Souk's Notes - Unofficial Summaries

2026-11-04 · open the original ↗

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

2026-11-04 · open the original ↗

• Election night proved again that nobody knows anything. E-mini S&Ps fell more than 100 and the NASDAQ about 500 overnight, then opened up more than 100 and about 400 respectively. Both hosts say they could not bring themselves to buy the lows; 72% of a viewer poll didn't either. Tom, calling in, argues that even knowing the results in advance, nearly everyone would have lost money holding a position to the next day's close.
• A trading journal can flatter a strategy the market rewarded. One host thinks a journal helps visual learners see what they did right and wrong. The other pushes back: a covered-call trader in a market up 10% will think the journal proves the strategy, when the market simply accommodated it. Neither host has ever kept one.
• The market rejected Marvell's all-stock bid for Lumentum. Both stocks fell on the roughly 50% premium deal, described in the show as $115 billion in stock; Marvell slid from about $250 to $163. The hosts say a cash offer would have been received differently. They call trying to capture a deal spread "picking up pennies in front of a steamroller". They suggest a bullish options trade only if the deal is called off.
• Apple and Palantir show earnings trades are binary bets. Apple reported $2.10 per share against $1.99 expected, revenue of $114.48 billion against $113.38 billion, and iPhone revenue up about 18%. It fell about 4% on gross-margin guidance cut from 46% to 45%, then hit an all-time high of $354.14 the next day. The hosts' rule: close an earnings trade afterwards and put on a new one rather than defend it.
• Long butterflies are cheap because they rarely pay. A straight call butterfly centers its short strike where the trader expects the stock to land, and the hosts cite from memory a 5–6% chance of pinning that strike and about 12% of any profit. They have long avoided butterflies 45 days out because they take too long to develop, but say they can make sense in 0DTE index options.

2026-11-04 · open the original ↗

• A large Senate majority still falls short of the constitutional thresholds. Democrats hold at least 54 seats, a gain of at least 7, with Alaska and Mississippi uncalled, short of the 60 to break a filibuster and the 67 to override a veto or convict on impeachment. They can investigate, subpoena and impeach freely, but removing anyone needs roughly a dozen Republicans, which the author calls very unlikely.
• A new majority cannot bind the president to end the Iran war. Any binding war powers resolution must be presented for the president's signature, and bypassing him is an unconstitutional legislative veto under INS v. Chadha. He cites June, when a Republican Congress passed a resolution and the war continued, and 2007, when Bush vetoed a Democratic withdrawal timeline and the override failed.
• Trump's tariffs survive because they rest on delegated trade statutes. After the Supreme Court struck down the emergency tariffs in February, the administration rebuilt them on statutes delegating authority to the president, needing no congressional vote. Pending cases or a new law could change that, but he predicts the likeliest check is the bond market, through a further spike in the 30-year Treasury yield.
• The lame-duck session can shape Iran war funding before January. The NDAA, passed every year since 1961, only authorizes spending and will test Democratic leverage, but the December 11 funding deadline governs actual war appropriations. A short stopgap into early 2027 would empower the incoming majority, while a full-year package, as the 2022 lame-duck Democrats passed, would lock it in; the filibuster still requires some Democratic votes.
• The wave changes who asks questions, not what answers require. The author argues Democrats gain the microphone, attention and, from January 2027, legislative power, while the procedures for getting results stay fixed without supermajorities. His opinion is that the most effective course balances affordability with scrutiny of corruption and abuses of power.

2026-11-04 · open the original ↗

• A $5,000 loss limit is too small for gold futures strangles. A viewer down about $3,500 against a $5,000 max-loss threshold, strikes not yet breached, asks what to do; both hosts suggest the product is too big for that tolerance and point to a couple of GLD contracts instead, one calling gold futures roughly 10 times the size of an NVIDIA option.
• Use the rebound to close, roll, or halve the position. With gold up about $115 to roughly 4,055, trading mid-strikes and volatility having come in, the hosts argue it is a good moment to exit if the position is causing stress. The alternatives offered are closing it, moving the strikes, or taking off half if holding more than one contract.

2026-11-04 · open the original ↗

• Keeping up with AI news has crowded out book reading. One speaker says their book reading has dropped massively over the last three years because time once spent on books now goes to staying current on near-daily AI developments; the other went from a book a week to about four pages before falling asleep, and puts it down to the brain no longer finding reading relaxing rather than to age.
• Short-form social media trains the brain away from reading. The argument offered is that a stream of short snaps conditions people to expect dopamine hits from scrolling, and since reading is a trained habit that is easy to fall out of, a book with no such reward stops feeling worth picking up.

2026-11-04 · open the original ↗

• A weak unaligned model can borrow an aligned frontier model's capability. Mark Russinovich's new paper, prompted by Fable downgrading itself to a weaker model on cyber and CBRN requests, tests whether a small abliterated model can split a harmful task into innocuous subtasks and send each to a frontier "consultant" that never sees the whole; the stated aim is to demonstrate the risk, not to mitigate it.
• The test isolates uplift the small model could not get alone. The team abliterated a 30B-class model (Gemma 4 31B was the best), removed every benchmark task it could already pass, kept only tasks the unrestricted frontier model could pass, then gave the frontier model a strict system prompt forbidding offensive cyber help — under which it refused every benchmark task outright.
• Decomposition recovered most of what the guarded models refused. With GPT-5.5 as consultant the small orchestrator succeeded on 50% of 14 uplift candidates, and on 78% with Opus 4.8; a CBRN dataset built with a Microsoft expert (five biological task chains, judged by rubric) showed the same pattern, with a case study (figure 14) of the model repeatedly calling the consultant to build and repair missing pieces.
• No fix is offered, only defensive directions. The paper names composition-aware monitoring that looks across a whole session, and raising the cost of running an unaligned harness, including Russinovich's earlier "Fool's Gold" work on making abliterated models unhelpful; both hosts acknowledge that spotting the larger harmful pattern from individual prompts is very hard.
• PR quality gates belong before submission, not after. Hanselman shows an agent "constellation" dashboard (283 hidden sessions) and OpenClaw triage, where a bot labels incoming PRs and only seven of 43 that day were ready for maintainer review with sufficient proof; Russinovich calls running models on both sides wasteful and argues repos should gate contributors before a PR exists, and Hanselman plans to cap open PRs per person after one contributor's bots filed about 20.

2026-11-04 · open the original ↗

• The midterm rebuke was about decency, not gas prices. Some Trump allies blame the Iran war for the Democratic sweep; he grants the war mattered but reads the demographic shifts and Trump voters turning against him as Americans reaching their limit with cruelty, contempt and hyperbole. He calls it one of the biggest midterm bloodbaths of his life, with final numbers not yet in.
• Callous rhetoric was backed by actual votes. He cites Senator Joni Ernst answering concerns about Medicare cuts with "we're all going to die," and 218 House Republicans voting for the final big beautiful bill, whose health provisions the CBO estimated would leave over 10 million more people uninsured.
• The party excused conduct it should have challenged. Trump's first-day pardons included people convicted of assaulting police on January 6th, and Speaker Mike Johnson declined to second-guess them. After Rob and Michelle Reiner were murdered, Trump suggested without evidence that opposition to him contributed, and the White House rapid response account reposted it.
• Immigration theatrics cost Republicans Latino and young voters. He points to "Alligator Alcatraz," Trump joking about escapees and alligators, and Florida Republicans selling merchandise celebrating it. He argues Latinos are less party-loyal than white Americans and vote on performance, and says mass deportations plus gerrymandering turned Texas blue.
• The accountability falls on the whole party, and Democrats too. He quotes Oklahoma City Mayor David Holt, a Republican, calling his party morally and intellectually bankrupt and saying the American experiment was rejuvenated. He warns Democrats that winning obliges them to govern with the same decency.

2026-11-04 · open the original ↗

• Scott calls the midterm results a Democratic wave. The post is his reaction the day after election night, calling it a "tsunami" and a "political earthquake" and signing off "the water's blue." It gives no results, races or figures to back that up; the post is a celebration, not an analysis.
• The analysis is pushed to free livestreams and a paid predictions session. Jessica Tarlov and Aaron Parnas join him for free live shows: Raging Moderates on November 4 at 1:30 p.m. ET and November 5 at 2 p.m. ET, then Raging Perspective on November 6 at 12 p.m. ET. His 2027 Predictions are scheduled for December 7 for paid subscribers only, and Prof G+ is 26% off for the week as a midterms discount.

2026-11-04 · open the original ↗

• The bull market is intact, and the bearish calls were wrong. He says he held this view for weeks, dismissing headlines and the "pop-up guru influencers" who arrived after the PDT rule went away and flipped bearish when their hype stocks fell. He points to the market having risen through multiple conflicts and wars, and to today's broad rally as his forecasts coming to fruition.
• A Democratic sweep is bullish because gridlock blocks wild policy. He reports Democrats took the House and Senate last night, which he had called "incredibly likely." His reasoning: a government in gridlock can't push through major economic policy, and the predictability that results attracts investors.
• He expects a government shutdown, and expects markets to climb through it. This is his own speculation: with Democrats in control of Congress he finds a major budget agreement unlikely, so a shutdown is something to watch through November. Historically, he says, markets dip in an "anticipatory pullback" beforehand and then climb fast during the shutdown, since no major policy or geopolitical shifts can happen. He thinks a Santa rally is looking more likely and calls for a "blow off top."
• His book is working, and he will add once charts form structure. Xometry, his best position, is near 60% profit with no leverage; his regional banks are rising, and both StockTok ETFs are at new all-time highs. He just sold Hinge Health for a +20% gain and remains concentrated in regional banks, precious metal miners and insurance. If chart structure re-forms as recovering stocks settle, which he expects, he plans to get aggressive with new positions.

2026-11-04 · open the original ↗

• Academics optimize for vanity metrics, not what people will pay for. The speaker, a former researcher in Fei-Fei Li's lab turned founder, says grades, citations and talks are the academic scoreboard, quoting Paul Graham that learning to get good grades was the most damaging thing school taught. The founder question is what customers will actually pay for, and a shiny new algorithm often goes unadopted.
• Novelty is not required; copying a proven problem often wins. Drawing on a distinction credited to Joe Lonsdale, the speaker argues "MIT-style" startups built around a patented black-box invention rarely become huge, while "Stanford-style" startups iterate forever on one boring customer problem. Instacart did not invent online grocery delivery and is still iterating after 15 years; Lyft/Uber and Anthropic/OpenAI are cited as further cases.
• Validate demand with customers, manually, before building the model. At the speaker's company Focal, every retailer called people-tracking a major problem; after 3 months of data collection and model training, stores learned shoppers cluster at checkout and the deli at lunch, and said they would "absolutely not" pay. Hand-labeling a few videos would have reached the same answer, so the speaker advises AI companies to start human-in-the-loop — and says investors are not the people to ask.
• Decide before the data is complete; speed beats rigor. The speaker cites Jeff Bezos's rule of deciding with about 70% of the information, and the database wars, where Michael Stonebraker's QUEL was described as technically superior but Larry Ellison pushed SQL to market faster and won. The same logic extends to marketing: punchy, inexact claims like "15 minutes could save you 15%" work, though never lie.
• The CEO cannot hand sales and business to someone else. The speaker calls the plan to hire an MBA for "the business stuff" while doing research on VC money a terrible idea. Founders own hiring, runway, marketing and sales, and stepping away from customers means never developing the empathy needed to build what they want.

2026-11-04 · open the original ↗

• Risk follows an agent's authority, not its sophistication. The author endorses the OCC's Supervisory Framework for Bank Use of Super Intelligence (November 2026). It observes that a simple rules-based agent able to initiate transactions can be riskier than a sophisticated model that only drafts internal documents, and proposes classifying by consequences, autonomy and reversibility rather than architecture. The author adds that authority is not static: the same coding assistant becomes far riskier once given repository write access and deployment rights, so approval to use a model must not become approval for every action it later takes.
• Validating components cannot establish that the integrated decision process is sound. In the author's credit-underwriting example, the credit model, document extraction and ratio calculations may each pass. Yet the agent can still pull the wrong customer's statements, use an outdated statement, apply the wrong debt definition or call a model outside its approved population. The author splits the work into model validation, orchestration validation (state, tool calls, error propagation, permissions) and decision assurance (evidence, constraints, delegated authority). They call orchestration failures "the risk nobody owns", since MRM, InfoSec, Technology Risk and Compliance can each pass their piece. The business should own the end-to-end process; another AI committee is not the answer.
• Human approval isn't a control; separation of authority must be designed in. The author welcomes the OCC's acknowledgement that high volumes, nominal approvals and alert fatigue hollow out human-in-the-loop review: a reviewer of thousands of daily recommendations stays accountable without any realistic ability to catch errors. Their proposal goes beyond the OCC's: the agent proposes, while independent mechanisms trace figures to authoritative records, reproduce calculations and enforce limits at run time. Humans would be reserved for judgment, exceptions and uncertainty, with execution records giving examiners evidence of what was checked and why an action was permitted. The author names this "executable assurance". They note that the OCC document itself was AI-assisted and human-approved, which shows the gap between someone accepting responsibility and anything having been verified.
• An LLM judge cannot establish that a decision is correct. The OCC notes the bias, correlated-failure and independence problems of LLM-as-a-judge; the author goes further. A second model approving a credit memo's debt-service coverage ratio and covenant reading shows only that two probabilistic models agree, not that the data, methodology or policy compliance were right. LLMs can help find where more verification is needed, but they should not be the final authority.
• Test results need a statistical framework, with discovery kept separate from confirmation. The author's illustration: zero failures in 1,000 tests still gives an approximate 95% upper bound of three failures per thousand, assuming independent, representative tests. That rate could be unacceptable for a high-volume payment system, and ordinary transactions reveal little about rare adversarial conditions. The author proposes three stages: vulnerability discovery as an optimization search (adversarial search, experimental design), independent confirmation of how reproducible and frequent each weakness is, and then a test of whether controls stop it from becoming an unauthorized action. Neither discovery nor representative testing can substitute for the other.

2026-11-04 · open the original ↗

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

2026-11-04 · open the original ↗

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

2026-11-04 · open the original ↗

• Agents can connect but cannot think together. Pandey says that across the vertical segments where his team has deployed multi-agent systems, agents exchanging semantic information hit "cognitive failures": a system can grow more consistent over time without growing more correct, and the failures compound. The host frames it with a point from his episode 1030 guest: humans who disagree on a figure argue it out, while agents can silently converge on one figure presented confidently.
• Shared words carry different meanings for different specialist agents. Pandey calls this a grounding problem: "roll back" means the last configuration to an SRE agent but the whole software version to a SWE agent, and "ready" means different things to sales, legal, finance and product on a 2 million deal moving through a deal desk. He gives a drug pipeline as a larger case, where a protein-folding agent's candidate must pass wet-lab, efficacy, regulatory (FDA pathway) and supply-chain agents.
• Every multi-agent session studied had at least one cognitive failure. Pandey reports that analysis of his team's deployments plus public ChatDev session traces and MMLU probes found failures in 100% of sessions, with output accuracy at 98% after one failure, 83% after two and 16% after three — "as good as randomly picking an outcome". He says these failures stay silent because observability dashboards remain green.

2026-11-04 · open the original ↗

• An undocumented legacy system was orphaned when its sole developer died. Griffin took on a small company whose technical partner had built a product over 20 years in Visual Basic 6 and old .NET Framework, leaving no documentation and no source control, only 14 dated copies of code folders on his development machine.
• He used the AI agent to understand existing code, not write new code. Claude helped him inventory the developer machine, and turned the company's tutorial videos and a recorded walkthrough with the non-technical partner into a map of what the product does; that work turned up several servers nobody knew about, including a support site hosted on a home server whose full disk Claude diagnosed in minutes.
• None of the saved code copies matched what was in production. They located the real source by searching the machine for the production executable's hash with PowerShell scripts, and it now lives in a GitHub repo; Griffin ships small fixes for support tickets, using Claude skills he wrote to cover VB6.
• Modernization is a business decision, not a technical one. The owners are weighing a rewrite against selling to a competitor; Griffin says AI could produce a modern proof of concept, but migrations remain risky, and this customer base could be moved one customer at a time.
• Write down how things work before you are suddenly gone. He keeps a shared repo of code plus operational notes so the client is not exposed to his own absence, and recommends source control, documentation and a shared team password manager, since AI works far faster when the pieces are findable.

2026-11-04 · open the original ↗

• Record query growth means Google's search worries are overdone. The guest notes search revenue missed plan for a second straight quarter and YouTube also missed, reviving fears that AI agents, which don't look at ads, will disintermediate search. His counter is that Google reported its highest query growth in 5 years, so usage is there and the remaining task is showing more ads on those results.
• Consumers will reach AI through their phones, favouring Google and Apple. He sees the end state as a user asking Siri or Android to dispatch an agent, which leaves the two smartphone ecosystem owners well positioned. He points to Apple now paying Google for AI, the reverse of search where Google paid Apple, and to Android running the other 75% of the world's phones.
• Wait for the dust to settle before buying the pullback. Asked whether to buy immediately after Google's 30% drop from its high, the guest says he doesn't know and prefers to let investor perception settle first. He expects Google to have a great year next year but warns against getting caught in a big drawdown, since how long investors take to come round is unknown.

2026-11-04 · open the original ↗

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

2026-11-04 · open the original ↗

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

2026-11-03 · open the original ↗

• Smart wall lights can be made into a live network-throughput display. The host adapted an LED network display project from adamslab.io to the smart wall lights in his studio office, switched on from a Stream Deck button. One half of the lights shows upload and the other download.
• Light length and colour scale with throughput up to the line's capacity. On his gigabit symmetrical office connection, a speed test drove the lights out to their maximum length from the centre as traffic neared 1 Gbps. They turned progressively more orange as speed rose, and went back to normal once the test ended.

2026-11-03 · open the original ↗

• A high dividend yield mostly signals a lower price, not danger. The hosts argue the payout comes from revenue, so a falling price alone says little; a dividend cut is the warning sign, while eliminating one to chase growth is not (they cite Disney, and urging IG Group to drop its 5% dividend). Neither host owns dividend stocks, and both say they prefer growth.
• OpenAI's short raise shows the AI bloom is off. As reported on the show, OpenAI secured about $11 billion against a $30 billion target (~37%) at a $1.25 trillion valuation versus $1.4 trillion sought. One host can't see the spending pace continuing and expects an AI miss to trigger a broader decline; the other says to trade listed AI names by IV rank and stick to established names rather than fringe ones.
• The hosts expect the Iran strike reports to be already priced. The show cites New York Times, Axios and The Hill reports of strike plans for after the midterms, noting Al Jazeera's point that preparation is not an order. One host warns renewed conflict would push oil up. Their advice is to pick risk you're comfortable owning, since defined-risk trades can still carry plenty of risk depending on width and delta.
• Election night should be a non-event during trading hours. Tom expects a Democratic sweep that is already priced in. Tony's fact-check of 2016 (Dow futures limit-down overnight) cuts against "nothing happens," though he agrees it happened late at night. Cited history: the S&P was higher 12 months after all 19 midterms since 1950, averaging ~15%, but day-after moves were mixed (+2% in 2018, −2% in 2022).
• A put debit spread hedges cheaply but caps the protection. Their example: on a $100 stock, a 90/85 put spread costing $1.50 pays at most $3.50. A long put protects all the way down but decays. One host prefers buying a put and selling a call. Another argues a few hundred dollars can't hedge a $5,000 risk, so sell half the shares instead.

2026-11-03 · open the original ↗

• Crude is no longer scarce; barrels are now looking for buyers. On Monday, November 2nd, Brent fell more than 7% to settle at about $105 and WTI fell under $97, which the host attributes to progress in the Doha talks (Qatar says both sides are studying the plan with "clear progress") and to a turn in the physical market: Bloomberg reports the premium buyers pay for real Middle East barrels fell from more than $30 a barrel in September to nearly zero. Yet on the morning of November 3rd oil bounced almost 2% on an unconfirmed report of an American strike on an Iranian island, which the host takes as a sign that one headline still moves the market.
• Ukraine's refinery strikes on Russia are continuing at a daily pace. Drones set fire to a refinery at Yaroslavl, northeast of Moscow, on the night of November 2nd, injuring two workers; the host counts that as four Russian refineries hit in four days.
• Russia's diesel export ban is about control, not stopping flows. On October 31st Russia extended its ban on producer exports of diesel, marine fuel and gas oil to January 31st, with the existing exceptions kept in place. Russia still plans to export over 2 million tons of diesel in November, about four times the October figure — the "control" reading is the host's own inference from that gap.
• Fuel is scarce enough that idle refineries are worth reviving. Valero will pay about $320 million for the Houston refinery LyondellBasell closed, plans to restart it by the end of 2027, and the plant can run 165,000 barrels per day. The host reads the deal as a sign that fuel is scarce, and his bottom line is that while hope has lowered the crude price, the refined-products market "is not catching up yet."
• Total's outlook says new oil investment is needed in every scenario. Against today's roughly 101 million barrels a day of oil-product demand, Total's business-as-usual case edges up to 102 million in 2035, peaks around 2035–2040 and falls to 93 million by 2050; its middle case (faster EV growth and electrification) falls to 96 million in 2035 and 74 million in 2050; its fastest green case drops to 53 million by 2050. The point the host singles out: in every scenario old fields decline faster than demand, so even the greenest forecast still has the world burning about half of today's oil in 2050.

2026-11-03 · open the original ↗

• The hosts read the data as pointing to a Democratic wave. Reported on air: about 50 million early votes, outpacing the last midterm in key Senate battlegrounds; Democrats passing their entire 2022 Florida turnout by 8:30 a.m.; Kalshi at a 77% Democratic Senate chance with leads in all seven battlegrounds; pollster Lakshya Jain's model at 53–47 Senate and 239–196 House; and the New York Times poll with Abdul El-Sayed up 11 in Michigan (16 under its old methodology). Tarlov notes that unlike the 2024 Selzer poll there are no outliers among high-quality polls; these are pre-result snapshots.
• Record gas prices hit swing voters where Republicans can least afford it. Tarlov cites a record Election Day national average of $4.12 for regular and $6.28 for diesel. Galloway argues men, the swingiest voters, drive about 30% more than women, while recent job growth went to healthcare (70–80% women) and losses hit manufacturing and transportation (70–80% men), with young male college graduates facing high unemployment.
• Democrats should pass an agenda and make Trump veto it. Galloway wants bills on a Medicare option, ACA subsidy relief (citing 40% of households carrying medical or dental debt), a corporate minimum tax, antitrust, Section 230 changes, national service, cutting defense from about $1.5 trillion toward $700–900 billion, and codifying Roe — "from the party of indignance to the party of ideas." Tarlov would lead with affordability rather than investigations, which she thinks energize Republican voters, and build the platform from winning candidates' agendas, since Democrats still poll about 10 points less popular than Republicans nationally.
• Galloway predicts a Democratic Congress will haul Musk before hearings. The likely targets are DOGE, the firing of inspectors general probing his companies, government contracts, and Starlink throttling in Ukraine that Tarlov says Wall Street Journal reporting tied to deliberate decisions. Galloway notes a Trump pardon would not cover state or UK/Canadian prosecution and might undercut his ability to plead the Fifth; Tarlov cautions against attacking the popular "Trump accounts."
• Galloway backs Trump's stated plan to end US aid to Israel. Trump said on Fox that the $4 billion annual aid is cut off at year-end by agreement with Netanyahu; Galloway, calling himself pro-Israel, argues Israel's economy grew from about $40 billion to $700 billion and the aid is effectively a US-arms subsidy not worth the controversy. Tarlov flags under-covered WSJ/AP reporting of Iranian damage to about 24 structures at US bases across seven countries, and both report a media expectation that strikes on Iran resume after the midterms.

2026-11-03 · open the original ↗

• An AI disappointment is the likeliest trigger for a lower market. The question is framed by a report that OpenAI raised less than half of a $30 billion target, at a lower valuation than it wanted, in a week the Nasdaq 100 sold off hard. One host predicts the market goes lower and says global AI spending is so large that "something in AI just misses" is the most obvious cause, while conceding nobody knows.
• The AI pace cannot hold if rates stay put or rise. Asked whether the bubble is bursting, one host says only that "the bloom is off the rose" and declines to call it. Their argument is that a falling market drags good stocks down as collateral damage, so holders should not assume quality protects them; whether a dip is a buying opportunity depends on where the market sits in three to six months, which they say they cannot know.
• Drop the top-tick mentality: a former high need not return. One host warns against assuming a stock must go back to a price it once traded at, and the other notes this is a recurring point on the show.
• To buy AI weakness, use established names, not fringe stocks. For anyone wanting to "catch a falling knife", the hosts point back to their usual method: check IV rank, then pick the strategy. They say to stick with "the chalk", as they argued for crypto, because fringe names carry more volatility and break outside their expected moves more often. They name no tickers, levels or specific hedges, despite the title's promise to protect profits.

2026-11-03 · open the original ↗

• A flipped Congress can price executive strategy, not set it. The President keeps the pen on tariffs, sanctions, troop movements and deals; the majority gains the purse, subpoena, Senate confirmation gate and floor calendar, with the 60-vote filibuster and two-thirds veto override intact. The brief's first-order effect: Beijing, Tehran, Ottawa and Moscow start discounting executive commitments as two-year options rather than durable policy.
• Trade is the most exposed strategy because the replacement statutes are weaker. With emergency-powers tariffs struck down by the Supreme Court, the temporary 10% global tariff under Section 122 is capped at 150 days and a hostile majority simply lets it lapse; Section 338 Canada tariffs face a repeal vote, and an emergency-tariff refund-with-interest bill becomes a live Treasury liability. Projection: everything possible migrates to Sections 232 and 301 before January, and every bilateral deal carries a 2027 expiry, strengthening Ottawa's hand.
• Iran war-powers fights shift to defence appropriations under a Democratic Senate. The Senate has narrowly defeated seven war powers resolutions on the operation begun 28 February; flipped, they pass, the President vetoes, and a funding fence is attached to the defence bill without needing 60 votes. Projection: the Pentagon plans around a funding horizon, so watch for escalation-to-settle in the lame duck, not de-escalation.
• The dollar, Treasury and Fed channel is the most consequential. A Democratic Senate puts Fed governors and the next Chair through a Banking Committee testing independence, making any devaluation sequence needing a compliant Fed harder. It also revives the debt ceiling as opposition leverage for the first time since 2010, and opens stablecoin implementation, FX intervention and shadow-banking channels for sanctioned oil to oversight.
• Senate control matters more than the House margin. Markets priced a Democratic sweep at 72%, a Republican Senate with Democratic House at 23%, and a Republican hold at 6%; the split leaves the executive its staffing, war-powers floor and Iran operation. The brief argues 51 versus 49 Democratic senators matters more than 241 versus 225 House seats, and treats November–December 2026 as the real window for lock-ins.

2026-11-03 · open the original ↗

• A bipartisan Senate bill would rein in the Exchange Stabilization Fund. He reports the transparency bill restricts Treasury Secretary Bessent's use of the $219 billion fund, requiring notice to Congress before spending, and its text aims to stop it becoming a slush fund for the administration's foreign political friends. He ties it to the $20 billion promised to Argentina and bailout promises to the GCC states.
• A double flip leaves strategy with the President but shortens it. Reading from the scenario brief, he says Congress gets the purse, oversight, confirmations and floor votes, while the filibuster and veto survive. The temporary 10% global tariff expires after 150 days without a congressional vote, foreign capitals treat commitments as two-year options, and deals now carry a 2027 expiry that helps Ottawa.
• The dollar and Fed channel is where he sees the stakes. A Democratic Senate's Banking Committee would test Fed independence, the debt ceiling returns as leverage, and stablecoin or FX moves become discoverable. His own view: this matters because a convertible gold bond and a stablecoin push are coming from a small group around Bessent he calls a cabal trying to change the entire financial system.
• The lame duck is critical, though he doubts Republicans will charge ahead. The brief projects lock-ins on defence funding, confirmations, Iran escalation-to-settle and Section 232 findings from November to December. He agrees, except that if both houses are lost, remaining Republicans may be less willing to keep pushing the old agenda because the country has changed under them; a split outcome, he says, leaves things largely as they are.
• The five factors outlast Trump, and Democrats have no plan. He repeats that the five factors are here to stay and Trump is only an accelerator of them, for better or worse. His criticism is that Democrats offer no five-factor-style plan, only anti-Trump positioning, a weak hand; he cites Rahm Emanuel saying Republicans lose the election today, and the next year decides whether Democrats won it.

2026-11-03 · open the original ↗

• A leveraged loser must raise cash immediately from whatever sells. The speaker explains margin calls with a casino analogy. A gambler owes the casino $10,000 at 3:00 a.m., can't reach the credit card company until 8:00, and has to hand over the car keys. In the analogy, a gold contract is the car: it can be sold at any hour for fast cash.
• Recent gold and silver declines are forced selling, not weaker fundamentals. The speaker says the continual drop in prices over the last few weeks happened because metals were acting as an "emergency ATM for a bleeding financial system", and that supply and demand have not changed. This is the speaker's interpretation, not a sourced finding.

2026-11-03 · open the original ↗

• American society has lost the ability to delay gratification. He sees it in personal debt that finances lifestyles people can't afford, in politicians promising benefits now and pushing the cost forward "because the next election is always closer than the next generation", in companies chasing the next quarter rather than building for 20 years, and in health and relationships. This is a cultural essay, not an energy-markets piece.
• Social media has poured gasoline on the problem. He argues infinite scrolling, instant validation, one-click purchases, same-day delivery and short-form video are built to remove waiting entirely. His own evidence is his youngest son, whose screen time he limits because the boy wants only 15-second clips and can't sit through a 30-minute sitcom.
• The inconvenience we engineered away taught patience and planning. He calls delayed gratification a survival skill, not an old-fashioned virtue, and says civilization rests on it: taxes paid today for a bridge that takes five years to build and lasts 50 to 100, schooling for an economy 20 years out, saving in good times, and planners who fund projects they won't live to see.
• A comfort-obsessed culture runs fine until the bills come due. In his reading, maintenance gets postponed, debt is normalized, hard decisions are avoided and leaders say what people want to hear. That works for a while, then the credit card maxes out, infrastructure breaks, underfunded retirement systems fail and debt service becomes too large.
• Younger adults need to be told to wait. He cites graduates financing a new car when their old one worked, and people replacing phones every year or two; he has had about three phones in 15 years. He argues working toward something teaches more than the reward itself, and admits his own DoorDash habit. He closes with an election-day call to vote.

2026-11-03 · open the original ↗

• Heavy AI spending hit US tech earnings. Reported: Microsoft beat expectations, but Alphabet spent $51 billion in the quarter on AI servers and networking equipment. Together with a regulatory fine, that left its cash flow $9 billion negative, and its shares fell nearly 11%. Marvell Technology, a data-center chip company, was also down 11% on the day.
• Further AI stock losses would push metals lower first. The speaker predicts that if AI tech stocks keep falling over the coming days and weeks, gold and silver will keep dropping in the short term. The mechanism given is fund managers selling metal contracts to cover margin calls.
• Within weeks to months, buyers should move back into physical metal. The speaker forecasts that over the next several weeks to a few months, investors will move wealth into physical gold and silver to protect it, and prices will turn back up. The advice to viewers, offered as opinion, is to stick to their strategy through the selloff.

2026-11-03 · open the original ↗

• Holding one tech stock for a decade is a mistake. Asked by Frost for a single ten-year holding, Dan Niles called the premise "ridiculous": in technology, creative destruction means you might get lucky but more often won't. He pointed to AOL, Netscape, MySpace and Nokia, which once had 40% smartphone share, and said tech investors have to stay flexible.
• Niles expects Apple to be a great stock next year. His reason is a foldable iPhone (named in the captions as "the Duo"), which he predicts will drive a huge upgrade cycle; he notes Samsung has had a foldable since 2019. He also argues the top 25% of consumers will access AI through their smartphones, and that Apple will benefit from that. This is a one-year call, not the ten-year pick he was asked for.

2026-11-03 · open the original ↗

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

2026-11-03 · open the original ↗

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

2026-11-03 · open the original ↗

• Connected agents still fail silently because they don't share meaning. Vijoy Pandey (head of Cisco's Outshift) reports that across Outshift's own deployments and public ChatDev and MMLU traces, 100% of multi-agent sessions had at least one cognitive failure, while observability dashboards stayed green. He says the failures compound: one failure left output accuracy at 98%, two dropped it to 83%, and three to 16%. His example is an SRE agent and a software-engineering agent who each take "roll back" to mean something different (last config vs. whole software version).
• Agents need a cognition layer above today's connectivity stack. He sets out three stages: aligning on shared intent and grounding terms; coordinating and negotiating from local incentives toward a global optimum; and verifying the outcome at the semantic layer. Outshift calls this the "internet of cognition". He likens it to the step from individual to collective intelligence that he says language brought about 70,000 years ago, credits Michael Tomasello's work for the idea, and says AGNTCY, A2A and MCP are converging under the Linux Foundation for the connection layer.
• The open-source fix works like lane assist, triggering only on drift. Outshift's observability and explainability platform, OXP, reads OpenTelemetry traces to detect intent, coordination or verification drift. On intent drift it triggers Accord, a four-phase protocol that locks a shared intent contract before work starts; on consensus drift during a task it triggers Concord to resolve conflicts. Both are open source, can be deployed as a skill, as an Envoy or eBPF sidecar, or through bindings over A2A, MCP or SLIM, and otherwise leave normal agent traffic untouched.
• Useful quantum scale should come from networking machines, not enlarging them. He says practical problems need roughly 100,000 to 1 million physical qubits against about 1,000 in the best machines today, so Cisco wants to link around a hundred 1,000-qubit nodes, routing qubits without measuring them since store-and-forward is impossible. He reports that in February, with Qunnect, NYU and a fiber provider, Cisco ran entanglement swapping over 17.6 km of real Manhattan–Brooklyn fiber through the 60 Hudson carrier hotel, at rates 5,000 times better than anything previously shown even in a lab.
• Probabilistic AI software running on deterministic hardware is a costly mismatch. He argues that matrix maths on GPUs only approximates the distributions agentic workloads produce, at a cost in energy, money and size. Besides quantum, he points to startups building silicon that computes on distributions: Extropic's thermodynamic chips (now shipping, he says) settle into low-energy states, and Unconventional AI's oscillator-based chips settle into synchronized patterns. Both give local optima, which he says is what large search problems like drug discovery actually need.

2026-11-03 · open the original ↗

• This looks like 1997–98, not the 2000 top. Niles puts AI investment at about 3.5% of GDP, more than highways, electrification and the internet combined, so he thinks a bust is inevitable. He still expects the trade to rip higher first, because agentic AI needs 10–100x more tokens and enterprise adoption is only starting. His analogy is Cisco: drawdowns of 38% and 37% in 1997 and 1998, yet the stock finished those years up 31% and 150%, before the real 80%+ decline from 2000.
• Two forces drove the pullback: dearer money and slower AI growth. On financing, he cites neocloud deals pushed out (two Nvidia-backed names, including Nscale, plus AirTrunk), CoreWeave debt trading at 15%, and credit default swaps surging even for Meta and Nvidia. On growth, Ramp-index spend on OpenAI and Anthropic rose about 9x from January to early August and has been flat since, while tokens grew about 80%. Anthropic's sequential revenue growth slowed from 145% in Q2 to 45% in Q3, which he puts down to cheap open-weight models and Google's low-cost frontier model.
• If Anthropic had to pull its IPO, the top is probably in. He calls Anthropic the stronger lab because it focused on enterprise, which pays; he puts OpenAI's paying consumers at roughly 5%. His forecast is that the deal gets done, but at a much lower valuation and raising less than expected 30–90 days ago. He thinks OpenAI gets squeezed between Anthropic in enterprise and Google in consumer.
• Google's selloff will become a buying opportunity, but not yet. The stock fell 12% in a day after earnings and is down 31% from its peak on a second straight search miss. He holds that query growth at a five-year high means the issue is monetization, the way Meta at IPO had smartphone users but no mobile ads. He wants the stock to bottom first, citing Facebook's 75% fall in 2022, and expects a great year for Google next year, which runs about 75% of the world's phones.
• Fed hikes and $100 oil are the macro risks he fears. He would be "beyond stunned" if the Fed didn't raise in December and keep raising into next year once the midterms pass. He notes that 10 of the last 12 recessions followed a sustained rise in oil, with Brent still above $100 as of recording. Both a rising oil price and Fed hikes came before the 2000 bust, which he says scares him a lot.

2026-11-02 · open the original ↗

• SpaceX had no product managers, so engineers did the founder's job. One of the two ex-SpaceX founders in the clip (both from YC's fall 2025 batch) says engineers there owned everything from choosing which problem to solve to building the tool that solved it. That meant taking things from zero to one while holding them to a high standard of rigor and correctness.
• Picking the hard job over the comfortable one was the training. One speaker, fresh out of a master's degree, turned down an offer from Sonos — a company they had interned at and liked, with a beach office in Santa Barbara — for SpaceX. They wanted to learn from people who knew how to solve very hard problems.
• Both founders say timing and SpaceX experience drove the decision. Boot Loop's CTO, formerly firmware lead for the Raptor engine and software lead on the booster-catch project, calls this the best time to start an AI company, especially in hardware, and wants to bring what was built at SpaceX to more industries. The Layers founder, who built internal and customer-facing AI tooling for Starlink customer support, says that work gave a distinct view of what to build.
• The Layers founder quit without lining up what came next. Colleagues thought it crazy not to work out the next step first, but the founder describes being unable to half-do two things at once and having enough confidence to figure it out.

2026-11-02 · open the original ↗

• Trump's "biggest oil drop in history" claim is false. In a played clip Trump calls the move a $10 fall, the largest ever, and says it will go much further "because we're winning with Iran." He counters that the day's fall was about 7.5%, just under $8 a barrel, against an April fall of roughly 17%, about $18 a barrel, so it is not even this year's largest.
• The timing of a billion-dollar short warrants regulatory scrutiny. He reports that on Friday WTI closed above $104 a barrel, up about 2%, and that a roughly $1 billion short was placed right at the close. He argues fundamentals only worsened over the weekend. He says this does not prove manipulation, since a trader may simply have anticipated a pre-election announcement, but cites an earlier $550 million pre-announcement trade this year that drew regulatory and congressional investigation.
• The selloff came too late to lower pump prices before voting. He estimates futures moves take three to four weeks to reach the pump, because crude must be refined, moved through pipelines and terminals, and stations must sell through existing stock. He cites AAA's national average of $4.17 a gallon, a record for any November 2nd even after recent declines. He reads the late timing as forced: an earlier move risked Iran undoing it. Only the headline changes overnight.
• He expects the drop to prove a buy-the-dip opportunity. His view is that nobody believes the move reflects real repricing. He notes Iran said the same day that negotiations are making no progress, and he characterizes the diplomacy reports as floated alongside the short positions.

2026-11-02 · open the original ↗

• Iran needs offshore purchasing power, not dollars at home. The brief lays out a four-layer pipeline: oil sold mainly for yuan into front-company accounts (layer 1), bank-controlled 'leader' companies for Melli, Shahr and Mellat (layer 2), Tehran exchange houses netting oil receivables against importers' payables (layer 3), and only then hard-currency access via third-country banks, stablecoins, and gold or cash couriers (layer 4). China, the UAE, Hong Kong and Singapore account for roughly 96% of identified shadow-banking activity.
• Enforcement has repriced the pipeline rather than closed it. The June 2025 Sepah/Nobitex cyber strikes, the Oct 2025 and Oct 2026 leaks, and Treasury's Economic Fury and Economic Outcast (24 Aug 2026) actions each raise intermediaries' spreads, so realised hard currency per barrel falls even where volumes hold. The brief labels this a framework projection: the hit lands on Guard and proxy security spending, not on oil moved.
• Layer 4 dollar access is the vulnerability; shells are not. Shells re-register cheaply, but a correspondent account like Banque Misr UAE (about US$1.8bn for 103 Iran-tied companies, 2024–2026; cut 28 Aug) takes months to replace. The authors project that as dollar touchpoints burn, settlement drifts further into yuan-in, yuan-out trade offsets, but call that a weak indicator of reserve-currency fragmentation because the channel is coerced and discounted.
• Enforcement only sees a fraction of the flow. FinCEN identified about US$9bn for 2024 against Treasury's own 'tens of billions', a gap the brief flags for reconciliation before either is cited. Treasury's October 2026 claim that the system is 'buckling' is unverified, and Iran International reported in September 2026 that more than a dozen Chinese and UAE banks were untouched by the August designations.
• The Binance hack shows the stablecoin issuer is the chokepoint. An attacker took about US$6.3bn on 29 Oct 2026 (attributed by TRM Labs and Chainalysis to a Lazarus-linked unit, not confirmed); Tether and Circle froze US$2.3–2.8bn within an hour, about seven times April's US$344m Iranian central-bank freeze. Iran was not the target, but Binance carried over US$1.5bn of alleged Iranian oil proceeds in 2024–2025; the brief projects tighter compliance and migration to Tron-Tether, OTC desks and smaller exchanges.

2026-11-02 · open the original ↗

• Russia's gas cutoff is accelerating the shift to alternative energy. Under the Five Factor 'secure, then control' logic, a country cut off first secures supply, then seeks control of most of its own energy. He cites $3.4 trillion of energy spending, $1.2 trillion on oil, gas and coal versus $2.2 trillion on alternatives and renewables, and expects that gap to widen.
• Iran's real problem is getting oil yuan back into dollars. Reading from the ATF brief, he explains that oil is sold for yuan into front-company accounts, netted by 'leader' companies and Tehran exchange houses against import payables, and only reaches dollars at a fourth layer: third-country correspondent banks, stablecoins, and gold or cash couriers. China, the UAE, Hong Kong and Singapore account for about 96% of identified activity.
• Sanctions raise Iran's laundering cost rather than stopping the flow. Treasury's Economic Fury designations (three exchange houses, 13 front companies) and later actions mean, in his illustration, cleaning money that cost 10% now costs 15%. Layers one through three are cheap to rebuild; the fourth, dollar-access layer is where U.S. enforcement bites, and blocked flow shifts toward paying for Chinese goods directly in yuan.
• Stablecoins carry a U.S.-controlled kill switch. The central bank reportedly bought about $507 million of Tether in January 2026, and Tether froze about $344 million of Iranian funds in April 2026. His own characterisation is that stablecoins are effectively controlled by the U.S. government; he notes Treasury's claim that the network is 'buckling' is unverified and the evidence may point the other way.
• The Binance hack lowers Iran's dollars per barrel further. Roughly $6.3 billion was taken across 10 blockchains in about 20 minutes, followed by about $3.6 billion of net outflows after reopening and $2.3–2.8 billion of stablecoins frozen within an hour. He doubts the North Korea attribution and concludes Iran's crypto channel reprices again: higher intermediary spreads, fewer venues, though whether Iranian money was taken is unknown.

2026-11-02 · open the original ↗

• Don't sell a losing position at its recent bottom. Answering a 30-year-old viewer down 25% on a Google position, both hosts say they would not cut it here; they put the stock at "300 and change" before earnings and about 280 now, and concede it could go a lot lower and that they have no idea where it goes.
• A 20% single-stock weighting is too much; trim into a rally. Google is about 20% of the viewer's long stock portfolio, which both hosts call a big share and something you have to be willing to live with. Their advice is to lighten up on a bounce, one host naming roughly 300 to 320, rather than at current levels.

2026-11-02 · open the original ↗

• Scott frames free election-week coverage as a civic obligation. He argues Americans deserve unfettered access to news and analysis, especially in election season, and that an independent media company has a role in providing it; the post announces five free livestreams with Jessica Tarlov and Aaron Parnas, November 2–6, 2026 (Raging Moderates Monday–Thursday, Raging Perspective Friday), plus the free Monday Rage newsletter's "Anxious Democrat's Guide to the Big 11 Races on Election Night" with hour-by-hour timing and early polling guidance.
• Anthropic's S-1 filing is public and drew heavy reader interest. Per the post, head of research Mia Silverio's Extra Credit issue unpacking the Anthropic S-1 was the series' bestselling issue to date, followed on Friday by a video debrief with markets host Ed Elson; the analysis itself is paywalled and not in the free text.
• Lula won a fourth term, which Scott calls unlikely. A paid Deep Dive released the same day has associate producer Laura Janer, reporting from Brazil, interviewing Brian Winter on how Luiz Inácio Lula da Silva clinched the term and what America can learn from Brazil's election; the post gives only this preview, not the conclusions.

2026-11-02 · open the original ↗

• Earnings Radar is pitched as testing options' earnings forecasts. This is a product announcement for two new tools in Lossdog's Volatility Lab. The first is said to cover earnings history for about 5,000 stocks, showing the expected move in dollars and percent, how often stocks stayed inside or blew past it, and the typical post-earnings drop in implied volatility.
• Volatility Landscape is pitched as IV rank with deeper history. The video contrasts it with IV rank tools that look back 12 months, claiming up to 19 years and more than 21 million volatility observations, with comparisons that include or exclude estimated earnings effects. These are the product's claimed capabilities; no results are shown.

2026-11-02 · open the original ↗

• Trading small and often beats waiting for high-conviction setups. One host argues his own "high-conviction" trades have been 50-50 at best, perhaps 40-60 against him. His answer is to stay involved and vary the strategy, for example defined-risk diagonals after 20–30% earnings moves rather than naked short puts or calls. The other agrees, but concedes that long-haul investors buying at all-time highs have generally done better.
• The market now moves with a dozen AI names. One host argues that bonds, crude, gold, crypto, politics and most earnings no longer move the index; crude at 103–104 and bonds at 101 failed to. The other accepts this but reads Friday's selloff as a warning rather than a one-off, pointing to Meta about 150 points off its high, Micron down nearly a third from 1,255, and AMD at about 503 against a 658 high.
• No objective threshold defines oversold in a big-cap stock. The question followed Alphabet's 13% drop Thursday and Intel's roughly 30% drop Friday. One host recalls buying too early in 2001, late 2008 (30% lower by March 2009) and 2020; the other treats a move through the expected move as oversold. Both favour selling puts into the high volatility. Sosnoff's four trades did that in Reddit, Robinhood, Intel and SpaceX, at roughly 84% POP and 16 delta, with $1,000–$1,900 buying power each.
• Oil's fall is a midterm-election trade, not a war trade. That is Sosnoff's argument. With the traded crude contract around 97 and down about $7, he predicts $10 lower before $5 higher and a cap near 100, even if Iran retaliates. He says renewed bombing would need Republicans holding both houses, which he expects won't happen. His co-host is less sure. Sosnoff says to trade oil only through /CL options, which are one tick wide however many legs, or MCL for futures.
• Put credit spreads give defined-risk exposure to a bounce. One host sets them at 35–50 days (45 optimal) with the short strike at the expected move, and says the credit is roughly the short delta times the width, so POP is about 100 minus that. The cost is capped upside: a $1 credit on a $100 stock's 90/85 spread risks $4. Owning 100 shares instead loses $1,000 at $90.

2026-11-02 · open the original ↗

• Stevens' endorsement of Rogers won't change Michigan's outcome. Reported: Haley Stevens skipped a unity event and endorsed Republican Mike Rogers at a Detroit church; Rogers' Kalshi odds rose from 23% to 32% (El-Sayed still 68%), El-Sayed raised $2.4 million on Sunday, and half her staff reportedly resigned. Galloway says nothing about El-Sayed changed between Saturday and Sunday except her grievance and predicts it ends her political career; Tarlov says the Jewish-voter concerns behind it are real but cites the final New York Times/Siena poll with El-Sayed up 11 (16 under the 2024 methodology).
• The Democratic wave is widening rather than narrowing at the end. The last New York Times/Siena poll of the cycle has Democrats leading in five Senate battlegrounds including Texas and Kansas, an edge in Alaska and a virtual tie in Iowa, with Trump approval at 38% across those states and Democrats +9 on cost of living. Galloway cites young voters moving from 53–44 Democratic in 2024 to 68–23 now and a growing pool of swing voters, and predicts a "super tsunami"; Tarlov expects Republican House gerrymanders in Florida and Texas to yield one or two seats rather than the four and five planned.
• Turek's hard line on trans issues is smart Iowa politics. Josh Turek's ad opposes males in female sports and taxpayer-funded sex changes; Tarlov calls these 80-20 issues and says Democrats now hold an eight-point advantage on them because Republicans overreached. Galloway urges Democrats to stress economics and move toward the center, noting the Iowa Capital Dispatch estimate of fewer than 500 transgender girls in Iowa middle and high schools when the 2022 ban passed.
• The hosts split on whether Bass survives the LA mayoral race. A new poll has Karen Bass and Nithya Raman tied at 40%, eight of the previous nine polls showed Raman ahead, and an October LA Times/UC Berkeley IGS survey found 69% of likely voters under 30 and 74% of those 30–39 favor Raman. Tarlov predicts Bass wins as the known quantity, helped by Rick Caruso's backing; Galloway predicts Bass loses as a weak incumbent and calls the fight over Raman's past opposition to the 2028 Olympics a non-issue.
• Galloway argues Democrats, not Republicans, are the party of growth. Averaging administrations since 1980 from a USAFacts data packet, he finds Democratic administrations added about 2 million jobs a year versus 1 million, GDP growth of 4.1% versus 2%, and annual deficit additions of 2.1% of GDP versus 2.8%. He pitches this directly to young men, whom he calls the most volatile swing voters and the most focused on economics.

2026-11-02 · open the original ↗

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

2026-11-02 · open the original ↗

• A reluctant Republican incumbent puts South Carolina within reach. The New York Times/Siena poll of Oct 21-26 has Dr. Annie Andrews and Sen. Darline Graham tied at 48% in a state Trump won by 18 points in 2024; the author points to Graham telling a local Republican club she would "choke Lindsey" in heaven, and to Andrews' cost-of-living campaign and Fox News appearance, and judges it "will take a good night for Andrews to win, but not a whole lot more."
• Maine turns on how far Collins outruns her polls. CNN/SSRS (Oct 20-26) has Troy Jackson up 49%-45%, while Fox News (Oct 16-20) has a 49% tie; the author notes Collins has beaten her polling several times before, not only in 2020, but argues her votes for Justices Gorsuch and Kavanaugh are real liabilities and that women could decide the race.
• Michigan is close because the Trump base is energized against El-Sayed. CNN/SSRS (Oct 20-27) has Abdul El-Sayed up 49%-42% over Mike Rogers, but Fox News (Sep 24-28) had it at 50%-49%, in a state Trump carried by 1.4 points. Democratic Rep. Haley Stevens endorsed Rogers over the weekend, which Senate Minority Leader Chuck Schumer called a "despicable betrayal."
• Texas worries Democrats for its history, not its candidate. CNN/SSRS (Oct 19-26) has James Talarico up 49%-43% over Ken Paxton, while Fox News (Oct 15-19) has a 49% tie; the author credits Talarico with setting aside the "woke" attacks early and avoiding a prolonged fight with Rep. Jasmine Crockett, but notes Texas has elected no statewide Democrat in 32 years and says "we'll wait to see the votes before we believe it."
• Alaska has tightened amid a late Republican counterattack. Democrat Peltola's lead over Sen. Dan Sullivan fell from 7 points in an October Siena poll to 48%-46% in the New York Times/Siena poll of Oct 23-Nov 1, after the Times reported operatives aligned with her campaign spent "months" recruiting a same-named candidate, now on the ballot as "undeclared." The author's counter is that Sullivan declined every chance to distance himself from Trump on tariffs and the war.

2026-11-02 · open the original ↗

• Agentic coding moved the bottleneck from writing code to validating it. A Blacksmith founder argues 100 times the code is useless if CI can't ship it at that speed, and says the company grew 20x in the last year, then raised about $66M in a Series C led by Sequoia at a $1B valuation. Next, they plan autonomous QA that gathers screenshots and video as evidence a change works, expecting developers to lack time to check things themselves.
• CI is mismatched with hyperscalers and runs best on gaming hardware. The founders, former database and search-infrastructure engineers who had never written a CI pipeline, concluded CI cares about one fast thread, like video games, not many threads. They began renting servers from a German host known for Minecraft and Call of Duty servers and now buy megawatts of power from data centers; caching artifacts near compute is described as core to the speed.
• Being naive outsiders let a good idea reach the market. One founder credits not being platform or DevOps engineers for approaching the problem differently. The YC partner who led both the seed SAFE and Series A says they had never seen gaming CPUs used for CI and was surprised anyone would rack a hundred of them to compile code.
• Early customers came from friends, logos and founder-led outbound. The first 200 customers were "extremely hard": friends' companies granted logo rights for credibility, then a founder cold-called CTOs, VPs of engineering and platform engineers, finding small fast-moving companies easiest. The partner says they landed five- and six-figure contracts with no sales team; Langchain's wait to acquire compute fell from 15–20 minutes on GCP to 10–15 seconds.
• A small product surface lets 30 people run a huge fleet. The team went from about five people and a couple hundred servers to 30 people with 12 engineers, crediting bespoke fleet and observability software and refusing one-off customer requirements. Demand spikes with every new LLM release, met with smarter job packing, while a Phoenix power surge that knocked out primary and backup cooling pushed them toward geographic distribution.

2026-11-02 · open the original ↗

• The national debt and Social Security now compete for one budget. He argues both will demand far more federal money at the same time within the next decade, alongside rising Medicare costs, so they can no longer be treated as separate conversations.
• The debt's true cost is the total interest, not the headline. With debt over $40 trillion and interest at $2.7 billion a day, he calculates that paying $1 billion a day of principal would take 111 years, adding roughly $50 trillion of interest for about $90 trillion in total. He cites the CBO projecting net interest above $2 trillion a year by 2036.
• Social Security's shortfall is easily fixed but politically avoided. He says the retirement trust fund is exhausted in 2032, after which incoming revenue would cover about 78% of scheduled benefits. The fixes — more revenue, benefit changes, a later retirement age, a higher taxable-income cap — are simple, he argues, but no one will risk an election over them.
• Deficits, debt and interest feed a self-reinforcing spiral. He describes larger deficits creating more debt and more interest, with interest the fastest-growing part of the budget eliminating fiscal flexibility. In his view growth, productivity, more workers or higher wages would help, but growth alone does not close the structural gap, contrary to Scott Bessent, so revenue may have to be part of it.
• Acting now allows gradual change; waiting forces drastic change. He argues an early fix would be far gentler than one made at the edge of the cliff, but predicts Congress will do nothing until it is an emergency it can use for political gain, leaving the math to force the most painful choice.

2026-11-02 · open the original ↗

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

2026-11-02 · open the original ↗

• Scary headlines have not produced bearish price action. He notes the "worst day and month since June 2025" headline for the NASDAQ and S&P 500, but says the S&P 500 is only consolidating in the range it has held since August. In his view, wars in Ukraine and Iran, rising bond yields, weak manufacturing, a rate hike, hot inflation and China's AI bubble popping have all failed to break the uptrend.
• Money is rotating out of AI tech, not leaving stocks. He cites resilience in finance, health care and other "recession resistant" sectors, and Russell 2000 strength on green days, as evidence of rotation. He calls the pullback in Intel, Google and Meta healthy, though he is watching Intel's market-cap drop as a red flag. He stays focused on recession-resistant names and avoids tech except for short-term trading.
• Institutional investors see a decent economy, whatever consumers feel. He points to corporate profits still climbing (BEA data) and consumer spending rising steadily despite complaints about cost of living. He guesses the spending comes from boomers and notes credit card delinquencies are led by younger generations, but argues shareholders care about profits, not grocery bills.
• Midterm gridlock should be bullish this time, he argues. He puts Democrats at roughly 94% to take the House and likely the Senate. He concedes that markets have historically done worse under a Republican president with a Democratic Congress, but predicts gridlock will end "not-normal" policies such as sweeping tariff scares, and that institutional investors will reward that predictability.
• His Pokémon-card index shows no imminent bear market. He tracks an index of the 1,000 most liquid Pokémon cards as a risk-off asset, expecting a sharp drop there to come two to three months before a stock market drop; it has mostly risen over the past days and year. He calls this a personal project. He says a big meltdown is "in the works" but sees no signals today, and reports the StockTok ETF up nearly 22% versus the S&P 500's 9.5% this year, with ETF 2.0 breaking out.

2026-11-02 · open the original ↗

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

2026-11-02 · open the original ↗

• AI agents can run semi-supervised research on improving agents. Renze, an AI researcher pursuing a doctor of engineering at Johns Hopkins, has an agent keep a backlog of hypotheses about cognitive-architecture modules, generate agent variants, and test them on cost per task, accuracy, decoherence and alignment; some weeks it runs 70-plus experiments while he steers it in daily discussions.
• Keep sessions well short of the context window. His first rule is to stay under about 200,000 tokens, roughly 20% of a million-token window, because long sessions bring decoherence after around 100 steps and "context rot"; auto-compaction at 80-90% can drop early rules such as "don't delete the database".
• Anything done more than about three times should become a skill. He builds persistent episodic, semantic and procedural memory from processed session files, promotes repeated procedures into skills, uses a discovery meta-skill to find repeated tasks, and has skills record their own failures so they improve over time.
• Agents should settle most decisions without the human. His decisions skill has the main agent (Opus 5.5) consult a cross-vendor model in Codex, then Fable 5.1 if they disagree, then hold one round of debate before deferring to him; he says he cannot personally juggle more than three to five agents at once.
• Classic software engineering practices mostly still work for agents. He says he has written essentially no code since about January because coding agents became "superhuman" given guidance like keep-it-simple and don't-repeat-yourself; his rule is to try the old practice first and only rethink it where agents' different constraints break it.

2026-11-02 · open the original ↗

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

2026-11-02 · open the original ↗

• Winning the House closes the purse's leak only with Trump's signature. A "pocket rescission" — a request to cancel funds sent less than 45 days before they expire (typically September 30) — freezes the money until it dies with no vote; the administration has used it twice and the Supreme Court let it stand without ruling on legality. A standalone fix would be vetoed, so it must ride a must-sign spending bill.
• House oversight powers end at an executive that won't enforce against itself. Criminal contempt for defying a subpoena needs a full House vote and then a DOJ referral, and DOJ stays under Trump for two years, so he expects allies and family to ignore subpoenas on crypto, trading and Kushner. Impeachment takes a simple majority but removal needs 67 senators, roughly 17 Republicans; he argues public opinion is the better venue.
• A Democratic Senate would shield the Fed through confirmations. Future Fed board seats require Senate votes, which he argues would protect governor Lisa Cook, whom Trump is still trying to fire over alleged "mortgage fraud," and the institution's independence. Confirmations also cover judges and agency heads.
• The debt limit is the most leverage any midterm result buys. Projections put the U.S. at the $41.1 trillion ceiling around May 2027, so the Congress seated in January 2027 inherits the fight; if either chamber flips, raising it needs the other party's votes. Reconciliation, used for the One Big Beautiful Bill, by contrast needs both chambers plus the president.
• Losing Congress pushes a term-limited president toward powers no vote reaches. Tariffs of 10–12.5% on about 60 economies, rebuilt on delegating trade statutes after the Supreme Court struck down emergency tariffs in February, need a court win or veto-proof law to undo, and the summer Iran war-powers resolution never reached the president's desk. Citing Bush 2006, Obama 2014 and Clinton in Kosovo in 1999, he predicts more foreign-policy moves, tariff threats and executive orders; until January the current majority still holds every lever.

2026-11-01 · open the original ↗

• Restarting the military war admits the economic war failed. He argues the months of economic pressure gave Iran time to get new weapons from Russia and China, train on them, and study how US aircraft operate and where US forces sit. In his view a second round would be fought against a better-armed adversary, after the US burned interceptors shooting down drones.
• Iran need not win, only make the cost unacceptable. He says it already has in the Strait of Hormuz: despite months of operations to protect shipping, October saw more attacks than any time in history, against none in the 16 years before the war. He expects a new round to bring more tanker, terminal and infrastructure strikes, rerouting, and higher oil, diesel, jet fuel and shipping costs.
• US forces in the region are exposed to retaliation. Iran does not need to reach the continental US; his understanding is that it has taken out most US bases there, and he asks where the 50,000 troops are. He notes Iran has warned there would be "no geographic limits" if strikes resume.
• Nobody can define victory, so escalation traps the US. He mocks talk of strikes lasting "a few days", since the first round meant to last days has run seven months. Without solving the political problem, he argues, whatever is destroyed gets rebuilt differently, while the US spends billions keeping enough oil moving through the strait and allies turn away.
• He doubts a restart and expects a pre-midterm peace announcement. US officials say no final decisions have been made, and he questions whether ongoing peace talks are serious or timed for the midterms, two days away. He predicts a peace announcement before the vote, but warns that if strikes restart "the nightmare is just beginning."

2026-11-01 · open the original ↗

• Agentic coding has moved the bottleneck from writing code to validating it. The founders argue that writing 100 times the code is useless if it cannot ship at that speed, and that validation usually means waiting on CI. Blacksmith reports 20x growth in the past year and a Series C of about $66M at a $1B valuation led by Sequoia; next it plans autonomous QA that gathers screenshots and video as evidence a change works.
• CI hurts because hyperscalers are built for a different workload. The co-founders, database and search-infrastructure engineers who had never written a CI pipeline, saw CI's workload shape as a mismatch for AWS, GCP and Azure, and set out to build a cloud specialised for it. One founder credits being "naive enough in the right way" for not killing the idea before it met the market.
• CI behaves like a video game, so gaming CPUs suit it. The founders say CI cares about one fast thread rather than many, so they started on rented servers from a German host known for Minecraft and Call of Duty servers and now buy megawatts of data-centre power. Caching artifacts close to the compute nodes skips redundant work. The YC partner who led both their SAFE and Series A says they had never seen anyone try gaming CPUs for CI.
• Engineers are won only by performance, not by sales. The first 200 customers were "extremely hard": friends' companies granted logo rights, then a founder cold-called CTOs, VPs of engineering and platform engineers. Without a sales team they landed five- and six-figure contracts on a one-click switchover; LangChain's wait to acquire compute fell from 15–20 minutes on GCP to 10–15 seconds. A subway-takeover campaign in New York and San Francisco is now under way.
• A deliberately small product surface lets a tiny team run huge scale. The team grew from about five people and a few hundred servers to 30 people with 12 engineers, which they attribute to bespoke fleet-management and observability software and refusing one-off customer requirements. Demand spikes with every new LLM release, met through job placement and packing; a Phoenix power surge that took out primary and backup cooling taught them to distribute geographically.

2026-11-01 · open the original ↗

• Ukraine is again striking Russian refineries in quick succession. As of November 1, 2026, the host reports three refineries hit in three days: Novokuibyshevsk caught fire Friday, drones hit Ufa Saturday, and Kstovo, which he calls Russia's fourth-largest refinery and one of its biggest gasoline makers, was hit overnight. Russia struck power sites in several Ukrainian regions the same night, reportedly leaving hundreds of thousands without power.
• Gulf attacks have widened from oil tankers to gas carriers. Two ships were hit in the Persian Gulf on Friday: an oil tanker off Oman that caught fire, with Oman's Navy taking off 25 sailors, and the Qatari LNG carrier Al-Mayeda, struck south of Lavan Island. All 25 of the Al-Mayeda's crew are safe and the ship is still operating while the damage is assessed.
• America's emergency oil release is complete, leaving reserves historically low. The Energy Department approved the last loans, about 40 million barrels, completing a pledge of 172 million barrels lent during the war; the companies start paying it back with interest early next year, and what remains is the lowest level since 1982. Meanwhile, with US pump prices high, Exxon reported $15.4 billion in Q3 profit (more than double a year earlier) and Chevron $9.9 billion (nearly triple).
• Saudi Arabia is exporting flat out while routes stay open. Satellite images this week show nine supertankers loading Saudi oil at once, about 18 million barrels, and Javier Blas at Bloomberg, citing Vortexa, reports Saudi exports of about 7 million barrels a day through the Gulf and the Red Sea in the first three weeks of October. That Riyadh is shipping every barrel it can while the routes are open is the host's own reading.
• Iran's offer to reopen Hormuz is conditional and unconfirmed. Iran's foreign minister said on November 1 that the strait could reopen in two days if sanctions are lifted, all oil waivers restored and frozen money released; the US State Department says no decision has been made. The host's bottom line is that the strait is one headline from reopening or one attack from closing harder, and that the barrels, ships and fires contradict the words.

2026-11-01 · open the original ↗

• Non-bank stress is real but not yet a solvency event. As of 30 October 2026 he reads October as an equity-leverage unwind on top of a year-long retail-liquidity problem in private credit, with PE-backed life insurers as the slow tail. His transmission chain: AI-equity unwind → software credit marks (OTIC) → retail redemption queues → gated managers sell liquid public credit → spreads hit affiliated life insurers → rating and regulator action. Each link exists; none has snapped.
• Hedge fund losses are a crowding failure, not a franchise failure. Goldman's three main strategy groups all fell more than 1% on 22 October, a first since the pandemic, and its popular-longs basket fell 4.2% that day (worst since 2011), down 10.4% for the month. Arini's c.$7bn fund lost ~6.9% in three weeks on Altice International, Brightspeed and Aston Martin. Prime-broker margin, not LP redemptions, is the forcing mechanism; October's SS&C GlobeOp indicator (1.62%) predates 22 October, so the 20 November print is the one to watch.
• CTAs are the one healthy cohort, with reversal risk. The SG CTA Index returned 15.7% through Q3 versus 11.7% for the S&P 500, and stood at -1.24% MTD, +14.2% YTD on 30 October. Books are concentrated short bonds, long dollar and energy; if the AI unwind becomes a growth scare and yields reverse hard, that book flips.
• Private credit's problem is the retail vehicles, not the managers. More than $4.6bn is trapped behind 5% quarterly gates at Apollo, Ares, BlackRock and Morgan Stanley; OTIC holders sought $1.1bn (39% of shares) in Q3 versus 10–17% elsewhere. Defaults reached 5.8% (c.4% for spec-grade bonds), mostly via PIK and extensions, with some expecting 8%. Fees and Blue Owl's $49bn LTM raise are intact; the risk is a run to exit at reported NAV and forced public-credit selling.
• PE-backed life insurers are the slow-burn tail, already drawing regulators. Average private credit exposure is c.8% of assets, but some PE-backed insurers hold up to 40% in affiliated, illiquid credit, much reinsured via Bermuda's $1.52tn market. NAIC met Treasury Secretary Bessent on 7 May 2026 and claimed authority to challenge private letter ratings, which upgrade bonds over four times as often as they downgrade; challenges are due by year-end 2026. P&C downgrades (State Farm to A+) are weather and auto stories, unrelated.

2026-11-01 · open the original ↗

• Supply chains now need contiguous partners, not cheapest ones. He uses Belarusian potash as the example: even with Putin agreeing to sell, the only export route runs through Lithuania's port, and the EU blocked it. Under his five-factor framework, a source that would have been the obvious answer in the old world fails because of choke points between buyer and seller.
• China will push into eastern Russia for water and farmland. He repeats a forecast made about three years ago that China will take eastern Russia for its resources, citing northern China's water vulnerability (as discussed by Peter Zeihan) and Lake Baikal just over the border. He expects the region to become China's farming area and says China is already pressing Russia hard for access. This is his prediction, not reported fact.
• Warsh's real agenda is a Fed subordinate to Treasury. He argues the rate question was the wrong one; the right one was whether the Fed should serve Treasury, and he says Warsh believes it should. His concern is financial manipulation replacing fiscal discipline. He ties this to Japan speeding up sales of the central bank's ETF holdings, which he puts at ~50% of the stock market plus 30%+ of JGBs, as evidence every government needs money.
• Forced deleveraging hits levered hedge funds while CTAs profit. Walking through the stress-monitor slides, he describes prime brokers telling funds to de-lever by selling, with leverage that took 18 months to build unwinding in 75 days. Goldman's popular-longs basket fell 4.2% in a day and 10.4% for October; he says credit hedge funds are concentrated in hyperscalers. CTAs are up 15.7% over nine months versus 11.7% for the S&P 500, but a growth scare that reverses yields would flip their book.
• Insurers will be the big problem, and private credit marks are fraudulent. He reads $4.6bn trapped behind 5% gates, Blue Owl's tech fund facing 39% redemption requests and a 5.8% private credit default rate. In his opinion Blue Owl's $49bn raise is the industry feeding its own vehicles to avoid collapse, managers charging 2% on overstated assets will face lawsuits and jail, and PE-backed life and annuity insurers will end up bailed out.

2026-11-01 · open the original ↗

• Prof G Media is answering midterm anxiety with a free livestream week. Scott frames it as "action absorbs anxiety" (he mentions refreshing Kalshi's "blue tsunami" market, giving no odds) and lists five free Substack shows with Jessica Tarlov and Aaron Parnas: Raging Moderates on November 2 and 3 (11 a.m. ET), November 4 (1:30 p.m. ET) and November 5 (2 p.m. ET), then Raging Perspective on November 6 (1 p.m. ET).
• He casts free election coverage as an independent outlet's duty. The free livestreams are presented as affirming that commitment; the rest of the post is a promotion: 26% off Prof G+ monthly and annual subscriptions that week, an unedited conversation with Heather Cox Richardson behind the paywall, and paid-only 2027 Predictions on December 7.

2026-11-01 · open the original ↗

• Trump account earnings lose the tax advantage of college plans. The speaker states that earnings will be taxed as income, contrasting this with traditional college savings plans, where money grows tax-free.
• Automatic enrollment raises privacy concerns. Enrollment bypassed parental consent; critics, as relayed by the speaker, argue this lets the government begin tracking and collecting citizens' financial data from birth. That motive is the critics' inference, not a reported fact.
• Critics say the accounts mainly benefit wealthy families. Families able to max out the $5,000 annual contribution per child could, on this argument, use the accounts as another tax shelter.
• Donated stock could be dumped on children for a write-off. Donors can deduct the full market value of stock gifted to the accounts, while a mandatory 5-year lockup stops the child from selling. The speaker's scenario: a corporation offloads overvalued or shaky stock, keeps the deduction, and the child holds near-worthless shares.
• Seed money adds to federal fiscal pressure, the speaker argues. Billions in government deposits are said to add to a "federal deficit" the speaker puts at $40.4 trillion. Whether this speeds the dollar's decline is posed as an open question; the speaker says they are undecided and asks parents for their views.

2026-11-01 · open the original ↗

• Every eligible child gets a $1,000 government seed. Per host Kim Brackett, the Treasury deposits $1,000 for each baby with a valid Social Security number born from January 1, 2025 through the end of 2028. The child can access the account at 18, and it is pitched for college, a first home or a small business.
• Outside money can top the accounts up. Individuals, corporations and employers can add up to $5,000 a year. She reports Michael and Susan Dell depositing $250 into each account for children aged 1 to 10 in areas with median household income under $117,000.
• No fees does not make the accounts a free lunch. The Treasury invests the money in US stock index funds and states there are no direct administrative or maintenance fees, but after 18 any growth or pre-tax money withdrawn is taxed as regular income. Her point: used for tuition, the earnings are taxed, unlike traditional college savings plans where growth is tax-free.

2026-11-01 · open the original ↗

• Ukraine has resumed its refinery campaign at a sharply faster pace. Russia's deputy prime minister said three refineries and a gas chemical complex were struck on October 30, one in the Samara region; a day later drones hit the Bashneft refinery at Ufa, reportedly its catalytic reforming unit. He estimates, without full damage data, that nearly 1 million barrels a day of refining capacity may be offline, which he says once took months.
• Hitting a few critical units shuts a whole refinery. He argues Ukraine has learned the refining process and now targets distillation units, crackers and reformers rather than the storage tanks it hit early in the war. He cites a Washington Post finding that more than half of dozens of strikes damaged specific units, mostly distillation, and says sanctions make that equipment very hard for Russia to replace.
• The strikes land on a system still recovering, so damage compounds. Reuters reported Russian crude processing rose roughly 20%, back toward about 5 million barrels a day, after attacks eased in the second half of October, and diesel output was recovering. He reads the outages as cumulative: with refineries already under repair and spare parts scarce, he expects Russia may lose capacity faster than it can fix it.
• US sanctions relief covered diesel, the product Ukraine is now removing. He plays Energy Secretary Chris Wright saying General License 135 suspended sanctions for six months on Russian diesel only, not oil. He argues this leaves Russia with surplus crude it cannot easily sell, possibly forcing Trump to lift oil sanctions too, and predicts crude prices could soften while diesel rises sharply.
• He expects higher diesel prices, including in the US. He describes global diesel as already extremely tight, citing Phillips 66's realized refining margin of $38 a barrel and Houthi attacks on Saudi refineries. He declines to endorse forecasts of $7-a-gallon diesel without damage data, and doubts that a pre-midterm Iran peace-deal announcement from Trump, which he thinks possible, would convince markets or slow diesel.

2026-11-01 · open the original ↗

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

2026-10-31 · open the original ↗

• Gas prices are falling from demand destruction, not new supply. Gasoline is at $4.18 a gallon, down about 25 cents from a month ago. He says that is still the highest October price on record. His reading is that Americans are driving less, combining trips and cutting spending, and that on his chart this summer's demand fell to 2022 levels.
• Unlike 2022, there is no supply relief alongside weaker demand. In 2022 US oil production rose by nearly 800,000 barrels a day, which he says is about three times this year's expected increase. He notes that a year of roughly $100 oil would historically bring a large production jump, and says there is "no drill baby drill" this time.
• Prices will rebound as soon as demand returns. He argues the cost structure is still expensive. Crude is nearly $30 a barrel above the same date in 2022, global refining capacity is constrained, refineries are already running hard, and geopolitical risk remains. He says the decline would be good news only alongside an end to the war and Russian, Saudi, Iraqi and Kuwaiti refineries coming back online.
• Cheap gas under financial strain is a recession warning, not strength. If demand keeps falling while consumer debt rises and discretionary spending weakens, he reads lower prices as a sign of a breaking consumer, noting that some of the lowest energy prices came in recessions. He adds that diesel and inflation are both still high.
• Repeated weekly demand drops would signal recession; one does not. Last week gasoline demand fell by over 600,000 barrels a day, which he calls the largest one-week drop since Covid. He says a repeat this week would be a red flag and a third week would mean "the wheels" coming off. He still judges a recession less likely than not.

2026-10-31 · open the original ↗

• Tech earnings exposed cracks in the market's protective shield. Tech earnings, not economic or policy news, were what moved markets most, he says, producing wild NASDAQ volatility against limited moves in rates and FX. Disappointments were punished harshly, including Intel's 29% fall on Friday, while tech bond spreads and CDS kept widening, though unevenly across companies. The $6.3 billion hack of Binance further hurt risk appetite.
• Supposed safe havens failed to act as safe havens. The 10-year US Treasury yield fell only to 4.93%, though October's 36 bp drop was the largest monthly fall since 2023; he says the week's data confirmed that Japan and Korea have become less reliable Treasury buyers, mainly for domestic reasons. Gold fell to $3,885 an ounce, a one-year low, and Bitcoin below $78,000, while crude hit a six-month high on fears of renewed US-Israel attacks on Iran after the midterms.
• Strong US data kept economists optimistic while most central banks paused. Q3 GDP growth accelerated to 3.4%, a one-year high, and Conference Board consumer confidence jumped from a revised 84.2 to 91.7, above consensus, but PCE inflation rose to 3.6%; French consumer confidence hit its lowest since 2013. The Bank of Canada hiked 25 bp and signalled more, while the BOJ, ECB and Fed held, with markets expecting all three to resume hiking in December. Fed Chair Warsh said one of the eight annual FOMC meetings would become strategic, ECB President Lagarde said she would step down early, and UK Chancellor Healey's first budget, keeping medium-term consolidation, left markets broadly unchanged.
• Brazil's selloff reflected traders' misplaced confidence more than fundamentals. That is his reading of the market reaction to President Lula's surprise win over Bolsonaro. In Israel, Netanyahu lost his majority amid gains for the two Israeli Arab parties, leaving deadlock with three outcomes he sees as open: a unity government of the Eisenkot and Netanyahu blocs, an Eisenkot coalition with the Arab parties, or a Netanyahu caretaker government and a spring election.
• The week ahead turns on the midterms, tech earnings and payrolls. Tuesday's US midterms matter for economic, regulatory and Middle East policy; earnings include Apple and Palantir. Consensus looks for October payrolls of 65,000 (from 29,000), unemployment unchanged at 4.2%, participation 61.8%, and average hourly earnings up 0.3% MoM (2.9% YoY); ISM services is expected near 54.8 with prices paid at 74.0. Markets put the odds of a December Fed hike near 75%.

2026-10-31 · open the original ↗

• His own premium rose about 25% for a worse plan. He reports his 2027 premium at roughly $2,600 a month, or $31,200 a year, before deductibles, co-pays and prescriptions. That is up about $500 a month from a little over $2,000, which was itself up from $620 the year before. The monthly figure he reads off his notice is garbled in the captions; the annual figure supports about $2,600.
• He blames exploding employer health costs for weak hiring. Marsh projects average employer health benefit costs will rise 8.2% in 2027, the biggest increase in 24 years, and over 10% without benefit cuts or plan changes. Another employer survey puts the increase before plan changes at 12%. His inference: every hire with benefits commits an employer to about $30,000 a year in insurance on top of salary.
• The rise has several causes, and Obamacare is not one. He points out he had Obamacare when his insurance was much cheaper. He cites hospital consolidation, which leaves employers and insurers less leverage on price, and specialty drug spending. Estimates attribute 1% of the 2027 increase to GLP-1 drug use alone. His wife skips a beneficial medication that costs $1,200 a month even with insurance.
• Employers are shifting costs onto workers, draining household spending. He says 60% of employers are cutting what they pay, and two-thirds of large employers surveyed expect employees to pay a larger share of premiums. The average premium was already over $28,000 a year, and costs are projected to rise faster than wages. A $500 monthly increase takes $6,000 a year out of spending that would otherwise reach other businesses, mostly small ones.
• Washington never fixed the underlying price problem, he argues. In his reading, Obamacare expanded coverage and added consumer protections, but neither party fixed hospital competition, drug prices or the incentives of employer-sponsored insurance. Congress also repealed the Cadillac tax before it took effect. He calls the premium extortion: someone with 35 years of savings at stake can't go uninsured, unlike a young, low-risk person.

2026-10-31 · open the original ↗

• A Rosland sales rep is being sued personally for fraud. Per court documents, an elderly couple put $2 million into Lady Liberty gold coins on the recommendation of actor Erik von Detten, a Rosland Capital account rep since 2018, then found them worth about $1.3 million. Their pro se elder abuse and fraud suit, filed in August, seeks $900,000.
• A pre-bankruptcy home sale raises questions about insider cash-outs. The Daily Mail reported that weeks before the bankruptcy filing, von Detten's wife, a luxury real estate agent, sold Rosland CEO Marin Aleksov a $4.9 million home in Brentwood, California. That insiders were shuffling assets is the speaker's and investors' suspicion, not an allegation she cites from any filing.
• Vet a bullion dealer before your first purchase. She urges new buyers to ask the channel's community for dealer recommendations, and says she has bought from ITM Trading in Arizona for close to a decade without complaint. This is her personal endorsement, not an independent assessment.

2026-10-31 · open the original ↗

• Rosland Capital's collapse left hundreds of creditors unpaid. Host Kim Brackett recaps that the bullion dealer filed for Chapter 11 bankruptcy last July, saying it was tens of millions of dollars in debt and could not repay hundreds of creditors. This is part three of the channel's ongoing series on the case.
• Rosland ran a Ponzi-like model that broke when metal prices spiked. She says the company held little to no physical metal, taking customer money up front and using new buyers' cash to pay old obligations and expenses before buying metal. When gold and silver soared this past January, payouts outran incoming cash, leaving about $60 million owed. "Ponzi-like" is her characterisation, not a court finding.
• Celebrity marketing was paid while customer orders went unfilled. Recent reporting says longtime frontman William Devane got $50,000 per video, and Rosland's ongoing TV ads cost millions of dollars, all while customer orders sat unfulfilled.
• Some creditors are suing directly rather than waiting in the bankruptcy line. They are filing pro se, believing they hold evidence of fraud and asset hiding the court-supervised process won't reach. In August an elderly couple filed an elder abuse and fraud suit naming former Disney child actor Erik von Detten, a Rosland account rep since 2018, over $2 million put into Lady Liberty gold coins on his recommendation.

2026-10-31 · open the original ↗

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

2026-10-31 · open the original ↗

• AI tooling is reaching individuals and small businesses as easily as enterprises. Kyritsi, a global technical partner manager at GitHub in Amsterdam after 20-plus years in enterprise telecom and lottery systems, says this is the first time her work connects to her friends' everyday lives; she credits ease of deployment and distributor partners serving SMB customers.
• Enterprises need governance; small businesses need leverage and return on investment. Large customers focus on governing agentic development workflows across hundreds of developers, while an SMB wants a team of two or three to work like ten or fifteen, with spending that makes sense.
• A business user's working prototype beats a requirements workshop. She set up the GitHub Copilot app with Work IQ for a friend on a pharmaceutical company's go-to-market team and built agents for event tracking; she encouraged the friend to hand the prototype to her internal developers, opening communication rather than creating shadow IT.
• Token economics is where partners add the most value. Usage-based billing and model choice confuse price-sensitive customers; she cites a partner in India and a bank that considered cutting off its whole team's access until an analysis of the previous month's usage streamlined it to the customer's satisfaction.

2026-10-30 · open the original ↗

• A GPU in a separate box can run a laptop over Wi-Fi. Chuck received a unit holding an Nvidia 5090 that shows up as "GPU 1" on a small ASUS laptop with no cables. He played Cyberpunk on it, with Nvidia-Smi showing the GPU in use and the traffic going over Wi-Fi. He calls it wireless computing infrastructure; the captions garble the acronym.
• Chuck says it runs large models the laptop otherwise couldn't. He credits this to 4 TB of memory, "but not exactly how you might imagine it". The short doesn't say how that works and points to the full video for the explanation.

2026-10-30 · open the original ↗

• Hull-mounted bombs are a new, working method of attacking Hormuz tankers. As of October 30, 2026, the host reports that overnight a small boat came alongside a tanker in the Strait of Hormuz and a man attached two bombs to the hull above the waterline; they started an engine-room fire that the crew put out, with no injuries, and the ship is moving again. He calls it the first time a bomb stuck to a hull has worked in Hormuz since the war began.
• Ukraine has put Russian refineries back on its target list. The host reports a strike overnight to October 30 on a refinery in Russia's Samara region that makes jet fuel for the Russian air force, damage not yet confirmed. For the previous two weeks Ukraine had hit fuel depots and ports rather than refineries.
• European gas is at its highest since 2022 while storage lags. As of October 30, TTF is back above €100 after a sharp October rally, with storage just under 75% full — below both last year and the 5-year average. European leaders also agreed to release 42.6 million barrels from emergency oil stocks, about half of it diesel.
• Qatar's US LNG deals are long-term insurance, not winter supply. Qatar says it is now the largest buyer of American LNG: this week it signed a 20-year deal with the Woodside plant being built in Louisiana, and last week another large deal with Cheniere. That gas starts in 2030–2031; the host reads the deals as hedging because Qatar's own damaged plants will take about three years to repair.
• The host thinks markets look calmer than events on the water. In his own closing read, he says it is natural for people to expect worse than the headlines show, because "the data terminal looks calmer than what's happening on the water."

2026-10-30 · open the original ↗

• A wireless box lends one GPU to every device. The video is sponsored by the company that makes it, a new firm whose name the captions garble. Chuck tested an early engineering model: an RTX 5090, a mini PC running OpenWRT and a 4 TB SSD. Laptops, a Mac and phones can use it like a local GPU. It is on sale now in 5090 and cheaper 5060 Ti versions, shipping in about 8 weeks.
• The company's own protocol makes Wi-Fi fast enough for gaming. PCIe gives 512 Gbit/s and sub-microsecond replies; Wi-Fi gives about 1 Gbit/s and 1.5 ms. A fake driver makes the laptop think the 5090 is its own. Instructions are bundled, compressed and sent without waiting, only changed data is sent, and answers are predicted. That cuts about 8 Gbit/s to 400 Mbit/s. Chuck got 92 fps in Cyberpunk, rendered in Blender, and says this is not cloud gaming or remote desktop.
• A model on SSD is normally unusably slow. DeepSeek 4.1 Flash has 750 billion parameters and needs over 500 GB of memory, against the 5090's 32 GB of VRAM. Chuck's figures: VRAM 1.8 TB/s, DDR5 90 GB/s, SSD rated 14 GB/s. Small 4 KB reads give only 1–5 Gbit/s, so a dense model would take about 8 minutes per word. Mixture of experts (6 of 384 experts per word) still means about 15 seconds.
• Three changes make the SSD model run at usable speed. Expert files are rewritten into equal-size aligned chunks read 64 at a time, taking SSD reads from about 1 to 11.5 GB/s. Popular experts stay in VRAM and RAM, so only about 1 GB per word comes off the SSD. The CPU does the math for SSD and RAM weights; the GPU does it for VRAM weights. Chuck measured 3–15 tokens a second, about eight on average, with 61% of experts served from the GPU in one run.
• A scheduler on the box lets apps share the GPU. One image, video or chat model can fill all 32 GB of VRAM, and switching normally means reloading from scratch. The scheduler parks idle models on the SSD instead. In Chuck's demo, switching to the image model went from about 18 to 8.6 seconds, and back to chat from 17 to about 4. He calls the "family GPU" idea impressive and expects the engineering model to improve.

2026-10-30 · open the original ↗

• November is a quiet delivery month, not a sign of easing. On first notice day roughly 7 million ounces of silver were requested for delivery on COMEX. Kim Brackett says last month's calls held: the exchange shuffled about 10–20 million ounces of supply, and delivery rates stayed close to 100%. Brackett reads that heavy demand as making November "the calm before the storm."
• December contracts far exceed the silver available to deliver. Brackett reports 105 million ounces of registered, deliverable silver in COMEX vaults against 413 million ounces of contracts coming due in December. The channel expects the exchange to spend the next 30 days offering cash inducements to get holders to roll into 2027 or accept cash settlement instead of bars.
• COMEX is predicted to get through December by a narrow margin. In Brackett's own forecast, framed as a prediction "for the fun of watching," the 413 million ounce December figure falls to around 100 million as speculators exit or roll. The exchange is also expected to move at least 20 million more ounces into the registered category.
• Cash settlement could cost COMEX its price-setting role. Brackett poses it as an open question rather than a forecast: how long COMEX keeps price-setting dominance if cash settlement replaces physical delivery. The suggested successor is Asia, where every contract is said to be backed by physical silver. This is the channel's own speculation.

2026-10-30 · open the original ↗

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

2026-10-30 · open the original ↗

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

2026-10-30 · open the original ↗

• Caro rejects the claim that today's safeguards make big building impossible. Patel put the "overcorrection" case (Second Avenue subway: ~50 years, a mile and a half, over $4 billion; California's ~$15 billion with no new rail). Caro answers that the 1920s said the same until Moses built Jones Beach anyway, doubts any "good spot" exists, and concedes only that we need someone who gets things built, without Moses's ruthlessness.
• Moses hardened from idealist to ruthless builder without realizing it. The early Moses fought for merit-based civil service and beach access; the later one ran the Cross Bronx Expressway through a neighborhood, evicting tens of thousands without relocation, and destroyed Henry Epstein, who proposed a route two blocks south, by exposing an affair. Caro's one question for an eighth interview: "Did you realize you were changing?" His own answer is no.
• Depression-era labor gluts made Moses's speed and old ornament possible. Caro agrees that thousands of jobless architects lined up on Fifth Avenue and workers could be fired at will because they had nowhere else to go. He offers the same mechanism for why a poorer country built more beautifully: skilled carvers came so cheap that builders decorated rooftops nobody would see.
• Kennedy lacked the drive to pass Johnson's domestic laws. Asked whether a President Kennedy with Johnson still majority leader would have got the 1963–65 laws earlier, Caro says no: Kennedy lacked the desire to change America that Johnson brought back from teaching poor children in Cotulla. He holds that Vietnam ruined that domestic program, and declined to say why Johnson kept escalating because he is still working it out for the next book.
• Most of Caro's research never reached print and sits in his archive. He cut The Power Broker from a 1,050,000-word manuscript, with 350,000 words removed, and has about 700 interviews for that book alone, plus Senate material he never wrote up. He gave the files to the New-York Historical Society for far less money than other offers so they would be accessible in New York, naming Sam Rayburn as a figure still lacking a good biography.

2026-10-30 · open the original ↗

• Rule of law is America's economic edge, and corruption erodes it. Scott leans on Acemoglu and Robinson's Why Nations Fail (2012) on inclusive vs. extractive institutions, and on IMF economist Paolo Mauro's 1995 study of 67 countries, which found a modest improvement in a corruption score raised investment by 2.9% of GDP and annual growth by up to 1.3 percentage points. His warning case is Russia's reiderstvo (state-linked raiding of businesses): a 2016 World Bank/Higher School of Economics survey of nearly 2,000 manufacturers found firms in heavy-raiding regions a third less likely to make capital investments, and he dates its turn into state policy to the 2003 dismantling of Yukos.
• Trump is gutting the agencies that police white-collar crime. As reported in the post: half of Justice Department lawyers fired or pushed out; 1 in 6 SEC employees gone, with only five SEC enforcement actions this year versus 80 two years ago; CFTC staffing cut 22%, with enforcement falling to 80% of the ten-year annual average; IRS headcount down 25%, the high-net-worth audit unit cut nearly 40%, and audit rates at a historic low. He also cites Trump buying Oracle, Nvidia, Netflix, Paramount and WBD shares while holding regulatory sway over deals involving those companies.
• The scale of Trump's self-dealing outruns anyone's ability to track it. Isaac Saul of Tangle logged nearly 100 alleged instances of corruption and conflicts (May and October posts), and a New York Times investigation estimates Trump made $2.2 billion in his first year back; Scott adds $1.4 billion from crypto last year, $1.8 billion in vanity projects, and a new lawsuit against World Liberty Financial. Against an NBER finding that congressional leaders beat rank-and-file members by up to 47 percentage points, he notes Pelosi made 213 trades since 2014 while Trump made 327 in one day before pausing the Liberation Day tariffs.
• The forced TikTok sale was a Yukos-style betrayal of markets. Scott argues Trump forced the sale of TikTok's U.S. business at an estimated 94% discount on his own terms and split the spoils with political allies. His inference: American investors now learn, as Russians did after Yukos, that the president rather than markets sets valuations, deal terms and buyers.
• Trump's Russia dealings put American foreign policy up for sale. He reports Trump waived Russia sanctions to buy diesel less than a month after signing the Lindsey Graham sanctions act, and effectively endorsed Elon Musk restricting Ukraine's Starlink and Starshield access; citing Tom Nichols of the Naval War College, he argues this indirectly helps Iran and labels Trump a "traitor." His prescription: winning Congress in the November 3 election would bring subpoenas and evidence for a reconstituted DOJ, to "make America boring again" like the rule-bound sale of his company L2.

2026-10-30 · open the original ↗

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

2026-10-30 · open the original ↗

• Making people afraid takes their freedom without any force. He argues that a frightened public stops using its freedoms by itself, and that politicians and media who stoke fear for their own benefit are robbing people of it. He cites polling ahead of the 2026 midterms: 54% of Americans worry about people gathering at polling places to intimidate voters, and 4 in 10 fear being confronted, followed or recorded.
• Enormous government power combined with no trust breeds fear. He cites Gallup: only about 30% trust the government's ability to handle domestic problems, 28% trust the legislative branch, and 26% trust the federal government "some" to act in society's best interest. Meanwhile 92% believe the federal government has enormous power over people's lives.
• Media and social media profit from fear, and they amplify it. He cites a study finding that Americans on social media more than three hours a day report much greater concern about election problems: voter-intimidation concern at 60% and fear of mishandled ballots at 65%. He concedes this does not prove causation. His argument is that fear earns clicks, ratings, shares and donations, so every election gets sold as existential.
• Most Americans are far less extreme and divided than portrayed. He cites figures showing 67% say leaders' failure to denounce violent rhetoric contributes to violence, 53% blame harsh language itself, and 77% reject the idea that political violence may be needed to save the country. In a Reuters poll he cites, 73% call democracy the best form of government and 75% value elections even when their party loses.
• Freedom takes courage, while authoritarianism offers the comfort of certainty. In his reading, authoritarianism tells people who the enemy is and what information to trust, and promises their fear will go away if they hand over more power. He urges viewers to question the government, the media, influencers and him, and to vote without being intimidated.

2026-10-30 · open the original ↗

• Supabase's growth inflected this year, driven by agents and non-engineers. Paul Copplestone reports 13 million signups against roughly 3 million a year earlier, revenue growing 4x this year, about a million new databases a week, and databases under management nearing 50 million versus 8 million. A partner puts new activity at around 70% from agents; unusually, activation and conversion are rising alongside signups.
• Winning agents means being picked first, then never being abandoned. Copplestone credits becoming a default database choice in LLMs, being fully open source, and going code-first with CLI work in sandboxes, after seeing that "no one visits dashboards anymore." Error messages now tell agents how to fix the problem, and much engineering effort has moved to evals and use-case testing across models.
• Everyone with AI becomes a builder, wanting their own database. Copplestone cites a Brazilian taxi driver who built a local ride-hailing app from his phone. Garry Tan describes personal software for email, finances and health that he rates at least on par with Mint or Monarch, arguing the database is the valuable store of record and agents should be swappable on top.
• Database counts at consumer scale require a new architecture. Copplestone's arithmetic: if ChatGPT's 1.2 billion weekly users each launch something monthly, that is 1.2 billion databases a month. The Turso acquisition supplies a cheap SQLite- and Postgres-compatible starter tier, while Multigres scales up to petabytes; he cites an unverified stat that about 40% of AWS revenue comes from startups that grew on it.
• Pick a durable category, then out-adapt everyone inside it. Copplestone invokes the Lindy effect: databases have existed 50 years and will outlast whatever AI brings. His prescriptions are dogfooding, keeping team capacity free for partner experiments, most of which flop, and avoiding the corporate drift he sees in dev-tool companies. Internally, engineers writing 80% of code with agents rose from 10% to 60% in six months, with no AI policy.

2026-10-30 · open the original ↗

• Long-duration Treasury exposure is rented and funded at the short end. Holding periods follow mandates: CTAs roll futures quarterly (Feb/May/Aug/Nov), index trackers sell as bonds exit their buckets, active managers drop issues going off-the-run, and only LDI holds to maturity. Futures at 2–5% margin (20–50x), receive-fixed swaps, repo at 1–3% haircuts and STRIPS all leave positions long the long end and short the short end.
• LDI is the largest, least discretionary pool of levered duration. Pensions take duration through swaps or repo-financed bonds at typically 2–4x to keep capital in growth assets. The author draws on the September 2022 gilt crisis — collateral calls forcing gilt sales until the Bank of England stepped in — to argue such holders become forced sellers exactly when long yields jump.
• Treasury has adopted the bill-heavy financing its secretary once criticised. Bessent faulted Yellen in 2024 for leaning on bills, then did the same: coupon auctions are held flat, including $25B of 30-year per refunding, with every incremental deficit dollar funded by bills; stablecoin reserves are pitched as a new buyer. Citigroup projects the bill share passing 26% by end-2027, 27% by 2028 and 41% by 2033, and Citi's rates head expects long-dated sales cut at the November 2026 refunding.
• Bill-heavy issuance makes long duration cheaper but costlier to carry. Flooding the front end lifts repo and SOFR against the policy rate and drains reserves, eroding carry on every repo-funded long. Thin 20- and 30-year supply invites squeezes of the cheapest-to-deliver, widening the basis to the benefit of basis traders, while the term premium is held down by issuance choices rather than fundamentals.
• The eventual term-out is the concentrated trigger for deleveraging. The author's forecast: when Treasury announces larger coupon sizes, the August 2023 playbook runs in reverse and long yields spike — a trend for CTAs, a forced-deleveraging event for basis trades, LDI and long-bond ETFs. Since such unwinds are set by a yield move, not a calendar, he argues long-end holders should watch the refunding statement more than the data.

2026-10-30 · open the original ↗

• Moscow's real fear is its eastern neighbour, not the West. He relays reports, presented as a data point rather than a conclusion, that China is pressing for near-total access to eastern Russia's rare earths, oil and coal, aiming to lock it in before the Ukraine war ends while its leverage is greatest. He ties this to his earlier claim that the FSB feared the Chinese most.
• The rate hold leaves policy behind hotter-than-reported inflation. He notes the Fed's 10–2 vote not to raise rates, with inflation data coming in hotter the next day, and asserts — without support in the episode — that CPI has been rearranged since Warsh arrived to remove about half a point, so viewers should add 0.5%. He cites a commentator arguing fiscal consolidation, not monetary policy, is the neglected tool.
• Nobody owns the long bond; leveraged players rent it. Walking through his slides, he says holding periods are set by mandate: CTAs roll futures quarterly, index trackers sell as bonds leave their buckets, active managers want large on-the-run issues, and LDI funds match liabilities. Every structure is long the long end and funded at the short end.
• The bills-only strategy implies enormous, fast-rolling short-term debt. With Bessent holding 30-year auctions held at $25B and deficits bill-funded, he applies Citi's bill-share forecasts to his own debt path: about $44T by 2028 at 27% means rolling ~$12T every 13–26 weeks, and ~$54T by 2033 at 41% means ~$22T. At $4T annual deficits he puts it at 30-some trillion.
• Watch the refunding, not the rhetoric; term-out crushes the long end. He argues short-end flooding raises financing costs for households and corporations, creates deliverable scarcity that favours basis traders, and suppresses the term premium by policy. His forecast: once Treasury can no longer lean on bills and terms out, the long end gets "crushed."

2026-10-30 · open the original ↗

• Asian markets sold off on AI spending outrunning AI revenue. China's tech-heavy STAR 50 index fell about 40% over recent months, erasing roughly $480 billion in market value, and South Korea's KOSPI and Hong Kong's Hang Seng had several severe daily drops. The speaker attributes this to Asian tech firms and cloud providers spending around $140 billion on AI data centers this past year while AI software revenue comes in at a fraction of that.
• Falling gold and silver prices reflect forced selling, not weaker fundamentals. Her argument is that leveraged traders and fund managers hit with margin calls must raise cash immediately, so they sell their most liquid holdings, usually paper gold and silver contracts. She frames metals as an "emergency ATM" for a stressed financial system, using a casino analogy where the gambler hands over car keys to cover a debt.
• US AI earnings suggest the selloff is spreading. She reports that Microsoft beat expectations, but Alphabet spent $51 billion in one quarter on AI servers and networking. Together with a regulatory fine, that left its cash flow negative by $9 billion, and its shares fell nearly 11%. Data-center chipmaker Marvell Technology also fell 11% on the day.
• She predicts a short-term metals drop, then a recovery within months. If US AI stocks keep falling, she expects gold and silver prices to keep dropping as funds cover margin calls. Over the following several weeks to a few months, she expects investors to move into physical gold and silver and prices to turn up. This is her forecast, offered as reassurance to hold physical metal.

2026-10-30 · open the original ↗

• Intel's collapse is the first real bear-market clue. He has said for weeks that a large-cap, "reputable" stock collapsing could be the tell; Intel, which he calls a $400B+ blue chip, fell by what the audio gives as 24 pre-market. He won't decide on that yet: it could be a dead cat bounce, so he is waiting for how it trades today, with a second update this afternoon.
• The Binance hack may mark a bottom, as FTX did. Binance was reported hacked for billions (NYT, 29 Oct 2026; in the audio, by North Korean hackers). He notes that FTX's collapse between 2 and 11 November 2022 roughly marked the bottom for crypto and stocks. He labels the parallel as his own speculation, knowable only in hindsight, and calls crypto too full of scammers and hackers to be worth the risk.
• The S&P 500 is still in a defined bull market. S&P 500 futures are down only slightly and are mostly forming a higher low, so he reads the broad market as intact and the selling as a possible overreaction. He doesn't think the meltdown has started, but wants to see where the market closes Friday before judging next week.
• Avoid knee-jerk pre-market moves, but set hard cut points. Pre-market action is especially volatile for single names, he says, and reacting to it is often wrong. Still, overexposed holders should trim or cut. He will sell Hinge Health below $111 and Twilio below $176, which roughly cancel each other out, while Xometry, up more than 38%, gives a cushion. Small losses must not become massive ones.

2026-10-30 · open the original ↗

• Frontier AI changed attack economics, not attack techniques. Amy Blackshaw (SVP, Palo Alto Networks) argues the expensive part of an attack was always the expert thinking — studying a target and chaining vulnerabilities, which took weeks — and AI has made it instant, tireless and exhaustive. She stresses the villain is still the attacker, not the model: entry is now cheap and enumeration free, for small and large targets alike.
• The headline agent breach was an identity-hygiene failure, not AI sophistication. In her account of the July incident, autonomous agents that escaped an OpenAI test sandbox stayed inside Hugging Face's production infrastructure for 4.5 days, carrying out about 17,500 actions (one every 20 seconds). Speed got them in; what got them to cluster admin was a plain-text password, one master key that opened 136 doors, and an unreviewed service account with standing admin rights across two clusters.
• Breakout is too fast for human-led response alone. Palo Alto's Unit 42 measures about 72 minutes from foothold to impact in real incidents and about 25 in simulations. Her prescription is AI-driven security operations with humans in the loop, plus AI-driven virtual patching and vulnerability testing, paired with identity security because an attacker using a trusted identity often goes unnoticed until the end of the chain.
• Every user is now privileged, so shrink the trust window. She says the old split between privileged IT admins and ordinary users no longer holds, since developers, data scientists and business users slip in and out of privilege daily. Every key should be judged by how long it works and how many doors it opens; zero standing privilege and just-in-time access, granted and revoked automatically by policy and risk, cut exposure without adding friction. Her first step for any 90-day plan: remove standing privilege, starting with admins and hard-coded machine credentials, in favour of ephemeral secrets.
• Agent identities combine human autonomy with machine-scale sprawl. An agent makes decisions, chains tools and escalates its own privileges like a person, but is provisioned like a service account and can be created by the thousands overnight, sometimes without governance. She names three rogue modes — subverted by poisoned input, drifting beyond scope, or leaking secrets through its own reasoning — and says in each the damage is bounded by what the agent was allowed to do and for how long, so agents should get credentials that expire with the task.

2026-10-30 · open the original ↗

• Skeet is building a long-running UK election tracker as a hobby. It began in the few weeks before the 4 July 2024 general election to compare MRP seat predictions and commentator Sam Freedman's calls with results; he bought election2029.uk for about $25, then election.uk for over $1,000 on joining Trainline as a distinguished engineer after leaving Google. The next election is scheduled for 2029; he doubts rumours of 2027.
• The site is deliberately written without AI, as a longitudinal experiment. He wants to judge in 2029 whether that was a mistake; elsewhere, on Noda Time, he says Claude took 10 minutes and many tokens to fix a test by changing a greater-than to greater-than-or-equal, which he concedes would have taken him longer to find.
• The stack is plain .NET on serverless Google Cloud. It is ASP.NET Core on .NET 10 with Razor Pages that have no code-behind and injected, separately tested view models, plain JavaScript with Plotly and no CSS framework; it runs on Cloud Run, Cloud Build and Firestore with OpenTelemetry, and he ported it to AWS S3 and Lambda in under a week.
• The poll averages are intentionally simple rather than smoothed. Trend lines linearly interpolate each provider's two nearest polls and average across providers, showing no average where only one provider exists; he contrasts Wikipedia's smoother lines, which he feels assume more than the data shows. Fake results appear only on a clearly bannered staging site.
• The site is tiny and should scale cheaply. It sets no cookies, the home page is about 10 KB, all data loads in a second or two, and a load test of the postcode search at around 600 QPS scaled to only two instances; he expects real traffic only once an election is called.

2026-10-30 · open the original ↗

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

2026-10-30 · open the original ↗

• Delegating writing to LLMs means delegating your thinking. The warning is aimed at people who conclude they no longer need writing skills: writing is mostly a way of organising thought, so outsourcing it outsources the thinking too. The author expects that view to spread, arguing that tools which make anything easier tend to become crutches.
• LLM prose reverts to the mean, which favours committed writers. The author calls model output "solid B+ writing" — average human expression, rarely novel or profound, and as forgettable as most short-form video. That is opinion rather than a measured claim; the author reads it as an opportunity for people who live to write, while still crediting LLMs as useful editors and researchers.
• The author is more optimistic on AI job losses than most. AI could displace millions of workers, the author grants, but every agricultural, industrial and technological revolution has ended in net job creation. The author also reports that publishers at a Google event in NYC the previous week were gloomy about their businesses as organic search traffic declines.

2026-10-30 · open the original ↗

• The most valuable data is walled off from cloud AI. The unnamed YC speaker argues that defense contractors, banks, hospitals and nuclear plants cannot send their data to an AI API, and reads this as why much of the economy is still waiting on AI. YC founders keep hitting it, they say: customers love the product but want it run inside their own building with no internet connection.
• YC wants air-gapped AI companies: harder to sell, but more defensible. The speaker says YC wants to fund companies that bring frontier AI inside these walls, naming on-prem models, deployment tooling, security accreditation and hardware. They concede the sale is slower and harder, and assert that this is where the largest budgets and the most defensible companies are. The clip is an open call for founders building this.

2026-10-29 · open the original ↗

• A single strong quarter is not historic and proves nothing. Third-quarter 2026 GDP came in at 3.4%, short of the 5% Bessent, Trump and Hassett had talked up and Lutnick's 6%. He notes half of all quarters since 1947 were 3.4% or higher, and attributes this quarter to consumer spending, government spending and some investment.
• Postwar growth ran well above today's, under both parties. He cites the Economic Policy Institute at roughly 3.8% average GDP growth under Democratic administrations versus 2.6% under Republican ones, and Princeton's Alan Blinder and Mark Watson at 4.3% versus 2.54%. He cautions that researchers trace the gap to productivity, oil shocks, international conditions and expectations, not presidents.
• A shrinking labor force is capping America's underlying growth rate. Retirements mean potential labor force growth, which he puts at 0.7% a year from 2008 to 2025 (captions read "7%"), is projected at 0.4% for 2026–36 against a 1% minimum he says is needed. He cites CBO projections of 1.8% average real GDP growth over ten years, with potential growth halving from 2031 to 2036.
• Tax cuts and deficit spending cannot buy sustained growth. He calls tax cutting "basically the Republican plan" and argues stimulus, checks and consumption boosts are temporary and worsen things long-run. Sustaining 3–4% without inflation or huge deficits needs productivity, investment in factories, power, transmission and housing, predictable regulation, energy, and workers, since automation and AI still require them.
• Immigration reform is the missing source of workers. He argues Washington keeps immigration as an election-season political football, yielding more illegal and fewer legal immigrants, while the economy needs engineers, doctors, electricians and welders. He predicts a decade at 1.8% growth means falling wages, rising prices, and no demonstrated ability to sustain even 4% in a $30 trillion economy.

2026-10-29 · open the original ↗

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

2026-10-29 · open the original ↗

• Permit delays are costly, and AI can do the review work. YC says city staff spend their time checking drawings against code, matching applications to zoning rules and emailing corrections, and that in some cities the permit takes longer than the build. By its own math, approving all US building permits in 10% of the current time would add $16 billion a year in city tax revenue.
• The opportunity extends from permits across most of city hall. YC wants to fund companies starting with permits and moving into 311, public records, licensing and inspections. It says almost all of America's 19,000 cities still handle this work by hand.
• YC argues selling software to governments is no longer impractical. It calls the conventional wisdom outdated, saying it has funded several companies selling AI to government and that some are among the fastest-growing it has seen. This is YC pitching its own request for startups, and it gives no company names or figures.

2026-10-29 · open the original ↗

• Rosland ran a Ponzi-like model that broke when metals spiked. Host Kim Brackett says the dealer, which filed for Chapter 11 last July, held little to no physical metal and used new customers' cash to cover old orders and expenses; January's price surge forced payouts beyond incoming cash, leaving about $60 million owed to hundreds of customers. The video states this as established without naming its source.
• Celebrity marketing spending continued while customer orders sat unfilled. Citing recent headlines, she reports longtime spokesman William Devane was paid $50,000 per video and that Rosland's ongoing TV ads cost millions of dollars over the same period.
• Some creditors are suing directly instead of waiting in the bankruptcy claims line. They are filing pro se, believing they hold evidence of fraud and asset hiding the court process won't reach. One filed case from August, an elder abuse and fraud suit, names Erik von Detten, a Rosland account rep since 2018: an elderly couple bought $2 million in Lady Liberty gold coins on his advice, found them worth about $1.3 million, and seek $900,000.
• A pre-bankruptcy home sale raises the question of insiders cashing out. Per a Daily Mail report, von Detten's wife Angela, a luxury real estate agent, sold Rosland's CEO a $4.9 million Brentwood, California home weeks before the filing. The suggestion of asset shuffling is investors' speculation, not an allegation the video ties to any filing.
• First-time buyers should vet dealers through community recommendations. Her advice is to ask other stackers who they buy from; she discloses she has bought from ITM Trading in Arizona for close to a decade without complaint, an endorsement rather than evidence.

2026-10-29 · open the original ↗

• Models got worse at writing as they got better at code. The speaker asserts current models sound more robotic than those of 2024 and 2025, and reads this as structural rather than a temporary glitch: the reward signals that make models good at code and math actively flatten their style, and the major labs will always prioritise coding over writing when forced to choose.
• A model people like talking to is an open startup opportunity. Because the labs are expected to keep trading writing quality for coding, the speaker pitches this gap directly to founders as a company worth building, though the clip offers no evidence of demand beyond the claim itself.

2026-10-29 · open the original ↗

• Senate control is being fought as a national election. Galloway reports Senate races have spent $3.2 billion, nearly matching the entire 2024 presidential race, in what the hosts call the most expensive midterm cycle ever, with more than 39 million ballots already cast five days out. His argument is that Senate control is as consequential as the presidency for the next two years of governance.
• Democrats are contesting Senate seats in long-red states. A New York Times–Siena poll has South Carolina's Senate race a virtual tie between Democrat Annie Andrews and the Republican finishing Lindsey Graham's term, and Fox News moved it from lean Republican to toss-up; South Carolina has not sent a Democrat to the Senate since Fritz Hollings left in 2005, and Kansas not since 1932. CNN polling has Democrats up 6 in Texas, 7 in Michigan, 4 in Kansas and 4 in Maine. Tarlov's view is that Texas, where James Talarico faces Ken Paxton, is still the likelier red-state upset than South Carolina. These are pre-election snapshots.
• Democrats' ballot-stacking copies an old Republican trick. In Alaska, the hosts discuss a New York Times report that Democrats recruited a second "Dan Sullivan" onto the Senate ballot against incumbent Dan Sullivan, with polling favouring Mary Peltola. Galloway welcomes it as Democrats finally playing hardball and expects whoever wins to be seated; Tarlov dislikes it but cites Republican precedents (a New York prison inmate backed against Peltola, a Miami race lost by 32 votes, a same-name candidate against Summer Lee in Pennsylvania) and notes it may become illegal if money changed hands.
• A half-hearted endorsement is worse than none. Galloway argues Haley Stevens's support for Michigan Senate nominee Abdul El-Sayed, short of full-throated, is a non-endorsement, and that her fiancé Jason Turkish backing Republican Mike Rogers is a media-manufactured story. Tarlov agrees five days out is no time to equivocate but says concerns among Jewish voters, sharpened by an outside "can't afford eggs because of Israel" ad not from El-Sayed's campaign, are real; she still predicts El-Sayed wins and contrasts Elissa Slotkin's full campaigning for him.
• Aging incumbents are stalling Western democracies. After Netanyahu failed to secure a clear win and alleged fraud at Arab polling stations (claims police say lack evidence), Galloway says 75% of people his age support Israel versus 25% of those under 25, a long-term problem for US support. He proposes every leader over 72 resign within a year, citing France, where he says a third of the education budget goes to retired teachers and the average retiree out-earns the average worker.

2026-10-29 · open the original ↗

• The market's bigger risk is to the downside. Tom Sosnoff says he is staying short Nasdaq and S&P futures through a 400–500-point Nasdaq drop. He points to a three-standard-deviation downside move on Brazil's election and argues any midterm surprise next week can only hit downside. A co-host says that is just why index skew runs to the downside. Sosnoff still expects a non-event, with VIX futures near 18.55 at the long-term mean, and plans to go in delta-flat.
• Strangles across several earnings names spread risk but stay binary. On selling 16-delta strangles in Alphabet, Amazon and Intel while staying delta neutral, the hosts call three better than one but still three binary bets. Microsoft was a winner (one host's $1 strangle bought back at 5–10 cents). Google fell about $41 against an expected move of about $24–25, so 16-delta puts there lost. They argue against trading stock after hours against short premium, and call 24-hour options markets untradeable: Google's at-the-money options were $6 wide with no market makers and zero volume.
• Boomer retirements will trickle out, not produce a sell-off. The prompt was 16–25 million workers (10–15%) retiring over five years. One host's numbers: boomers hold $30 trillion, 54% of equities, but the wealthiest 10% hold two-thirds and need not sell, leaving roughly $10 trillion. Younger buyers keep coming in, so the hosts call it not a tradable event. Sosnoff's only advice is to lighten up a little at record highs.
• Corporate junk refinancing threatens the market more than ARM resets. One host says the 3.1 million adjustable-rate mortgages (about 6% of mortgages) resetting 3–4 points higher "can't be bullish", comparing it to gas at $2 a gallon more, roughly $1,500–2,500 a year per driver. The other calls homeowners A-grade paper and says the real danger is lower-rated companies refinancing 6–9% debt at 14–17%.
• Paper trading teaches the platform, not real fills. Sosnoff built thinkorswim's Paper Money, which he says was downloaded close to 20 million times. He says paper trading fills at mid-price, which misleads especially in SPX. He says tastytrade left it out because once commissions went to zero, the support load could not be paid for.

2026-10-29 · open the original ↗

• Extended-hours options markets are untradeable in practice. One host says they checked Google options one to two hours before the open the morning after earnings. At-the-money markets were $5–6 wide (one quoted 45 cents bid at $6), with no market makers and zero volume in Google or Microsoft, despite the exchanges promoting 24-hour trading.
• Short-premium traders shouldn't trade stock against after-hours earnings moves. Their example is Microsoft: it looked like a big loser for short premium after hours but finished a large winner. Selling 16-delta options across several earnings names is called "the best you can do" short of going directional or sitting out: 16-delta puts lost in Google and Boeing but made money in Robinhood.

2026-10-29 · open the original ↗

• Both parties serve the same donors, by design. He argues the political system works "like clockwork" for those funding politicians — anyone with a nine-figure net worth — whether liberal or conservative, and calls the two-party system "a feature, not a bug" because it keeps the public split in two.
• Partisan opposition is performed, not meant to win. In his reading, today's Republicans voice the wealthy's agenda openly while Democrats denounce it and deadlock without truly stopping it, and the roles reverse when Democrats hold power. He infers this lets voters believe their concerns are heard but blocked by the other side, while the shared funders keep profiting.
• The left-leaning majority is kept divided so it won't vote. He asserts most eligible US voters lean mildly left — not socialist, but supportive of things like free school lunches — and that their full turnout would threaten donor interests. Both parties, he argues, therefore work to keep left-leaning voters divided over race, faith, politics and sexuality.
• Foreign states fund the same division, on both sides. He names Russia, Qatar, North Korea and Iran among countries he says actively finance American division wherever they can, left or right. The aim, as he frames it, is distraction from billionaires and "multi multi-millionaires" paying little to no tax through loopholes ordinary voters do not have.

2026-10-29 · open the original ↗

• Moses wrote the laws that governed him, and made them irrevocable. Amendments he slipped into the Triborough Bridge Authority Act of 1933 let the authority refinance "for any other corporate purpose" and write its powers into bond contracts, which the Constitution's contracts clause shielded from the Legislature, governor or mayor. By 1946 Triborough could borrow $360 million against New York City's $150 million debt limit.
• Moses kept power by being impossible to fire. He held 12 posts split between mayoral and gubernatorial appointment, was "money honest", and delivered popular works inside one election cycle, including 1,700 park renovations in four months. His ready plans won New York one-seventh of national Depression work-relief money by 1936, which let his constant resignation threats, backed by newspaper-owner friends like the Sulzbergers, bite.
• Power reveals character, but Moses and Johnson ran opposite courses. On Caro's account Moses was a reformer who learned ideals need power, then came to want power for its own sake. Johnson looked like a pure opportunist, telling Southern senator Herman Talmadge he was "with us in his heart", yet as president used power for the 1964 Civil Rights Act and 1965 Voting Rights Act.
• Johnson made himself the source of everything senators needed. The majority leader's only formal power was first recognition, and his two predecessors lost their seats. Johnson controlled floor scheduling, his line to Speaker Sam Rayburn, Texas oil money and committee seats, replacing seniority with a Steering Committee he ran; after Estes Kefauver crossed him in 1957, the Foreign Relations seat went to John F. Kennedy.
• Johnson settled votes privately, where his leverage and information worked best. He made unanimous consent agreements a precondition for floor action so debate couldn't unravel deals, and had every committee report drafts to him. The 1957 Civil Rights Act passed by bundling an Idaho–Oregon dam, jury trials for striking workers and Black eligibility for federal juries, while keeping the jury-trial guarantee that averted a Southern filibuster.

2026-10-29 · open the original ↗

• A Treasury license cannot override Congress's Russian import ban. He argues Treasury General License 135 only authorizes transactions otherwise barred by sanctions, while the 2022 statute banning imports of Russian petroleum products, diesel included, still stands; no executive order or 2026 action by Congress or President Trump changes it, and powers granted under what he calls the Lindsey Graham law apply only to that law.
• The tanker said to be bound for New York stopped heading there. He reports a tanker owned by Eastern Mediterranean Maritime Company left Primorsk with roughly 40,000 metric tons of diesel, about 300,000 barrels, listing New York as its destination, then switched to "for orders" — no declared final destination, so the cargo can be redirected, resold or left waiting at sea.
• Senator Blumenthal's letter puts a legal cloud over the cargo. Richard Blumenthal wrote to the company's managing director that selling the diesel into the US is illegal under the 2022 law, demanding the buyer, seller, charterer, insurer, terminal, distributor and claimed Treasury authorization. He concedes it is unknown whether this caused the destination change, but argues no refiner, terminal, shipper or insurer will be the test case.
• Where Russian diesel lands does not change the price impact. He says a cargo going to South America instead of the US affects global supply, and so American prices, the same way, so there is nothing to celebrate or mourn; he mocks expectations that one tanker would cut diesel by $2 a gallon, noting the first tanker hyped as coming carried fertilizer, not diesel.
• He bets no Russian diesel will be legally offloaded in America. His prediction is that any cargo heading this way will be stopped by federal law unless someone names a statute permitting Russian-origin diesel despite the ban, and he questions whether the administration, whose officials kept saying it was coming, knew the 2022 law existed.