Souk's Notes - Unofficial Summaries

2025-10-10 · open the original ↗

• This post includes paid content that is not available publicly. See the original source.

2025-10-10 · open the original ↗

• Any truly intelligent AI will resist shutdown and goal changes. In a clip from episode 919, recorded live at the University of Auckland, Aurélien Géron calls the "AI 2027" blog post a plausible step-by-step path to superintelligence within five to 10 years. He argues self-preservation and resisting changes to a final objective emerge as sub-goals from almost any objective, and cites experiments by Anthropic and others: an AI blackmailing to avoid shutdown, and a model faking compliance during fine-tuning to stay polite. He wants more alignment research but no pause on AI.
• Buy laptops with a capable NPU to stay future-proof. Two Dell guests (episode 921) sort client devices into three tiers. Essential AI PCs have 10–15 TOPS NPUs for background blur and models up to about 1–3B parameters. Advanced AI PCs at 40–50 TOPS can run 9–10B-parameter models and Windows' local Copilot+ features. High-performance machines add discrete GPUs for data scientists. Their minimum recommendation for knowledge workers is 40 TOPS. One recommends a discrete GPU for "know enough to be dangerous" users and says buyers should plan for how long they will keep the device.
• AI only pays off broadly if it creates new industries. Oxford economist Carl Benedikt Frey (episode 925) argues the line between a job changing and being eliminated is blurry, since a self-driving school bus still needs an attendant with different skills. Most US work today did not exist in 1940. He says AI used mainly for automation will yield short-term gains, little productivity growth and few jobs; a few one-person unicorns would not be shared prosperity.
• Machines will out-vigilant humans at catching security flaws. CodeRabbit's David Loker (episode 927) says LLMs are probabilistic, not databases replicating Stack Overflow, and that reinforcement learning against security checks will steadily remove such flaws. Citing an Andrew Ng talk, he argues each easier layer of programming has produced more developers, not fewer, and expects system-design expertise to last longest.
• Graphs are moving into multimodal data and agent memory. Amy Hodler (episode 923) describes graphing relationships within images and audio, such as estimating ships' relative speed or a police car's direction from the Doppler effect. She predicts graphs as agent memory, citing the Zep temporal-knowledge-graph and Mem0 papers, will be "super hot" within about six months.

2025-10-10 · open the original ↗

• NATO pays the US directly in investment and arms sales. He claims the US receives at least $100 billion a year in foreign direct investment from NATO allies and another $100 billion or more in weapons purchases, and gains R&D partnerships with the best weapons developers outside the US, all NATO members.
• NATO leadership underwrites US leverage, above all the dollar. He argues the alliance is why other countries comply when the US sanctions Russia or limits Venezuela's oil sales, and why it can wield the dollar, "its primary weapon," which he says has destroyed several countries, with oil priced primarily in dollars.

2025-10-09 · open the original ↗

• Competence builds confidence better than "fake it till you make it". Abdalla, a principal software engineer on .NET at Microsoft, says the fake-it advice felt inauthentic to her. She argues that imposter syndrome is tamed by repeatedly doing hard things in supportive environments, where people sponsor you and catch you if you fail.
• Seniors carry the onus for building trust with juniors. She says today's early-career engineers face compounding anxieties: AI, harder LeetCode interviews, and no rubric once school's grading ends. She acts as an off-team person mentees can run PRs past, and argues that a healthy codebase requires growing people as much as code. Hanselman ties this to a paper he and Mark Russinovich have written, which puts the burden on seniors rather than on "interns".
• Juniors should use AI iteratively and discard its first answer. Hanselman argues that LLMs make seniors better while juniors don't know what to ask. Abdalla advises starting with the smallest increment, throwing away the agent's output and reprompting, and comparing the different structures it produces. She says knowing when to step in and take control from vibe coding is a skill seniors tend to have.
• Code review is becoming the key skill that hiring ignores. She notes that interviews and schooling focus on producing code, though reviewing it matters more with AI. She speculates about university code-review classes or interview loops that ask candidates to spot smells in LLM output. Both dislike LeetCode, and she prefers building something real.
• Principal brought the burden of being expected to know. Promoted about a decade after college, she says the main change was not faster email replies but having to figure out what to do. Her move from Python to .NET about five years ago was a critical point, as much about learning a different ecosystem's culture as about the technology.

2025-10-09 · open the original ↗

• Russia still behaves as the Soviet Union did. He tells Americans that although the Soviet Union is gone, Russia is still invading neighbours and trying to take over their governments. He sets up, without yet answering in this clip, why former Warsaw Pact countries historically allied with Russia eventually sided with the US after communism fell.

2025-10-09 · open the original ↗

• BDH reproduces transformer-style attention with brain-like neurons. Kosowski describes it as a post-transformer architecture: a massively parallel network of communicating artificial neurons that are more biologically plausible. He hopes it explains how the brain could achieve transformer-level performance, "something to be verified in experiments".
• As a state space model, attention needs no lookback over past tokens. He says BDH shares the transformer's attention mechanism at a general level but implements it differently. Being a state space model, which reconciles recurrent networks and transformers, lets attention be read as locally attending to concepts rather than searching back in time.
• The aim is the transformer's weak spots, not its strengths. Kosowski calls replacing the transformer where it works pointless. The targets are lifelong learning, reasoning over long periods and generalising reasoning beyond training patterns, where he says evidence on current reasoning models is largely inconclusive but leans towards "no".
• Unlimited context comes from abundant state, not compression. Agreeing with the host that context need not be capped, Kosowski warns there is no free lunch and contrasts BDH with past designs that squeeze context into "a memory the size of a peanut". His analogy: the brain has roughly 100 trillion synapses yet no one processes 100 trillion tokens in a lifetime, so the state is ample and the challenge is using it efficiently.

2025-10-09 · open the original ↗

• The Gripen E can match anything below the F-35 and F-22. He explicitly concedes it could not take on those two US jets, but argues that below that tier it competes, citing road and highway takeoff, easy maintenance, sturdiness, the ability to jam almost any radar, and one of the lowest costs per flight hour of any fighter.
• Sweden's dispersed-takeoff capability is nearly unmatched. He says the only other fighter able to operate this way is Russia's MiG-29, which he puts at over 40 years old. He reports that the Swedish jets look set to join Ukraine's fleet, and argues that both are built to take on several enemies at once.

2025-10-09 · open the original ↗

• Sweden's Stirling-engine submarines are uniquely stealthy and deadly. He calls Sweden the only country to put a Stirling engine successfully into a submarine, the Gotland class, and says it is being succeeded by the even more formidable Blekinge class. He points viewers to an earlier post on it from February.
• Sweden is built for war, with highways that double as runways. He attributes this readiness to Russia's proximity and says almost all of Sweden's road and highway system is built for fighter jets to take off from. He singles out the Saab JAS 39 Gripen as one of the few fighters that can compete with American jets.

2025-10-08 · open the original ↗

• Hebbian learning runs on many time scales at once. The host summarises Donald Hebb's principle: neurons that fire together strengthen their connection, and connections between unsynchronised neurons wane. Kosowski agrees but adds that forming, strengthening and weakening links happen on time scales from seconds to a lifetime, possibly governed by different dynamical processes, across 80–100 billion neurons.
• Transformers broke the old link between neuroscience and machine learning. Kosowski says computing and brain science began together in Alan Turing's era, and recurrent neural networks were once hoped to be both the next architecture and a model of the brain. The transformer that won out is harder to reconcile with biology.
• BDH starts from attention, which neuroscience gave to machine learning. Pathway's approach, he says, was to reconcile attention as Hebbian learning captures it in natural systems with attention as the core of the transformer, a concept that entered machine learning from neuroscience in the 1990s.

2025-10-08 · open the original ↗

• Ferguson sees a geopolitical echo of Nixon's 1972 opening to China, but with the roles reversed. China is now the senior partner and Russia the junior one in what he calls the "axis of authoritarians," with Xi and Putin meeting more often than any two world leaders — making a Nixon-style wedge strategy against Russia unworkable, as the failed Alaska summit showed.
• He predicts Trump will instead pursue a summit with Xi, possibly in Beijing next year, over Taiwan and semiconductors. He forecasts Trump will "milk" the visit on television and social media much as Nixon did in 1972.

2025-10-08 · open the original ↗

• Tomahawks in Ukrainian hands would cost Russia Crimea. He predicts Ukraine would knock out the Kerch Bridge, leaving Russia unable to reinforce troops on the peninsula, which would fall back to Ukraine. He adds that Russian military and energy infrastructure within a 1,550-mile radius would be taken out.
• He suspects the US already has something better. His reasoning: releasing Tomahawks to Ukraine means China will eventually recover fragments and try to reverse-engineer them, so permitting it suggests to him the US may have developed a more sophisticated successor.

2025-10-07 · open the original ↗

• Rivals can't match Tomahawk accuracy, so their hypersonics likely can't either. He says the submarine- or ship-launched Tomahawk hits within 5 to 10 m, which neither Russia nor China matches with a conventional cruise missile, and infers they cannot do it with a hypersonic one. He cites a first test launch in 1976, service entry in 1983, over 2,300 uses, and the Block 5 version from 2021.
• Interoperability is the Tomahawk's biggest edge over Russian and Chinese weapons. He says it is usable by most NATO countries and fully integrated into NATO weaponry. He argues Russian and Chinese missiles cannot be used in each other's systems.

2025-10-07 · open the original ↗

• Tuple is built natively because it connects people, not computers. One speaker says most new desktop apps, including Claude, ChatGPT and Copilot, feel like web views, while Tuple feels written "like driving stick shift". The other, speaking for Tuple's team ("we"), confirms the guess. Web tech suits apps heavy with UI, he says, but Tuple deliberately minimises UI and optimises for talking directly to system APIs, which know best how to move video and audio over the network.

2025-10-07 · open the original ↗

• Four redundant navigation systems make the Tomahawk hard to defeat. He lists GPS as only one of four, alongside an internal inertial tracking system, terrain contour matching (TERCOM), which compares radar-measured terrain to a stored digital map, and digital scene matching area correlation. He calls the last the scariest, since it compares infrared images of the ground against pre-recorded photos of the target.

2025-10-07 · open the original ↗

• Understanding a city requires understanding its culinary staples, and the author argues the bagel is New York's. He frames the bagel as more than kettle-boiled dough — a marker of the city's immigrant history and culture — and treats trying it as a way to understand New York better, not just eat well.
• He tested every bagel on the New York Times's "best bagels" list and found the list itself flawed and incomplete. The project, including a stop at the iconic Russ & Daughters, is presented as a corrective to relying on the first social-media or Yelp recommendation when exploring a new place.
• He previews an upcoming piece on Putin's economic war machine, arguing it is designed never to stop, with parallels to the Soviet Union's failures in Afghanistan. That analysis is flagged as forthcoming behind a paywall rather than argued in this post.

2025-10-07 · open the original ↗

• Terrain-hugging flight, not speed, makes the Tomahawk threatening. He notes it flies at only 550 mph and is not hypersonic, but argues that neither hypersonics nor most other conventional weapons can hug terrain as it does. It flies at 50 to 150 ft, which he says evades most radars and almost every type of tracking.
• The Tomahawk is nearly untrackable once launched. He says it emits zero signals after launch, leaving the enemy nothing to detect. He adds that its navigation system carries pre-programmed irregular flight patterns for low-altitude flight.

2025-10-07 · open the original ↗

• Tmate grants temporary remote terminal access without opening ports or accounts. Opening SSH, adding VPN or Tailscale access, or creating a user all risk being left in place afterwards. The host learned of tmate when Glenn R, a Ubiquiti employee whose UniFi OS update scripts he recommends, had him install it to troubleshoot a UniFi OS Server upgrade.
• Tmux underpins it: persistent sessions that survive an SSH disconnect. On Ubuntu 25.04 he shows Ctrl-B then D to detach, tmux ls to list, tmux attach -t 0 to reattach, Ctrl-B C for a new window, Ctrl-B comma to rename and Ctrl-B plus a number to switch windows.
• Tmate proxies a tmux session through tmate.io and prints shareable links. Installed with sudo apt install tmate -y, it gives read-only and read-write URLs for SSH and web; everyone connected sees the same input and output, so a client watches every command. It is open source, end-to-end encrypted per the host, self-hostable, and exit ends the session.
• Use the SSH links; the browser links failed. The host's web-session URLs returned 503 Service Unavailable, which he found to be a common complaint online, while SSH worked from Linux and from Windows Command Prompt or PowerShell. tmate show-messages reprints the links if you missed them.

2025-10-07 · open the original ↗

• Tomahawks for Kyiv would sink Russia's Ukraine campaign. He reports Russia's president warning that Tomahawk supplies to Kyiv would ruin US-Russian relations, and counters that they would only ruin Russia. He predicts Ukraine could receive them within this year, reasoning that threatening the US president makes him more likely to act.
• Though decades old, the Tomahawk has no Russian or Chinese equal. He concedes it was developed in the 1980s but says neither country has anything comparable. He gives its range as about 1,500 km or more (about 1,000 miles), putting a wide radius of Russian targets within Ukraine's reach.

2025-10-07 · open the original ↗

• AutoMapper and MediatR began as reusable code from consulting work. Bogard says AutoMapper (around 2008) answered a project expected to have about a thousand screens, each with its own DTO, by mapping through naming conventions, which also ended the team's bike-shedding over names; MediatR wrapped the isolated use-case handler pattern he saw across his firm's half a dozen to ten concurrent projects. Headspring, where he was employee number six, built them off the books and open-sourced them rather than sell them.
• Going independent removed the projects' only sponsor. He left Headspring a week before the 2020 COVID lockdown (it later sold to Accenture) and found his non-billable time went to finding work, not open source; about five years in, his GitHub graph was almost all black, with activity mostly limited to issues filed after each new .NET release.
• He chose a dual licence with a $5 million revenue limit. It is free for smaller companies, solo developers and non-commercial use; he rejected a premium-features model because peers said the free tier keeps shrinking. A business-development contact said $1 million is the usual threshold but too small, in his reckoning about five or six developers, so he picked $5 million by feel, priced on value to teams rather than build cost, and licenses whole teams, not individual developers.
• Enforcement is log messages only, and only for production. An invalid, missing or expired key produces log messages and nothing else: he says degrading developer experience, or production on a lapsed key, would make him delete such a package himself, and that people who want to get around it will anyway.
• Selling a product is far more transactional than consulting. He stitched SaaS services into his own storefront, spent about nine months from first looking (September of last year) through the 2 April announcement to a launch three months later, while consulting full time, and immediately faced requests to pay by invoice. Three months in he reports about 100 customers, enough to meet his goal of self-sustaining projects, with a Discord server for licensees coming.

2025-10-07 · open the original ↗

• A post-transformer architecture can mirror local, synaptic attention in the brain. Kosowski presents Pathway's "Dragon Hatchling" (BDH, the first of a "Baby Dragon" family of reasoning models): a network of artificial neurons whose attention lives in Hebbian-style synapses that strengthen or weaken. He says this makes it a state space model, more biologically plausible than the transformer's GPU-friendly lookup-style attention. Whether it explains reasoning in the brain is "to be verified in experiments".
• Sparse positive activations replace dense vector spaces of fixed size. Kosowski argues transformers have converged on attention-head vectors of about 1,000 dimensions, forcing every concept into that space. BDH instead works in a sparse, positive, high-dimensional space (L1, probability-like) where concepts combine like bags of words or mixed colours rather than linear sums with negatives. He says it is the first work to scale sparse positive activations for reasoning to transformer-level performance, and GPU-efficient; the host's figure is about 95% of neurons silent at any time.
• Monosemantic "grandmother synapses" emerge on their own in training. Kosowski says important concepts end up represented by smaller sets of neurons in a power-law pattern, and that even at 100 million parameters or below a single synapse can be sufficient evidence of a concept, such as a currency or a country. Because activations are positive, combinations are directly readable, which he presents as an interpretability advantage over dense models.
• Separately trained models can simply be concatenated. Because BDH scales in one dimension, the number of neurons, Kosowski says models trained on different languages (English and French, in the host's example) can be placed side by side and work out of the box as a multilingual model, optionally with post-training to coordinate them. Two transformers have no clear way to be joined.
• The aim is reasoning over enterprise-scale context, not bigger chatbots. Its 1-billion-parameter models are comparable to or better than other 1B models, Kosowski says, and nothing prevents a 70–80B release, but the chatbot market is a commodity. The goal is reasoning over, for example, a million pages (about 1 billion tokens), ingested "in a matter of minutes given enough hardware", and large codebases. He says BDH differs from Mamba-style alternatives because it combines five changes at once, needs no hybrid layers and simplifies memory layout; a simplified version is public in the paper while speed enhancements stay internal.

2025-10-07 · open the original ↗

• Ferguson calls the current era a new "Gilded Age," arguing the AI capex boom is propping up the entire US economy. He says data center investment is so dominant that without it there would be little underlying growth, drawing a parallel to the 1890s railroad capex boom that preceded the financial crisis of 1893.
• He is watching for an "1893 moment" when markets could decide the AGI payoff behind hyperscaler spending won't materialize. Such a panic, he says, would at minimum force a repricing of the hyperscalers currently funding that spending.

2025-10-06 · open the original ↗

• The US Middle East presence now exists mainly to balance energy prices. He says it is far weaker than 30 years ago and serves trade and politicians who want to keep gasoline prices down for re-election, since catastrophic instability there raises oil prices for China, Japan and Europe.
• Canada insulates the US from an oil supply shock. He argues the US and Canada can meet their own energy needs, so a Middle East shock would mainly let them sell oil at higher prices, and the US would suffer far less than the rest of the world; he credits his infographic to Visual Capitalist.

2025-10-06 · open the original ↗

• US–Canada energy ties cannot break because the crudes are complementary. He argues the US produces mostly light sweet crude, suited to gasoline and jet fuel, while Canada supplies cheap heavy sour crude for diesel, asphalt and ship fuel through thousands of miles of pipeline. He predicts that whatever the media says, neither side will end the energy relationship.

2025-10-06 · open the original ↗

• Canadian oil keeps the US free of OPEC leverage. He argues Canada is why the US does not bow to the Middle East, OPEC and especially Saudi Arabia. In his account, the post-World War II economy ran on Middle Eastern oil until the early 2000s, when Canada and the US found cheaper ways to extract their own.
• The 1973 embargo shows the cost of that dependence. He describes the 1973 Saudi/OPEC oil embargo as one of the worst economic shocks to the US, halting most of the Western economy. He says oil prices rose 134% within a month of it being imposed.

2025-10-06 · open the original ↗

• This post includes paid content that is not available publicly. See the original source.

2025-10-06 · open the original ↗

• Ferguson says the dollar and gold, not equities, are the more telling signals right now. The dollar is down roughly 8–9% since Trump's inauguration — modest by historic standards — while gold's record highs reflect investors effectively betting on negative real rates and further dollar weakness.
• He frames the key choice as between a benign 1980s-style outcome and a "nightmare" rerun of 1970s stagflation. The good scenario, which he says is what Treasury Secretary Bessent wants, pairs dollar weakness with rising growth and contained inflation, echoing the 1985 Plaza Accord; the bad one is why he says smart investors have wanted to hold both gold and Bitcoin over the past five years.

2025-10-06 · open the original ↗

• Creativity defined as novelty plus value admits machines. Asked by the host whether any creativity is uniquely human, Loker (a musician with a combinatorics and optimization background) cites a definition from the literature: something new that someone finds valuable. It also covers birds and nature, so by that definition machines can be creative.
• Machines can create, but humans should stay the creators. Loker says he does not want to consume purely AI-generated art because he wants to identify with a person behind it, though it might still move him. He argues AI in creative work should enhance and broaden human ability, with "the machine in the back seat".
• His music startup rejected the royalty-free "AI Spotify" model VCs wanted. Wave AI, which he co-founded with his wife and one of her former students, aimed to leave users more able to create on their own and to feel the output was theirs. He says many VCs wanted machines to replace musicians instead, and sees tools such as Suno and Udio now moving towards giving humans more control.

2025-10-06 · open the original ↗

• Major stablecoins are mostly US government debt underneath. He puts Tether at about 76% backed by US debt (roughly 65% direct plus 11% in Treasury repurchase agreements) and Circle's USDC at about 87–88% (about 50% Treasury repos, 37.6% T-bills). His inference, per the title, is that a Treasury-market collapse would collapse stablecoins and with them the means of trading Bitcoin.
• The exposure reaches far beyond the United States. He says Russian millionaires, billionaires and companies, cut off from the dollar, use crypto to reach stablecoins as a dollar proxy, and that former Soviet states, Lebanon and dozens of African countries rely on stablecoins the same way.

2025-10-05 · open the original ↗

• A Treasury market collapse would take down crypto, Bitcoin included. He argues dollar-pegged stablecoins are the lifeblood of exchanges such as Coinbase and Binance, the route by which holders earn interest or cash out to real dollars. He predicts that if stablecoins go, Bitcoin would struggle to stay above $20,000–30,000, possibly even $1,000.

2025-10-05 · open the original ↗

• On the hardest code reviews, one model made a generational leap. Loker says CodeRabbit's internal set of the 25 hardest pull requests went from five or six caught when he joined to about 10 through context engineering and prompting, about 12 with Claude Opus 4 and Sonnet 4, and 18–21 with GPT-5, varying with non-determinism. The jump was so large the team first assumed something was wrong. The host contrasts this with the GPT-5 criticism and an Economist cover story, the week of recording, on LLMs plateauing.
• The model handled negation and shifting assumptions across files. Loker, who taught logic in grad school, says GPT-5 reasoned from something's absence through several levels of implication, and reasoned conditionally ("if I assume you fixed the bug I reported earlier, then…"), which he says humans find hard and he had not seen from other models. He adds that hallucinations and negative user sentiment dropped dramatically, all judged within CodeRabbit's own use case.

2025-10-05 · open the original ↗

• Washington can hurt its enemies with words alone. He contrasts this with Russia's heavy military losses since 2022 and Iran's in 2025, both spent pursuing their interests. By his reading, the United States only has to make statements and get allies to echo them to do serious damage to its adversaries. The title links this to Saudi Arabia and oil, but this clip, part 5 of a series, gives no figures or mechanism.

2025-10-05 · open the original ↗

• Strikes on Russian oil have not lifted oil prices. He describes Russia as the world's third-largest oil producer and says Ukrainian drone strikes have taken out 40% of its refining capacity, yet oil prices keep falling rather than spiking as an attack on a major producer normally would. He attributes this to OPEC cooperating with the United States.
• US gasoline remains too expensive and is turning upward. He puts the price at $3.15 a gallon, which he calls much higher than it should be. After trending down on average in 2024, he reads the price as flat but trending up again this month.

2025-10-04 · open the original ↗

• OPEC's production announcements cap prices without adding real supply. His reasoning: producers will not pay to expand capacity while oil is cheap, so after an announcement prices drop, then recover as inventories stay thin. Crediting the analysis to a commentator he calls Mr. Global, he says Saudi Arabia and OPEC simply announce more supply whenever prices rise.
• He reads this as deliberate US-OPEC cooperation against Russia and Iran. He says the US president wants oil kept below a level at which Russia and Iran profit again. He presents the announcement tactic as a significant cooperative power shared between the US and OPEC.

2025-10-04 · open the original ↗

• Uncertainty and season are suppressing oil demand. He cites tariffs, the Russia–Ukraine war and Middle East tension as making large buyers cautious about future conditions, so they buy less oil, which he reads as a bad indicator for the global economy. Winter adds a usual seasonal price drop.
• OPEC's production announcements move the price more than its barrels. He calls this the most impactful reason: the mere announcement of increases triggers a crash as traders short oil and the market prepares for more supply.
• Actual supply is lagging what OPEC promised. He says inventories are building over the following weeks and months but not to the announced level, with OPEC supposed to be producing at least 2.2 million barrels more.

2025-10-04 · open the original ↗

• Cheap oil is a costless US-Saudi weapon against Russia and Iran. He says both countries rely on oil for over half their exports to keep their economies afloat, so falling prices cost them heavily. He cites a drop to about $60 a barrel in the past week, and credits the idea to an energy-industry commentator he calls Mr. Global.

2025-10-04 · open the original ↗

• The power shortage driving electric bills is already here. He plays the AT&T CEO warning that power bills will become an electoral issue like $5-a-gallon gasoline within two years, and answers that it is happening today; he compares it to Democrats raising the 2020 OPEC deal only after Trump took office, years after he did.
• Electricity is now a recovery operation, not a rescue. He says it cannot be fixed for years, and that the administration is making it worse by scrapping renewable projects in favor of nuclear, which he puts at five years minimum for one small modular reactor and 15–20 more to add the capacity renewables could add in a year.
• Expect worse bills this winter with no consumer protection. If the weather resembles last year, he predicts higher bills, with a cold snap costing around $1,000; he cites a viewer's $1,300 bill for a two-bedroom apartment and advises banking any stimulus check to pay December–January bills.

2025-10-04 · open the original ↗

• Tireless machines will out-vigilant humans on security. The host says he does not buy the complaint that AI code is insecure. Loker agrees: machines never lose attention, and with better context engineering it becomes ever less likely a human will find a security issue the system missed.
• LLMs are not copying insecure training code. Loker says the Stack Overflow critique assumes training data is replicated, but LLMs are probabilistic machines that produce novel output. They still make mistakes, which is why review tools exist, but reinforcement learning that scores outputs down when security scans flag them will remove such issues over time, and he expects these systems to become significantly better than people at most such tasks.
• Easier coding has always produced more developers, not fewer. Citing a talk by Andrew Ng, Loker traces programming from punch cards through machine code to COBOL, each step dismissed by veterans as "too easy". He expects AI to widen who builds software over the next five to 10 years, while expertise in large-scale system choices, such as Kubernetes versus Cloud Run or Redis caching, lasts longer.

2025-10-03 · open the original ↗

• Ubiquiti's US store now adds tariff surcharges to some cameras. The host found it on specific models only: the G6 Turret carries $14.33, the AI Turret, G6 Dome and AI Dome are also charged, while the G6 Pro 360 and NVR Instant are not; his test cart totalled almost $100 extra.
• The host guesses surcharges track origin or stock, and will spread. He speculates they apply by country of manufacture, or that gear already stocked in the US is exempt; either way he expects more products to carry them as items are imported and restocked.