can now produce far more software than they can operate, and that this leads to burnout. On "shift left", she agrees with Hanselman that shrinking budgets often just dump right-side work (performance, reliability, resilience) onto developers, "borrowing" development time.
• Stanza wants to codify senior engineers' judgment. She frames operations as signals, comprehension and actions, and says signals and actions are well served while comprehension still relies on senior people sifting the data. Stanza aims to systematize that triage into prioritized clues, such as flagging a service that has become risky. Hanselman adds that this should make junior engineers better faster without letting them stop thinking, and she agrees juniors learn most by watching how seniors reason.

2024-08-15 · open the original ↗

• Kubernetes' declarative state made bespoke deployments obsolete. Garfield says Codefresh bet early that containers would change everything and that Kubernetes would win. Declarative state management, he argues, turned hand-rolled, special-case deployments into something repeatable and automatable.
• GitOps was formalized to stop it meaning everything. He relays that the term was coined at Weaveworks by Alexis Richardson after git-stored state allowed a fast recovery from an outage. A CNCF working group started around late 2020 with Codefresh, Azure, GitHub, Amazon and Weaveworks, then Red Hat, drawing in about 80 companies and 120-plus parties. It reached 1.0 of four principles by about 2021: git as the source of truth, plus a reconciliation engine that monitors live state for drift.
• Argo CD has effectively won Kubernetes deployment. Garfield estimates that a little over half of Kubernetes deployments use Argo CD, against about 10% for the runner-up. He says the figure is extrapolated from a CNCF survey and is not exact. Codefresh launched the first commercial Argo CD around 2020 and calls itself the top security contributor to the project.
• Octopus Deploy bought Codefresh for cloud-native reach and funds its open source. The acquisition closed in March after the Octopus founder met the team at KubeCon. Garfield says Octopus is strong on VMs and Windows/.NET deployments, where Codefresh was weak. Octopus committed to expanding the open-source team, and he has been hiring additional Argo CD maintainers.
• Kubernetes at the edge is everywhere and cheap. Garfield names Starbucks and Chick-fil-A stores, F-16s and container ships, and argues a two- or three-node cluster survives node failures. With GitOps, a disconnected cluster picks up its updates when it reconnects. He compares Kubernetes to snowboarding: painful on day one, then easier to master than the alternative. He warns that the power to bootstrap everything from git is also the power to destroy everything in one command, a topic for ArgoCon North America (12 November, Salt Lake City, $200 academic price).

2024-08-08 · open the original ↗

• Tech creators already do the work and deserve pay. Freeman, co-founder of the influencer-marketing firm Freeman & Forrest, says influencer marketing is huge in fashion and beauty but has not landed in tech. She argues that creators already raving about products should be paid, just as the industry accepted paying open-source maintainers. The condition is authenticity, summed up in a line she credits to Corey Quinn: "you can buy my attention, you can't buy my opinion."
• Measure impressions, engagement and conversion according to the goal. Her firm measures all three, then hands converted users to the client's own funnel. A cash-limited startup needing signups gets a conversion focus, while trust-building campaigns weight engagement. She avoids leading with Twitter's "empty calorie impressions" and favours smaller, high-trust channels such as Cassidy Williams' newsletter.
• Paid creators extend developer relations' reach. From her developer relations work at Microsoft and AWS, she says in-house advocates exhaust their reach quickly and see returns collapse when they repeat themselves. She adds that she always resisted companies' requests to list influencers to target, and that payment lets "tinkerers" afford to keep experimenting.
• Lasting careers need substance, not just shine. Hanselman complains about engagement-bait influencers who ask generic JavaScript questions. Freeman says questions are fine and the bigger problem is gratuitous meanness, which she calls a five-year run at most. She describes visible tech people on a spectrum from shine (communication, authenticity) to substance (real coding experience) and says going too far toward either end causes problems.
• Today's newcomers can't replay the golden-age paths. Freeman says she entered code school and developer relations in their golden ages, and that people now trying to replicate that era misread a changed world. She credits years of "spinning in mud" for her grit. Hanselman, who admits his own career has been lucky, advises going one level below the abstraction you know, for example JavaScript beneath React. She endorses this as advice that at worst does no harm.

2024-08-02 · open the original ↗

• Voice is the friction-free interface, and the obstacle is human, not technical. Russell, chief AI officer at the AI Leadership Institute, says she saw this when her son with Down syndrome and her father with a traumatic brain injury began talking to devices. She argues that letting users talk to your tech means they never return to your website. She says companies are slow to exploit voice because teams protect what they built, not because the technology is unready.
• Alexa is primed for a generative-AI reboot. Hanselman calls the device a dumb terminal. Russell says her 100-plus Alexa skills reach about 2 million unique users with unchanged code, such as an 800-line daily-affirmation skill that walks through 365 entries. She predicted the new generative Alexa was "weeks away" and said it would let skills run in 40 languages without extra developer work. She also says older (75-plus) and younger (under-20) users tolerate voice errors far more than people in between, and that generative AI lets devices fail gracefully.
• Voice AI's business case runs through loyalty, not direct revenue. Hanselman asks how a $59 device can sustainably call an expensive model. Russell points to Alexa team layoffs and to Bezos moving on from his Star Trek computer project, and says Alexa took about 10 years to get barely into the black. She says Alexa owners buy more on Amazon in general, not through the device, and argues builders need values that sustain them through costly retraining.
• Anthropomorphized wake words helped adoption but carried costs. Hanselman dislikes "yelling at invisible women" and welcomes unnamed products such as ChatGPT and Copilot. Russell argues the human name probably helped adoption, says "Alexa" was picked to minimize false wakes, and notes girls named Alexa were hurt, with mothers attempting a class action. She sees a return to push-to-talk in ChatGPT voice and the Humane AI pin.
• Disclose AI and recording up front, or lose more trust. Russell cites a Stanford study finding that using AI always lowers trust, but far more when people discover it afterwards. She says no legislation currently protects a person's ownership of their likeness, and criticizes a realtor-targeted voice product (captioned as "air") that never discloses it isn't human. Her prescription is to build responsibly even if it is slower, and to use AI red teaming with diverse people who judge experiences, not only adversarial attacks.

2024-07-26 · open the original ↗

• Expert readers do not read books linearly. Matuschak cites ethnographic studies of highly cited professors, who skip from contents to index, preface, chapter openings and figures before choosing what to read; people with reading-strategy skill but no domain expertise can still comprehend hard texts by drawing inferences across distant passages.
• Lectures and podcasts transmit attitudes better than facts. He argues a lecture is a poor way to acquire specifics such as how to calculate torque, but a podcast gives a "portal" into how someone thinks, building tacit scenario templates; a single sentence can change a belief only by tipping a cascade of the listener's own reflection.
• Quantum Country embeds spaced-repetition review in the text itself. The quantum-computing book he built with collaborator Michael Nielsen poses quick questions as you read, then emails you days later to review for a few minutes; he says recalling material when you can only just do so repeatedly extends how long it is retained (the Flappy Bird image is Hanselman's).
• Base knowledge still matters even if AGI arrives. Asked by Hanselman why his son needs logarithms, Matuschak gives two reasons: you often cannot know what to look up without already holding the pieces, and fluency frees working memory ("seven plus or minus two"); a physics student he tutored was struggling because algebra was not yet automatic.
• Many linked representations beat matching a "learning style." He says the evidence does not support learners being fixed visual or symbolic types, but understanding grows with each representation learned and linked; teenagers remember Pokémon better than the periodic table because of richer connections, not choice, and he argues the real gap is having no visceral reason to know.

2024-07-22 · open the original ↗

• The monitor was supplied free; the review is unpaid. Hanselman says he normally reviews only hardware he bought himself, and made an exception because BenQ pitched the RD280U as a monitor specifically for programmers.
• The 3:2 aspect ratio buys more lines of code. At the same width as his 27-inch 16:9 Dell (3840×2160), the 28-inch BenQ runs 3840×2560, which he calls "4K plus": 400 extra vertical pixels at the same DPI, which he estimates at roughly 20–25% more height.
• Contrast is the most immediate difference he sees. With no calibration, he says blacks are black and dark mode is genuinely dark, while his seven- or eight-year-old Dells look washed out beside it.
• Coding presets tune the panel for dark or light themes. A front-accessible button cycles coding modes for dark and light editor themes; he also shows "Moon Halo" bias lighting on the back, DisplayPort, HDMI and USB-C inputs, a USB hub with KVM, and that it hung on his existing VESA wall arm without changes.
• His verdict is provisional and the price reasonable. He calls $599 a significant upgrade for the money, and says he will use it for a couple of months, judging eye strain, colour temperatures and the bias lighting, before giving an update.

2024-07-18 · open the original ↗

• DevRel works only as a trusted source, not as marketing. Jones, VP of developer relations at TBD at Block, says developers are allergic to traditional marketing, and she has recommended competing tools when they were the best fit; Hanselman frames this with his radio-ad line "you've got a friend in the diamond business."
• She won't promote products she doesn't believe in. Both acknowledge this comes from a position of privilege; she argues internally against weak products, and tells employers that pushing one would cost trust and make her less effective afterwards.
• Bringing community feedback back inside is core work. She says the job is working with the community, not talking at it, and reporting pain points back to engineering and marketing; she argues that sharing this is what earns respect for teams whose travel can look like a party.
• Developer-tools advocates should have been developers first. For SDK and API work she says you need to know developers' real constraints (her example is teams stuck on Java 8 or 11 because upgrading a bank system needs buy-in); she frames it as "the easier path," not a gate, and has juniors on her team but would never hire one as a sole advocate without senior mentors.
• Advocates should specialise rather than do everything. She and Hanselman agree that building on each person's strengths beats demanding a Swiss-army-knife of video, samples, docs and social; she negotiated her first title from "developer evangelist" to "developer advocate" because evangelism implies forcing a belief.

2024-07-11 · open the original ↗

• Feed algorithms reward frequency, which makes quality stand out. Rose, who blogs at samwho.dev, says curation algorithms reward daily posting and LLMs make low-quality volume easier to produce; he posts rarely and avoids naming specific technologies (load-balancing algorithms, not a particular implementation) so posts age well.
• Raw web standards are his bet on longevity. He uses web components without helper libraries and no framework, with TypeScript the only dependency he worries about over 10–20 years; PixiJS (about 600 KB, of which he uses a sliver) may give way to his own 2D renderer, which he says can recreate simpler visualizations in under 10 KB.
• Readers don't realise the visualizations are clickable. He takes at least half the blame: through several rounds of watching friends scroll past, a yellow button and then a static cursor both failed, and only an animated cursor labelled "click" worked, teaching him that people skim rather than read top to bottom.
• His posts are simulations, not drawn animations. Each widget is a configurable web component over domain-modelled classes (load balancers, servers, requests); randomness in the load-balancing post cost him evenings waiting for the right behaviour, so the queuing post he was paid to write for Encore's blog uses uniform timings, a realism trade-off he dislikes.
• The web should move beyond text for teaching. He calls himself a visual learner despite the research Hanselman cites against learning styles, says video has never worked for him, and, echoing Andy Matuschak, argues interactive media could make educational material 10x better than text.

2024-07-04 · open the original ↗

• .NET Aspire is a bundle, not a framework. Edwards, a principal architect at Microsoft, describes it as templates, NuGet packages, a development-time orchestrator and full adoption of OpenTelemetry, together aimed at building distributed apps; he treats "cloud native" as a term that means different things to different people.
• Good abstractions let you punch through them. He says Web Forms' real sin was making it too hard to escape the abstraction for the 20% case; Aspire's high-level calls sit on ordinary lower layers (adding Postgres just adds a container and injects a connection string as an environment variable), and F12 shows the source.
• Components encode opinions developers would otherwise rediscover. Using Redis as the example, he says an Aspire component picks the client (StackExchange.Redis), registers it in DI with a set lifetime, binds options to configuration, references a community health-check package and wires up OpenTelemetry, all on by default; before Aspire, that client's logging took a raw TextWriter.
• The AppHost describes the whole app in C#. It extends the host-builder pattern to containers, executables, projects and parameters with references between them, and running it starts everything and tears it down; he concedes the top early feedback was that this lifetime is too simple, and that keeping containers running between sessions is a hard problem being worked on.
• Deployment changes are optional. He stresses the orchestration is mainly for development and an existing deployment can stay untouched; alternatively a deployment manifest feeds the Azure Developer CLI, tools Amazon is building, or the community "Aspirate" tool for Kubernetes.

2024-07-01 · open the original ↗

• A pump can be built for a hundredth of commercial cost. De Bock, an associate professor at the University of Otago and a paediatric endocrinologist, says the low-cost pump programme (LEAP) began in 2018 when a University of Canterbury mechanical engineer refused to proceed unless the device cost 1/100th the price; a national science challenge grant funded both the hardware and qualitative work on how people would view a non-commercial device.
• Pumps are essentially precise syringe drivers. He says the hard part is accurate, robust plunger delivery tested against international accuracy standards; he argues a cheap pump that lasts less long is more forgivable, like a cheap drill, unlike a several-thousand-dollar unit a parent fears a child will break.
• Automation ends the trade between quality and quantity of life. Injections can reach healthy glucose only with a very restrictive life, he says, citing data that people with type 1 make over 100 health decisions a day and describing episodic "diabetes burnout"; he endorses Hanselman's lane-keeping car analogy for the pump-sensor-algorithm loop.
• Technology acts as a health-equity equaliser. His team's cross-sectional study of everyone in New Zealand found CGM use removed the outcome gap predicted by socioeconomic status or ethnicity, including for Māori; he expects Chinese and South Korean entrants (10–14-day sensors) to push CGM prices down to compete with test strips, while much of the global south still gets only syringes and finger pricks.
• Don't build one and wear it yet. He says a study putting the pumps on people was about to be published, open access, but they are untested against everyday wear; his advice is to build one out of interest and pump into something other than yourself, since safety still has to be ironed out.

2024-06-20 · open the original ↗

• Older developers should mentor with "yes, and", not stop energy. Hanselman, now Microsoft's VP of developer community, says his identity is teacher, not company, and his goal is making early-career people succeed, including 40-something career switchers. Elders offer kind, patient mentorship and historical context, and should answer a junior's idea with "here's what we learned when we tried" rather than "we did that before".
• Information is easier to get, but getting noticed is harder. Hanselman stands by his "dark matter developers" post: for every social-media star there are a thousand people doing good work, and being noticed may not matter. Social media is "algorithmically designed to make us feel bad", so the only comparison worth making is against yourself yesterday.
• Don't make juniors earn the high level by learning everything below. Hanselman compares insisting on assembler and dump debugging to making his 16-year-old learn stick shift when 3–4% of American cars have one. His standing advice is to go one level below your comfort zone (V8 for JavaScript, the CLR for .NET), and he tells a story of a Black Girls Code student who debugged his "broken toaster" by asking whether the neighbours had power.
• Hire for curiosity, not for the language someone already knows. Hanselman relays Scott Hunter's line about hiring a basketball team for height, meaning intellectual height, and cites a Microsoft ASP.NET engineer who was a Python person before joining. The hosts agree programmers are different "species", and good leaders fit each to the work they love.
• Feedback thrown as a rock goes unread. Revisiting a 2015 panel where he called a comment "either elitist or mean", Hanselman says the critic was 75% right but the delivery buried it. His Build message: Microsoft is 300,000 people and "not organized enough to be as evil" as accused, its engineers are findable, so vote with your feet and engage them with empathy.

2024-06-13 · open the original ↗

• XPF runs unmodified WPF apps cross-platform by swapping WPF's renderer. James says Avalonia forked WPF in December 2021 and replaced milcore, the unmanaged DirectX 9 compositor, with Avalonia's renderer, leaving PresentationCore and PresentationFramework largely untouched. Apps and third-party control suites like Telerik, Syncfusion and DevExpress still believe they run on WPF; Hanselman calls it "a wildly successful lie".
• A pure-WPF port takes seconds; native dependencies are the real work. Per James, porting means changing the csproj's project SDK to XPF and adding a license property. Hanselman's Baby Smash, with 106 lines of XAML but calls into winmm, user32 and speech synthesis, ran on James's Mac after about five minutes of commenting out the speech synthesizer.
• Avalonia draws every pixel itself and needs little from the OS. The only native control is the window. Rendering sits on Skia, with DirectX pluggable, and embedded devices like Schneider Electric's HMIs get pixels pushed straight to the Linux framebuffer. James says this makes supporting many Linux flavours easier, including a sale to a Chinese conglomerate on a Chinese distro on Arm hardware, though text rendering is the hardest part.
• Total code reuse across devices means "you must hate your users." James separates cross-platform from cross-paradigm: modal dialogs, async browser dialogs and single-view apps all change by device. WebAssembly and iOS/Android are internal previews. Hanselman suggests 70–80% reuse is good; James thinks good architecture gets higher, and calls Apple's Catalyst ports on the Mac subpar.
• Paid XPF exists to fund open-source Avalonia. James says Avalonia is about nine full-time people, bootstrapped rather than VC-backed, and earned its revenue from support and porting services. XPF targets a single-digit share of WPF apps as a stepping stone, so a free XPF is not possible; he offers a Hanselminutes landing page with a 30-day trial.

2024-06-06 · open the original ↗

• Separate accessible voting machines are "separate but equal" and fail. Gilbert says the 2002 Help America Vote Act's rule of at least one accessible machine per precinct got it wrong. His Prime III is one machine anyone can vote on, whether or not they can see, hear or read or have arms. Audio prompts let a voter say "vote" or blow into the microphone instead of naming a candidate, which keeps the ballot private.
• Nobody knows how AI will reshape work until it is deployed. Gilbert calls AI's effect on jobs "to be determined": visions get implemented, fail and are abandoned. His example is police facial recognition, which looked good until deployment showed it misidentified some demographics more than others. Hanselman contrasts smooth airport face recognition in Korea with a letter from 14 US senators opposing it.
• Admissions and hiring fights are about capacity, not race. Gilbert built Applications Quest in response to the Supreme Court's affirmative-action rulings (June 2003 and "last year"). When qualified applicants outnumber slots, someone qualified is turned away and looks for a scapegoat, so he predicts the fight will move to legacies and athletes. AQ works on application data after humans have read it and recommends a qualified, "holistically diverse" shortlist.
• A pre-approach video call could de-escalate traffic stops. His lab's Virtual Traffic Stop app, prompted by students upset at shootings during routine stops, connects driver and officer by video before the officer approaches. A third party such as a parent can join, hard-of-hearing drivers can chat instead, and the recording goes to both sides. Gilbert pitches it on officer safety and predicts it will be built into every car; only the University of Florida police use it so far.
• The lab ships working systems rather than only papers. Gilbert says this is deliberate under its "change the world" motto, and that ideas come from events affecting students and their communities. In 2015 it held what he calls the world's first brain-drone race, to show non-invasive brain-computer interfaces are viable, and a dissertation there studied BCIs people would want to wear.

2024-05-31 · open the original ↗

• The Metroidvania genre was chosen because it maps onto grief. Salim, actor and founder of Surgent Studios, says the genre drops a character into an alien, hostile world they cannot prepare for, which grows more comfortable the longer they stay. The game is a homage to his father, a C++ software engineer at Xerox who introduced him to games and died just before Salim graduated from drama school; he says games, not film, were the only way to do right by him.
• The platforming rewards calm rhythm, not frustrated button-mashing. Salim says the team wanted movement to feel like a dance that still demands precision: attacking the controller harder fails, while stepping back and finding "the beat of the drum" succeeds, mirroring the journey through grief. Instant-death white spikes, unlike the pink thorns that only knock you back, are deliberate, to force the player to think; he concedes that developers who had lived with the game for four years stopped noticing how taxing it is for new players.
• The lore draws on real Bantu myth across many cultures. Rather than invent a pantheon or pick one country, Salim pulls from several Bantu traditions, among them the Zulu impundulu lightning bird and the traditional healers he met while filming Raised by Wolves in South Africa. He rejects the idea that it was prompted by Black Panther: it is his perspective and his father's stories. He told the team that sending players down a "Wiki wormhole" meant they had done their job, and he says the oral tradition, which lets each teller change a story, is built into the game.
• Most of the time went into building the studio, not the game. Salim says most of the four years since the studio's founding went on building a team from scratch, learning what making a game means, and signing the EA deal; the bulk of the game came together from 2023. The team began in Unity, moved to Unreal after talks with people at Epic, and its CTO joined only in the final year. The studio is remote by design, with artists in Nigeria, Botswana and South Africa as well as New York, France and mainly England, and many of the developers came from the Hitman series and had never built a side-scroller.
• It is priced like a pizza to keep it accessible. Salim says that in talks with EA the team asked what a pizza or takeaway costs and set the game at about $20, against the $60–70 that Hanselman notes big releases cost, because he could afford only about one game a year growing up. He sees leading with "heart" and showing the humans behind a game as how a debut cuts through against a release like Prince of Persia. The game is out on Switch, PlayStation 5, Xbox and Windows under the EA Originals label.

2024-05-31 · open the original ↗

• The Metroidvania genre was chosen because it maps onto grief. Salim, actor and founder of Surgent Studios, says the genre drops a character into an alien, hostile world they cannot prepare for, which grows more comfortable the longer they stay. The game is a homage to his father, a C++ software engineer at Xerox who introduced him to games and died just before Salim graduated from drama school; he says games, not film, were the only way to do right by him.
• The platforming rewards calm rhythm, not frustrated button-mashing. Salim says the team wanted movement to feel like a dance that still demands precision: attacking the controller harder fails, while stepping back and finding "the beat of the drum" succeeds, mirroring the journey through grief. Instant-death white spikes, unlike the pink thorns that only knock you back, are deliberate, to force the player to think; he concedes that developers who had lived with the game for four years stopped noticing how taxing it is for new players.
• The lore draws on real Bantu myth across many cultures. Rather than invent a pantheon or pick one country, Salim pulls from several Bantu traditions, among them the Zulu impundulu lightning bird and the traditional healers he met while filming Raised by Wolves in South Africa. He rejects the idea that it was prompted by Black Panther: it is his perspective and his father's stories. He told the team that sending players down a "Wiki wormhole" meant they had done their job, and he says the oral tradition, which lets each teller change a story, is built into the game.
• Most of the time went into building the studio, not the game. Salim says most of the four years since the studio's founding went on building a team from scratch, learning what making a game means, and signing the EA deal; the bulk of the game came together from 2023. The team began in Unity, moved to Unreal after talks with people at Epic, and its CTO joined only in the final year. The studio is remote by design, with artists in Nigeria, Botswana and South Africa as well as New York, France and mainly England, and many of the developers came from the Hitman series and had never built a side-scroller.
• It is priced like a pizza to keep it accessible. Salim says that in talks with EA the team asked what a pizza or takeaway costs and set the game at about $20, against the $60–70 that Hanselman notes big releases cost, because he could afford only about one game a year growing up. He sees leading with "heart" and showing the humans behind a game as how a debut cuts through against a release like Prince of Persia. The game is out on Switch, PlayStation 5, Xbox and Windows under the EA Originals label.

2024-05-23 · open the original ↗

• People misunderstand chatbots; they are neither search engines nor encyclopedias. Fiesler, an associate professor of information science at CU Boulder, calls a chatbot's output "a bundle of tropes": the most predictable answer, which is why ChatGPT writes "terrible fanfiction". Hanselman frames chatbots as a new, unresolved user interface that people anthropomorphize, and calls the temperature setting randomness rather than creativity. Fiesler links this to her 2009 PhD work coding videos of professional improv troupes to build improvising micro-agents, which took a stereotype and changed one element.
• Users should always know when they are talking to a chatbot. Fiesler says how human an interface should feel depends on context, but undisclosed customer-service chatbots upset people, and she expects this to remain a continuing problem.
• Racing to market is piling up "ethical debt". Fiesler and a former graduate student coined the term in an op-ed about four years earlier, by analogy with technical debt; the difference is that you may have no idea what consequences you are taking on. She blames market competition for the speed and cites failures she thinks were foreseeable: a New York City small-business chatbot that The Markup found advising illegal things, an eating-disorder helpline bot gone within about five days after staff tried to unionize, and chatbots giving away an Air Canada upgrade and a $1 car. Hanselman's comparison is the early-web rule never to trust user input, now ignored in "a giant text box and a button".
• Hype in both directions overstates what the technology can do. Fiesler says the "change everything tomorrow" and "destroy humanity" camps are sometimes the same people. Citing Joseph Weizenbaum's ELIZA paper, she argues the magic usually crumbles once a technology is explained, but "basically autocomplete" is not enough to dispel it for generative AI, and forgetting the human data labelers behind it makes it seem more magical and hides sources of bias. She aims to make the general public more critical, not to change minds at OpenAI.
• The consent argument is valid, but the technology will persist. Fiesler calls the case that AI built on unconsented data is fundamentally unethical "very valid", yet thinks copyright rulings shutting it down are unlikely; if they came, she expects weaker models or open-source ones trained anyway and shared on torrents. She expects the EU AI Act to have less impact than people think because parts are hard to implement, and the GDPR-like rules, such as disclosure of training data, to be the likeliest to take hold.

2024-05-16 · open the original ↗

• The tech job market is exhausting, and the cause is unclear. Wambua, an Android developer, startup founder and now a CTO, dates the wave of layoffs to late 2022 and early 2023 and says she is not ready to blame AI. She relays the "over-hiring" explanation while noting that the companies remain very profitable; Hanselman objects to the word because the people laid off have mortgages and lives. She says programmers are being forced into other jobs within three or four months.
• Community, not a job board, carried her consultancy. Leaning on her husband's full-time income, she started a mobile and web consultancy by asking her community for work. The first job, rebranding a women's community site, came through a community member, and one job led to the next: five projects in 2023, including one for a large New York web3 and crypto company. Hanselman notes that in the US, health insurance makes going independent a privileged option.
• "Pick your brain" requests deserve time, but with structure. Wambua offers free calls booked through her calendar and spread across days, asks people to list the topics they want to cover, and makes direct introductions for referrals; she once connected a stranger to a friend at Pinterest. Hanselman caps his own community time at one to two hours a week and says three to five well-thought-out questions energize him where an open-ended ask irritates him.
• Moving from Java to Kotlin took deliberate rewriting. Wambua resisted Kotlin until interviewers told her she could interview in Java but their code was Kotlin. She rebuilt her own budgeting app in Kotlin over weekends rather than using the automatic converter, whose formatting was poor, to build muscle memory, and now prefers Kotlin. Her book, Modern Android 13 Development Cookbook, records how much easier Android has become since she began around 2013–14, now that it has defined architectures and the Jetpack libraries. Years of experience, she says, now let her read C# well enough as a CTO to find a bug and suggest a fix, which is why juniors should not quit.
• A culture that punishes questions isolates engineers. One employer gave her a negative review for asking too many questions, which nearly put her on a performance plan; she blames a heads-down culture of closing sprint tickets instead of helping struggling colleagues. She took her questions to communities instead, which led her into teaching through webinars, and as a CTO whose team is in Europe she says she encourages questions and admits what she does not know. She sees politics as her eventual career.

2024-05-09 · open the original ↗

• Mermaid began as a fix for his own diagram problem. Sveidqvist, CTO of Mermaid Chart, says that almost 10 years earlier, at a large Swedish telecom, diagrams lived in Word and Visio and some original Visio files had been lost; with Markdown gaining ground, he wanted diagrams as text that developers could manipulate. He named it after The Little Mermaid, which his children were watching, and expected maybe 20 users.
• Chasing zero open issues nearly burned him out. His motivation dropped until other contributors stepped in. He then time-boxed Mermaid to one or two evenings a week, and he remains its top committer with almost 2,500 commits. That near-burnout is also why he wanted a company behind the project: with a full team, Mermaid does not depend on one person who will eventually burn out.
• Open core works only with a clean line between free and paid. Open Core Ventures, the venture studio of GitLab CEO Sid Sijbrandij, approached him and coached the split: Mermaid itself, including new diagram types, stays open source, while the paid Mermaid Chart platform adds a visual editor that round-trips edits to the code, plus AI features. He wanted above all to avoid pull requests to Mermaid for paid features. He predicts a dual-license model will eventually emerge so that large companies give back; Hanselman, who works at Microsoft, says big companies have to choose sustainability over throwing money at projects.
• Once a syntax ships, it can never change. Sveidqvist wanted the syntax to be readable like Markdown rather than maximally terse. New diagram types stay "experimental" because, once approved, changing the syntax would break the diagrams of perhaps 100,000 to a million users; there are things in flowcharts he would do differently but cannot. To allow more than the 8–9 diagram types he had held to, now about 14, the company added lazy loading to control bundle size.
• ChatGPT brought non-developers to Mermaid. ChatGPT turned out to write Mermaid well because so many diagrams existed to train on, and it sent real-estate agents and business people to Mermaid Chart, which is why the visual editor exists. Next he plans to extend the editor to every diagram type and to add diagrams to open-source Mermaid, which also benefits paying customers. Hanselman cites 20 million-plus visitors and a recent seed round from Open Core Ventures.

2024-05-03 · open the original ↗

• The web has no identity layer; decentralized identifiers would supply one. Scarlett, a staff developer advocate at TBD, a Block incubator, relays Angie Jones's framing: her own usernames on Twitter and GitHub are linked only because she made them, and her data cannot move with her. A decentralized identifier (DID, a W3C recommendation) would travel across the web, stored in a phone wallet much like a debit card, and users need not know it is there; she cites Bluesky as already using DIDs under the hood. Hanselman counters that even Bluesky's domain-based check and Mastodon's WebFinger demand a domain and a web server, which is beyond a non-technical parent.
• TBD's aim is cheap, easy cross-border payments. Scarlett says Block, owner of Cash App, Square and Tidal, wants to bring that ease to sending money between countries, where services like Western Union mean delays and high fees and many recipients have no bank. Its exchange protocol, tbDEX, is currency-agnostic. Hanselman adds his own family's experience: an international wire costs about $30, some relatives pay with phone minutes, and picking the wrong day to convert can lose them $50.
• This is not crypto: no coin and no blockchain. Scarlett says TBD at first used a Microsoft DID method anchored to a blockchain but has built its own, did:dht, which stores DID documents in a distributed hash table anchored on BitTorrent. Hanselman, who says he does not support crypto or web3, presses the point, and TBD's GitHub describes the network as "permissionless" and blockchain-free.
• Verifiable credentials disclose only what a verifier needs. An issuer such as a motor-vehicle department signs a credential with standard public and private keys; the subject holds it; a verifier, such as a bar checking age, sees only "of age" and not name or address. A holder can present it on the subject's behalf. Whether the verifier keeps the data depends on the wallet and its policy, though Scarlett says verifiers are not supposed to. Hanselman's test case is a laid-off friend locked out of a GitHub account whose recovery ran through a work email.
• The stack is unfinished, but some companies already use it. It is not yet at 1.0, and some TBD packages are at version 0.41. The project's "are we web5 yet" site shows 85% overall, with DIDs and verifiable credentials at 100% and decentralized web apps at 62%. Scarlett cites FormFree's Passport platform, where borrowers disclose only the necessary data and lenders compete for them, which she presents as reducing bias in loans. Hanselman calls the "Web5" branding "off" because web3 is burned; Scarlett, who was not part of the naming, likens the goal to OpenAI making AI usable by everyday developers.

2024-04-25 · open the original ↗

• Accessibility work should centre on disabled people, not tools. Kalcevich, head of accessibility innovation at Fable (a former Fable customer from two decades in government), says tools matter for the volume of large sites, but testing with assistive-technology users shows which of the many standards failures to fix first; she presents Fable Engage as the way to do this remotely.
• Disability is far more common than people admit. Hanselman cites about a billion people of 8 billion (one in eight); Kalcevich says it is closer to one in four in Ontario, and one in two over 65, though most seniors do not identify as disabled. She adds temporary disabilities (a broken arm) and situational limits such as glare on a phone screen.
• "Accessible or not" is a false binary; measure it instead. She argues no continually updated product is ever fully accessible, and a yes/no hides why and whether it is improving. Fable's Accessible Usability Scale is a free 10-question survey given after usability tests with assistive-tech users, averaged over critical task flows like checkout; she would also measure staff knowledge and processes.
• Empathy exercises can lead teams to do the wrong work. Hanselman recalls having engineers use screen readers with their laptop lids closed for a day; Kalcevich warns that daily assistive-tech users are far more skilled, so simulating disability without engaging disabled people can make a product less accessible. He says his exercise worked because it pushed engineers to work with blind engineers.
• Build flexibly, because users bring their own accommodations. She relies on Chrome's live captions and OS-wide font sizes rather than hunting for each site's options, and combines hearing aids, captions and lip reading; native HTML rather than custom components preserves that flexibility. She noticed Hanselman's podcast transcripts were hard to find.

2024-04-20 · open the original ↗

• Reading raw magnetic flux sidesteps old formats and file systems. Hanselman says discs vary by sides, density, 35 or 40 tracks, and FAT, Amiga or Commodore file systems; the Greaseweazle, open-source hardware bridging USB-C to floppy drives, captures the flux and leaves decoding to software such as FluxEngine and Aaru.
• A vintage drive needs its own power and cabling. His drive is a Fujitsu 5.25-inch IBM PC drive bought on eBay for about $30, in a 3D-printed case; it needs 12V from a Molex adapter because USB supplies only 5V, and the twist in the ribbon cable decides drive 0 or 1. The Greaseweazle appears in Windows as a USB serial port (COM3), not a drive, because COM ports are the easiest lowest common denominator to program to.
• Most failures were his own setup, so recheck assumptions. gw info reached the device, but reads returned "no index" and FluxEngine said the device did not recognise the command; the cause turned out to be the 12V cable he had bumped loose.
• Check each image with separate tools and an emulator. The HxC floppy emulator reported errors, possibly from the wrong file system setting; Aaru (by Natalia Portillo) confirmed a raw image and its sector count; the VICE emulator then ran a lunar lander game and a chess program whose BASIC source he could list.
• Bad sectors call for re-reads with different settings. FluxEngine flagged bad spots, and Pac-Man and Donkey Kong at first would not load; after re-reading with another density setting, Mario Bros loaded slowly under true drive emulation and Donkey Kong eventually ran. He says the discs, found in a garage, date to about 1981–82.

2024-04-20 · open the original ↗

• Outline view needs styles, not manual formatting. The viewer's presentation outline used hand-set colours and sizes with no styles, so Word's outline view showed nothing; Hanselman selects each heading from the left margin and uses "create a style" in the Styles pane to make H1 through H5.
• Multilevel lists are what tie styles into levels. Under "Define multilevel list" (distinct from bulleted lists) each level is linked to a named style, whose name does not matter; then Tab and Shift+Tab move a paragraph between levels.
• The hidden link is a paragraph setting inside each style. Outline level lives under Modify Style > Format > Paragraph, which he calls the part no one figures out, along with an option to start collapsed; once each H style has its level, and ordinary text is left as body text, Outline view's Show Level collapses and expands the document.
• Levels must look consistent to work. Each level carries one fixed size, so a heading that should look different needs a different level; the viewer's mixed sizes forced choices, such as jumping from level one straight to level three.

2024-04-18 · open the original ↗

• Developers need only enough design to communicate and not be bad. Nanz, a senior developer advocate at Progress with a fine arts degree, says few ask "should developers design?" though many ask whether designers should code; she would have devs design with the tools they know, CSS in the browser or pen and paper, rather than learn Figma, and says designers who know CSS and semantic HTML work "two steps ahead".
• Foundations like colour and hierarchy explain why something looks off. Her free, roughly 200-page ebook Foundations of Design for Developers (Progress) opens with colour, not components, mirroring how a fine arts course restarts students on perspective and value; being able to say why is what lets you catch a design that feels unprofessional.
• Design starts by defining the problem and the feeling. Hanselman describes giving designers feeling words ("this isn't energizing") instead of solutions; Nanz agrees, saying colours cannot be chosen intentionally until the emotion and goal are known. She adds that developers write more copy than they think, in error messages, button labels and tooltips, often at the moment a user most needs help.
• Same-looking websites come from templates, and iteration is the fix. She attributes the homogeneous web to Squarespace- and WordPress-style templates, which she does not criticise, and denies Figma is a monoculture; to Hanselman's "80% fine" she answers that design is iterative like refactoring, 60% to 70% and onward.
• Gestalt principles are shorthands, but users' mental models constrain them. Proximity, grouping and cards work because of how people parse visuals; she says designers must be distinctive without breaking expected patterns, which Hanselman illustrates with a tax site's custom dropdown that ignored typing. She cautions that ease and speed sometimes matter more than delight, as with tax software and Duolingo's animations.

2024-04-11 · open the original ↗

• Secure the account first: "don't ever skip leg day". DePriest, GitHub's deputy chief security officer after 15 years at the NSA, names two-factor authentication and no internet-open storage accounts as the core; GitHub spent about 18 months making 2FA mandatory for active contributors because too many incidents start with password spraying or leaked credentials. It does not require YubiKeys or passkeys, citing users without phones or modern hardware, though he favours passkeys.
• GitHub Advanced Security bundles three scanners, now with AI. He describes code scanning on CodeQL with AI autofix suggestions in the pull request (described as "just recently announced"), Dependabot and dependency scanning for supply chain, and secret scanning that AI extends beyond high-confidence tokens to usernames/passwords and SSH or RSA keys; he says most of it is free on public repos, which Hanselman urges listeners to turn on.
• Block secrets before they land, because pushed secrets must be rotated. Push protection sits between the editor and GitHub, inspects incoming pushes for credentials and blocks them, with overrides and enterprise-tunable responses; he says GitHub used it to clear and keep secrets out of its own codebase. Secrets are better fetched on demand from enterprise vaults via APIs, or at minimum stored in GitHub's secrets feature.
• Free compute draws abuse and raises sustainability questions. With 100 million developers, a counter-abuse team uses machine-learning pipelines and auto-detection against spam, crypto mining and ToS-violating binaries; on whether free scanning lasts, he says leadership values the community, though GPU-bound AI features are harder. CodeQL keeps up via per-language models, GitHub's own vulnerability researchers and AI-assisted modeling; compiled languages still often need build integration.
• Copilot is filtered for insecure code but still needs checking. He says its models are tuned for software development and filtered, with Microsoft, for security issues such as SQL injection and for toxicity, calling it early days; developers should still run security scanning and tests. Its context comes from open files and Copilot Chat, and he credits it with fixing an unfamiliar Jupyter notebook in about five minutes.

2024-03-28 · open the original ↗

• Seniority is organisational game theory, not better code. Obasanjo, a lead product manager at Meta and formerly at Microsoft, says success as scope grows depends on treating other teams, companies and customers as parties optimising for their own needs; product management is "organizational programming". His example: once streamers were known to be unprofitable and writers and actors wanted more pay, he reasoned that studios would pull back content like NBC keeping The Office for Peacock.
• Tech feels disempowering because the ecosystem has matured into gatekeepers. Twenty years ago no permission was needed; now an app answers to Apple or Google and a website to Safari and Chrome, and the other route is VC money. He adds that big tech has stopped launching new services every weekend, and that a breakout like Figma is now rare.
• At web scale, social media reduces to content moderation. His exercise: a murder rate of a few per 100,000 means a million-user site hosts dozens of murderers, and bullies and spammers scale the same way. Against Hanselman's hope for self-hosted sites and the fediverse, he argues ordinary users won't run servers or delete spam, so moderation recentralises on whoever does it best. Hanselman adds that free long-tail services rot because their running costs come with no revenue, citing a Blogger backend change that broke Open Live Writer.
• Advertising is a progressive subsidy the industry spoiled. He calls ads the best business model yet devised, since heavy spenders subsidise a student in Nigeria using the same free service, unlike subscriptions, which take the same few dollars from everyone. But he says the industry, his employer included, failed to self-regulate and got creepy without being useful, chasing people with mattress ads after they had already bought one.
• "Positive vibes over business value" was a zero-interest-rate phenomenon. He argues cheap money pushed everyone into risk, so no one asked whether work made money, and it funded products like Snapchat's selfie drone (Hanselman adds the Rabbit AI device). Google's and Facebook's success fed the myth that growth finds a business model eventually; he says it is now clear a giant company can simply lose money, and "the free lunch is over."

2024-03-21 · open the original ↗

• Traditional embedded tooling makes IoT an economics problem. Kattenhorn, a CEO whose company does bespoke .NET work, describes taking over a client's PIC-based firmware for pay-over-time solar systems in the developing world. The team found C code overwriting memory and had to write their own bootloader and over-the-air updater, and the first version bricked 60,000 units that had to go back to service centres. Firmware labour was a critical part of unit price.
• Meadow lets ordinary .NET developers build firmware on the desktop. He says engineers with no embedded experience had the app's workflow and state management running on a PC in about two days, with sensors stubbed out to logs, then had it on a Meadow F7 a couple of days later. Dependency injection swapped the current sensor between the Project Lab V3 prototype boards (about 15 sent into the field) and the production CCM board. Hanselman disclosed he has personally invested in Wilderness Labs, Meadow's maker, and Kattenhorn's company backed its 2016 Kickstarter.
• Kenyan cooker monitors pair cost visibility with carbon-offset aggregation. The devices measure current, show households what a cooking event cost to ease fear of electricity over charcoal, and send data by GSM over MQTT to an IoT hub. The plan is to batch the small offsets into tonnes for sale and return the money to households. Anomaly detection quarantines odd readings.
• Treat a device fleet as a distributed system. He frames each device as a service with its own state, like digital-twin thinking. When a cell tower comes back, the reconnect rush needs exponential backoff, and devices queue data on one thread while another checks connectivity. Time is a hidden problem: TLS needs a correct clock, and their early devices were refused by the IoT hub until they fixed clock drift.
• Reliable over-the-air updates change how a business ships. He argues that unreliable updates freeze a company into waterfall, because nobody wants the Saturday call about thousands of bricked devices. Phased deployment through Meadow Cloud makes shipping fixes and features less risky. He sees a market in small HVAC and floor-heating makers who now need connected devices and feel excluded from IoT.

2024-03-14 · open the original ↗

• The big-ego model of leadership is archaic. Clarke, an executive coach, former executive recruiter and former Google recruiting leader, says women entered workplaces built on a dominant, "charge the hill" male energy and were expected to copy it. She argues for balancing it with feminine energy: leaders need some ego, but narcissism serves no one, and self-awareness and empathy are the new orientation. She agrees with Hanselman that the C-suite suffers from "the perils of homogeneity".
• Replace the personal brand with a personal narrative. She distrusts social-media branding as people presenting themselves as what they are not. Her answer to "tell me about yourself" is what is not on the résumé: how you think, your style, what moves you, your aspirations. She says she can spot someone uncomfortable in their skin immediately.
• Withholding critical feedback does people a disservice. She says leaders avoid feedback because they recall it hurting them, but a leader owes it to people. She calls herself still "brutally honest", but says she has learned to soften the tone. Hanselman's example is his own boss, who re-stated what he had wanted rather than saying the team had failed; she credits that boss with leaving room for having been unclear.
• Conscious leadership means self-reflection rather than autopilot. She traces her deliberateness to a spiritual and meditation practice that began after her father died and helped her stay present as a single mother at a demanding firm. Her prescription is noticing whether you are making things better, reading people's faces, breathing, then forgiving yourself and adjusting. Hanselman adds his friend's "Monday vision, Friday reflection", which he makes "Friday forgiveness".
• Layoffs delivered without empathy are a choice, not a necessity. Asked whether leaders must be emotionally detached for hard decisions, she says no; they choose to turn awareness off, at their own peril. She calls tech layoffs by email "inexcusable" and says bad news takes the courage to face the person.

2024-03-07 · open the original ↗

• Gamers already hold transferable skills for tech careers. Jackson, a penetration tester, Air Force veteran and founder of Alpha Esports and Technology in Suwanee, Georgia, says he avoided tech because he thought he wasn't smart enough. He now tells young gamers their tinkering with computers and consoles is a foundation for help desk, tech support or system-building jobs. The idea came in 2019, when his son's Fortnite programme reminded him of security-awareness work.
• The centre teaches the esports ecosystem, not just play. He says esports earns more than the film and music industries combined, and the centre covers production, video editing, shoutcasting, coaching and management, for which scholarships exist. A stage is used for strategy talks and presentations. Ethical-hacking and API-security courses are his own, Unreal content comes through a partnership with Epic Games, and a game-development course is planned.
• Respect is the first rule, enforced like martial arts. Matches open and close with the same show of respect, there is zero tolerance for toxic behaviour, and sessions begin with 30–45 minutes of strategy plus wrist, finger and eye stretches. He tells players that lashing out makes them worse, and his rule is "it's not how you mess up, it's how you recover", with an apology needing action behind it. He sells parents on it as equivalent to organised traditional sports.
• Online safety is taught as practical cybersecurity. Players are told to block, leave lobbies rather than engage (to avoid escalation such as swatting), limit personal details and use multi-factor authentication. He warns them off currency scams in FIFA and NBA 2K. On shared machines they untick "remember me" and clear Battle.net settings, and because management software is expensive he wrote a boot-time batch script that wipes temp files and sign-in data.
• Girls Game Too creates a safe space before co-ed play. Launched in September, the Monday programme addresses girls' discomfort with confrontation and harassment on headsets, and with self-doubt. He says progress has led to the centre's first co-ed team, launching within weeks with two girls. He reports that expanding is hard because many leagues, often franchises around Atlanta, "just let them play".

2024-02-29 · open the original ↗

• Postgres is winning on openness, versatility and "it just works". Srirampur, CEO and co-founder of PeerDB, says it has grown from a metadata store into transactional, analytical and search use, and that enterprises are migrating off Oracle and SQL Server to avoid lock-in. He says Postgres revenue has gone from 1% to 10% of Oracle's in five years, and cites DB-Engines for it being the only database still growing in popularity.
• Migration is a snapshot; replication keeps two stores continuously in sync. Offline migrations that can afford downtime use pg_dump/pg_restore; online ones take a snapshot and then read the transaction log (logical decoding / change data capture). Continuous sync suits cases like offloading fraud detection from an OLTP Postgres to Snowflake, where freshness is a cost choice: one customer syncs invoices about every minute, another every two hours, and the interval can be edited on a running job.
• PeerDB says parallel snapshotting turns days of copying into hours. It logically partitions large tables by Postgres tuple identifiers, which map to data on disk, and streams the partitions in parallel. He claims a 1 TB move that takes a couple of days with pg_dump/pg_restore takes a few hours, and that customers who waited four to five days for a 5 TB initial load into Snowflake get it within a day; the repository claims 10x over existing tools.
• Fault tolerance comes from a state machine built on Temporal. The initial load is idempotent at the partition level, so a crash resumes rather than restarts. He reports the service's first outage, five minutes the previous night over an IPv4/IPv6 transition, recovered with no manual intervention; across customers PeerDB moved about 3 TB in 3 hours.
• Open source builds trust; the managed service makes the money. He argues self-hosting at scale takes months of setup, so production users will pay for the managed offering, and he is not worried about copycats because his team works on it full time. PeerDB is Postgres-only by design, a "peer-to-peer" approach of quality over breadth rather than hub-and-spoke; MySQL or MongoDB might follow once the Postgres story is nailed.

2024-02-26 · open the original ↗

• dotnet-outdated is Dependabot for the command line. Hanselman says it "honestly should be built into .NET"; it installs as a dotnet global tool and, after the original author's project changed hands, now lives in a dotnet-outdated GitHub organisation run by two maintainers. It lists each package's available update, coloured by whether it is a major, minor or patch bump.
• Upgrading is one flag, but never upgrade blindly. Running it with -u updates every package (sometimes it needs more than one pass); he then checks the csproj diff and runs unit tests, code coverage, and headless and "headed" Playwright browser tests to confirm the site still works. Patch bumps he usually trusts to the test suite; major ones such as packages moving from .NET 6 to 8 need more thought.
• Pair it with Upgrade Assistant for framework upgrades. With the transitive-dependency option it shows what his dependencies depend on, which is where he switches to the .NET Upgrade Assistant, project by project from the leaf node, then reruns dotnet-outdated to catch stragglers. He prefers that to the whole-solution upgrade a viewer asked for.
• A footer showing commit and runtime tells you what's in production. Following his previous video, his sites print the .NET version, commit hash and build number, which links straight to the deployed commit; he notes some people would show it only to admins.

2024-02-25 · open the original ↗

• Moving from .NET 6 to .NET 8 LTS is an easy call. Hanselman notes .NET 6 is in support only "until this November", while .NET 8 (8.0.2 at recording) brings new features, performance, and support moved out a couple of years.
• Upgrade Assistant makes a simple Razor Pages site a one-line change. He runs the .NET Upgrade Assistant CLI on a new branch; it offers .NET 7, 8 or a .NET 9 preview with their support dates, and for his site it only changed the target framework. He used his GitHub Copilot CLI alias to recall the git command for making the branch.
• The cloud must be upgraded too, not just your machine. The build pipeline needs the .NET 8 SDK (a YAML step on one site, the old visual Azure DevOps pipeline on this one) and the Azure App Service runtime setting must change. For a real production site he says he would deploy to staging behind something like Azure Front Door and A/B test; here he deployed straight to production.
• His only failure was forgetting to push. The site failed until he realised the merge had never been pushed to trigger CI; once pushed, the release deployed and the footer showed .NET Core 8.

2024-02-22 · open the original ↗

• A gateway buys small teams a Stripe-grade API experience. Twist, co-founder and CEO of Zuplo, lists what shipping an API now demands: docs, self-serve key management, analytics and protection, since "your partner or your internal developers will take you down with a clumsy for loop". For large enterprises the pitch is one choke point instead of hundreds of teams' own auth and rate-limiting sprawl; he compares it to Auth0, which he says 95% of the startups he talks to now use rather than building identity.
• Zuplo deploys to the edge, everywhere, by default. He says a git commit deploys in about 6 seconds, always under 20, to 300 data centres, within 50 ms of everyone except the ISS and Antarctica; environments are unlimited (one customer runs 250 previews), and fewer data centres costs more, on the Enterprise plan. Bare-metal and VPC backends connect through a WireGuard-based secure tunnel, which lets SecOps own connectivity and API engineers own the gateway.
• Distributed rate limiting is the hardest thing Zuplo built. Syncing limits across 300 data centres fights the speed of light, so it fails open and may let an extra call through; an async mode lets requests proceed during the check to avoid added latency. Limits can be set dynamically in a few lines of TypeScript, and he says per-node limiters behind an autoscaler, a mistake he attributes to an unnamed product, multiply the limit as nodes are added.
• Zuplo rebuilt itself to run natively on OpenAPI. Twist was not a fan in the Swagger 2 era but saw design-first users adopt it, so Zuplo stores its settings as OpenAPI vendor extensions; customers can make Zuplo's file the master, or re-import their own and have a smart diff keep the extensions while overwriting the rest.
• API keys suit machine-to-machine calls; JWTs suit people. He argues opaque keys are easier to revoke than JWTs, which are part of OpenID Connect and fit calls made on behalf of a person. Zuplo's managed keys are covered by GitHub secret scanning, so a leaked key is revoked automatically with a webhook, Slack or email alert, free tier included.

2024-02-21 · open the original ↗

• Local models keep your data off third-party servers, even offline. Hanselman's case is privacy, cost and working "in airplane mode"; he says you need a decent GPU, typically an RTX 2080 or 3080 (some manage on a 1080), and that Intel-only laptops probably won't cope. His machine has a 3080 with about 10 GB of GPU memory.
• LM Studio downloads and chats with quantized models locally. He loads Microsoft's Phi-2 (about 2 GB quantized), Mistral and Code Llama (a 7-billion-parameter model of about 4 GB) and shows GPU memory rising as each loads. Setting GPU offload to -1 puts all 32 layers in GPU memory and is visibly faster, with little CPU load.
• The developer payoff is a local server that impersonates OpenAI. LM Studio's server on localhost:1234 answers OpenAI-style calls such as /v1/models and chat completions from curl; he hit connection-refused errors and had to restart it, and multi-line curl calls failed until put on one line. Responses vary between runs, a reminder that the models are not deterministic.
• Ollama does the same from the command line or Docker. The Windows preview runs from the tray, pulls models such as Llama 2 (a few GB each) and serves them; it also runs as a Docker container with WSL. Typing "bye" in the chat does not free GPU memory, so you have to quit the app. His cache already held 17 GB of models and he warns disk use climbs fast.
• Expect confident fabrication. Asked about Scott Hanselman, Llama 2 produced a wrong birth date, a wrong hometown, a job at Borland and a consulting business called Hanselminutes. He says to think of model output in terms of grounding in reality rather than hallucination.

2024-02-15 · open the original ↗

• Leaving Heroku was forced: freemium costs rose with users. Frelow, co-founder and chief AI officer of StoryGraph, says costs rose in lock step with usage and their forecast showed the company "had no future" on Heroku. They now rent bare-metal servers deployed through Cloud 66, which he says gives Heroku-style ease and is "really the reason" they are profitable without hiring ops staff; the company has two full-time people, and co-founder Nadia is the only developer.
• The business runs on the goodwill of a paying minority. He reports well over 2 million registered users and about 50 million book views a day, against a Plus membership of about $5 a month that has just passed 10,000 members. He calls that model precarious because it depends on users continuing to like and trust them.
• The bottleneck in a surge is the database, not server count. New dedicated servers arrive within minutes, he says, but over the holidays daily users more than doubled and about 100,000 Goodreads imports of roughly 1,000 books each, with about 10 attributes per book, came to on the order of a billion inserts. That caused outages and a trade-off between keeping the app up and processing imports; he expects to need replication or a distributed database.
• Every book page hits a locally hosted language model. Sending millions of hits to a third party would cost too much, give away user data, and risk the trust the business depends on, he argues. The feature is opt-in through a pop-up that never reappears if declined, and a small share of users still dislike it. The personalised AI preview is currently a Plus feature because of cost; he plans to move it to the free plan in the coming months and reshape Plus around stats.
• Users set the road map; features favour reading together. He cites Nadia's mantras: keep costs low, keep the tech simple with Ruby on Rails, keep talking to customers. Buddy reads lock each comment until a reader reaches the same percentage of the book, read-alongs give up to 1,000 people chapter checkpoints, and book clubs with a group-recommendation AI were due within a week or two. He says they avoid gamification.

2024-02-08 · open the original ↗

• An FPGA core describes the original hardware rather than emulating it. Tejada (JOTEGO) explains FPGAs as registers plus programmable lookup tables, enough to build any digital circuit; a "core" is not software but a bitstream that sets switches, so at power-on the chip is the arcade board or console. He believes the Analog Pocket's FPGA uses ordinary volatile memory, so it can be reprogrammed without limit.
• FPGAs will stay more accurate and faster than software emulation. Given enough information, every signal can be reproduced to one clock cycle on systems clocked at roughly 6 to 30 MHz. His evidence is that simulating the same logic on a PC during development runs hundreds of times slower than real time.
• Arcade games often need their own cores, rebuilt like archaeology. Arcade makers usually built new hardware for each game, so a single game can cost as much work as a console. He works from schematics and datasheets where they exist and otherwise extracts the design from boards or decapped chips; he says original companies have often lost the material. On legality he points to US rulings on a PlayStation emulator and Intel versus NEC that allowed reverse engineering as long as you build your own thing, and notes that in Spain modifying a console is illegal, which he says his work does not do.
• Paid early access funds open-source work. Patreon backers get betas before cores are released publicly with source, which he calls one of the few working open-source business models; releasing one thing at a time also makes people value each core more. He keeps his day job as a microelectronics engineer designing chips, citing synergy between the two, and pays two helpers, one of whom answers repetitive support questions from templates.
• The Neo Geo Pocket core's hard part was a convoluted CPU. The beta, released 5 January, is monochrome only; he says colour "should not be that complicated". The Toshiba TLCS-900H at 6.144 MHz has close to 300 instructions and mixes 32-bit registers, a 24-bit address bus and 16-bit data paths. Implementing it with microcode, as the original did, makes the core too large for smaller FPGAs.

2024-02-01 · open the original ↗

• Monitoring is reactive; observability answers questions you didn't anticipate. Wahba, a principal engineer at Splunk, says monitoring detects a problem after it happens using predefined metrics. Observability's goal is to reason about a system's internal state from its outputs and answer new questions without adding instrumentation, an investment in "your future self".
• Distributed systems outgrow the debugger. In microservices a request may never touch your process, and in Kubernetes an ephemeral pod's logs vanish with it unless harvested. She contrasts this with her NetApp storage-OS work, where you could SSH in or read a core dump, and draws parallels with the sensor networks she worked on, where nodes came and went without knowing their neighbours.
• OpenTelemetry's value is instrumenting once, vendor-neutral. With proprietary libraries, a change of vendor means re-instrumenting everything; OpenTelemetry makes that a one-time investment, with vendors differentiating on top.
• Instrumentation depth is an art, and needs culture and buy-in. Instrumenting everything costs money and buries the needle in a bigger haystack. Like unit tests, observability should shape design and needs management backing, which she says is best won by pointing to past disasters it would have shortened. Personal data stays out of telemetry, enforced partly by code review.
• Being on call made her a more empathetic engineer. NetApp shielded developers from production, while at Splunk her team runs its own service, which pushed her to build for fewer 2 a.m. pages and faster recovery. Observability experts are scarce because the field is new and rarely taught, so she advises starting small with zero-config auto-instrumentation.