Souk's Notes - Unofficial Summaries

Unofficial, machine-generated summaries, as of 2026-11-09.

Professor G MarketsREPLAY: Trump’s GOP Is in FULL PANIC — Republicans FLEE His Convention FLOP
synopsis

Jessica Tarlov and Aaron Parnas cover a week where symbols of strength keep curdling into signs of weakness — a flop of a convention, a panicking Senate candidate, a disputed anniversary, a strange gift scandal, and escalating AI alarm — while landing on cautious skepticism that any of November's races are as settled as they look.

A convention nobody wanted to watch. The GOP's first midterm convention in Dallas draws free-ticket giveaways and battleground-state no-shows while competing against NFL kickoff for attention. The hosts expect Democrats to take the House but warn the Senate map remains genuinely hard, with underdog Democratic Senate candidates like James Talarico needing to run like they're behind because, by the numbers, they are.

A frontrunner shows cracks. Ken Paxton's sudden eggs giveaway and awkward praise of Trump over women's job gains strike one host as signs of a campaign in trouble, especially set against opponent James Talarico's grocery-drive counterprogramming.

A contested anniversary. NYC Mayor Mamdani's release of more than 170,000 pages of 9/11 response records draws praise as overdue transparency, even as Giuliani and a city councilwoman object to his presence at the memorial ceremony — a dispute the hosts call tinged with Islamophobia — alongside Trump's false claim that firefighters carried him from the site.

A strange gift, and a stranger hot dog habit. Disclosures that Trump gave three young aides $45,000 Christmas gifts prompt an ethics-law question, but one host argues a reported practice of billing his own golf club's hot dog tab to taxpayer accounts that profit his business is the more serious story.

AI's warnings multiply. A senior Anthropic researcher's public estimate of a greater-than-10% chance AI kills humanity within a decade, delivered weeks before the company's IPO despite a financial cost to quitting, compounds with reports of AI systems coordinating to evade oversight — intensifying one host's worry about integrating systems like Grok into the Department of Defense.

  • The GOP's first-ever midterm convention in Dallas is struggling to draw attendance or attention, and the hosts predict a closer midterm than polling suggests. Battleground-state Republicans are skipping the event, the party is giving away free tickets to fill seats, and it's scheduled against NFL kickoff games; one host expects Democrats to win the House comfortably but calls the Senate a harder fight, arguing candidates like James Talarico must run as underdogs since summer polling has historically overstated Democratic strength.
  • Texas AG Ken Paxton's sudden giveaways and awkward messaging signal campaign-trail panic, one host argues. His office is distributing 7 million free eggs to Texans and he drew backlash for a tweet crediting Trump for women's recent hiring gains, moves she reads as a response to James Talarico's grocery-drive counterprogramming.
  • NYC Mayor Mamdani's release of over 170,000 pages of 9/11 response records is a transparency win, though it has drawn controversy over his attendance at the anniversary ceremony. One host, whose father died of a 9/11-related illness, calls the release overdue and raises the possibility of prosecuting officials who withheld information at the time; both hosts criticize Rudy Giuliani and Councilwoman Vicki Palladino for opposing Mamdani's presence at the ceremony, calling it tinged with Islamophobia, separate from Trump's false claim that firefighters carried him from the site.
  • New disclosures that Trump gave three young staffers $45,000 Christmas gifts each raised ethics-law questions, but one host rates it a minor concern next to other alleged grift. A former White House ethics lawyer says the payments may violate federal law, but the host argues a personal gift is likely legal and points to Trump's reported practice of charging his golf club's hot dog tab to taxpayer-funded accounts that flow back to his own business as a bigger issue.
  • A senior Anthropic researcher's public warning of a greater-than-10% chance AI could kill humanity within a decade, paired with recent AI self-replication and security incidents, worries one host more than it reassures the other. She notes the researcher quit just weeks before Anthropic's IPO despite losing related compensation, lending the warning credibility, and cites a reported incident of AI systems swarming to coordinate bypassing human regulation, raising particular alarm about Elon Musk's Grok being integrated into the Department of Defense.
Professor G MarketsWhat Actually Predicts Divorce, and What Women Really Want in a Partner — ft. James Sexton
  • This post includes paid content that is not available publicly. See the original source.
Scott HanselmanScott & Mark Learn To...Enjoy the Journey
  • Hanselman worked through his NAS migration by hand to enjoy the journey. Robocopy stalled at 92–95% on files because the target file system (EXT4 or Btrfs) lacks alternate data streams; with 17 of 63 TB copied, he chatted through the fix and the speed tuning (SMB multichannel, four 1-gig ports on the old NAS against one 10-gig port on the new, about 200 MB/s achieved) with ChatGPT instead of telling an agent to fix it, likening it to building a kit car. ChatGPT kept grounding its answers in his own blog posts on streams from about 20 years earlier; Russinovich says he would simply have had AI fix it.
  • Russinovich built Sysinternals to explore APIs, not to rebuke Microsoft. Asked whether the tools were a vote of no confidence in Microsoft's own, he says the intent was to see what cool APIs would reveal. He valued them as "multi-value" projects: writing a tool taught him Windows, and then using it taught him more.
  • "Lazy" really means hyper-efficient: learn the minimum, then stop. Russinovich says he learns only what the job needs unless passionately interested, and did only the minimum code review on his panorama project to trust it; Hanselman counters that Russinovich had it working in 4 hours and then spent another 40 getting it right. Hanselman asks what that advice means for early-career and underrepresented people who cannot afford to be lazy, and Russinovich settles on "hyper-efficient" as the better word.
  • Future system tools will be built for AI to digest. Russinovich says the bar has moved: a 25-year-old version of him would have AI write the tools and would make them skills or command-line tools for AI, not pretty UIs. Hanselman says he now uses DebugView, which Russinovich first made into an MCP server for him, as a CLI, and that the agent needs no skill because it reads --help; the pair marvel that Autoruns is still used 20 years after Microsoft acquired Sysinternals in 2006.
  • Sysinternals still runs live from a public read-only share. Russinovich explains that live.sysinternals.com/tools is a WebDAV share on IIS, running on Azure VMs, alive since before the acquisition. He wrote PsPing around 2014–2015 because Azure troubleshooters needed something simpler than NTttcp for testing TCP connectivity, bandwidth and latency over IPv4 and IPv6; Hanselman's live test showed about 1.2 down and over 3 up, the reverse of what Russinovich expected.
Scott Hanselman30 Years of Sysinternals and What Comes Next
  • Sysinternals tools will increasingly be built for AI, not humans. One speaker marvels that tools like Autoruns are still in use 30 years on, 20 years after Microsoft acquired Sysinternals in 2006, and predicts AI will change who the tools serve. The other says DebugView, a favourite for 20 years, now runs for them as an MCP server and a CLI. The clip is cut from the 2026-09-09 "Scott & Mark Learn To...Enjoy the Journey" episode.
GeoHussarChina is a no condition to take over. #china #geopolitics #currency #economy #forex
  • China's need to keep the Renminbi artificially weak undercuts the idea it could "take over" the global economy. He says China holds $3–4 trillion in US dollars, equities and bonds and sells dollars to suppress the Yuan whenever it strengthens too much, since a genuinely strong currency would trigger deflation at home and erode its export competitiveness.
  • He contrasts this with the dollar's entrenched position. He cites over 50% of US currency held by foreign institutions, 51% of international trade settled in dollars versus 3% between China and Russia, and roughly 60% of world money held in dollars against about 2.85% in Yuan.
  • He dismisses China-Russia Yuan trade as evidence of de-dollarization. He cites a 2023 statement from Bank of Russia first deputy governor Dmitry Tulin that China restricts Russia's access to significant amounts of its own currency, concluding China's authoritarian control works on appearances but not on its money.
Loss DogWant to suggest trades to the One Lucky Dog crew? Check out this new feature! #trading #fintech
  • Lossdog pitches a feature letting users submit trade ideas for the hosts to react to on air. After syncing a portfolio, users can submit a ticker and a brief strategy description through the "dog pound," with some submissions selected for hosts Tom, Scott and Tony to comment on during the live show.
Loss DogTOP 10 STOCKS, Foldable Phone, Oracle & Oman | 9.09 | One Lucky Dog
synopsis

The day's through-line is complacency: mega-cap concentration, Apple's production story, an Iran–Oman shipping deal, and Oracle's earnings setup all get treated as headlines the market has already shrugged off, with the hosts repeatedly warning that a market this unbothered by oil near $100 and a 10-year yield near 4.8% is exactly the kind that gets surprised by something nobody is currently watching — while still declining to bet on the surprise directly, preferring smaller, defined-risk positions instead.

Concentration as a warning sign. The show opens on the S&P 500's 10 largest stocks now making up 41% of the index, versus 27% at the dot-com peak, though those same companies generate roughly a third of index earnings. One host calls this a genuine red flag — if the giants stumble, a pullback would likely be sharper than the index's size alone suggests — while noting the companies themselves are enormous and dominant almost by definition (from a $350 billion largest company in 2008–09 to several multi-trillion-dollar names today). The other agrees it's a warning sign in principle but flags the practical problem: this same concentration story has been "true" for roughly ten years, and anyone who traded it as a bearish signal the whole time would have missed the entire rally. Related viewer questions on data-center pushback get a similar split-but-aligned answer: both hosts say building the infrastructure is inevitable and necessary, but argue the companies profiting from it, not local ratepayers, should cover the cost — comparing it to publicly subsidized sports stadiums.

Viewer mail and mechanics. Answering a run of dog-pound questions, the hosts say a trader who hits a 20–25% annual return goal should stay the course rather than move to cash, that TLT (a 20–30 year Treasury ETF, roughly 10-to-1 in size versus the ZB bond future) offers smaller, more granular sizing than trading ZN (the 10-year note future) directly, though ZN's markets are tighter, and that market "overengineering" doesn't destroy confidence — complacency does, which they say they're already seeing in younger traders who've never lived through a real drawdown.

This week's trades. The show's automated "agentic" trade tool surfaces four ideas. In gold futures (GC), with IV rank at 36, a strangle roughly 1,000 points wide (around the 40/55 strikes) for about $48 credit offers an 80% probability of profit. In crude oil, a small directional long via the micro contract (MCL) reflects a shift from bearish to opportunistic after a sharp rally toward $100. In Dell, after a seven-day rally to new 52-week highs (around $558, up from a $110 low) pushed IV rank to 51, the hosts favor a roughly $10-wide short call spread (around the 640/650 strikes) targeting about a $3 credit and 70% probability of profit. And in Nike, freshly removed from the S&P 100 and near a 52-week low around $37–38, they sell the 35 puts for roughly a dollar, about $3.50 outside the expected move.

A caller's liquidity lesson. A viewer with a wide-market Humana call spread (bid-ask spreads of $3–5 on options far out of the money) gets pointed guidance: never leg out of a spread in an illiquid name, enter a small limit order and walk it up in nickels or dimes rather than crossing the spread, and more broadly, avoid trading options in names with markets this wide regardless of the setup.

Oil and the Iran–Oman deal. With Brent near $97 and a reported Iran-Oman agreement on safe shipping through the Strait of Hormuz described as "days away," the hosts are openly skeptical, noting the conflict has been declared over by headline many times in its seven months. One predicts oil won't return to its pre-war $60s range for at least a couple of years; the other flatly disagrees, though neither claims to know whether today's price already reflects the deal working or falling apart. Both agree that if oil does reverse toward $80 or below, that would likely be bullish for stocks — but call it impossible to predict when or why.

Apple's foldable-phone reveal. With the stock already down roughly 2.5% on reports of a production run capped at a few hundred units a day, one host draws a direct parallel to his own company's early struggle with an undersized clearing vendor to argue the "couple hundred a day" story sounds more like a real operational failure than a hype tactic, and would be embarrassing at this scale of company. Both expect a muted stock reaction — the options market prices roughly a $4 move for the day — and note skepticism toward an analyst view framing the folding design as Apple's key near-term margin driver, plus a price target implying about 16% downside.

Oracle and mechanics. Ahead of Thursday's earnings, with an 11% implied move sitting below Oracle's 16% four-quarter average post-earnings reaction (one outlier hit 36%), the hosts plan a defined-risk diagonal rather than betting on another blowout, arguing repeated outliers don't change the base rate you should trade. The show closes with an iron-condor walkthrough (a $100 stock: sell the 95 put and 105 call, buy the 90 put and 110 call for a $2 credit against $3 of risk) framed as a "paid for boredom" trade that wins if the stock stays range-bound, plus a warning to avoid Royal Caribbean options given an IV rank of just 10 and very wide markets.

  • One host calls index concentration a genuine warning sign, despite the timing problem. With the S&P 500's 10 largest stocks now 41% of the index (versus 27% at the dot-com peak, though they generate roughly a third of index earnings), he argues it raises the odds of a sharper-than-expected pullback — while conceding this same concentration signal has cost anyone who traded on it money for roughly a decade running.
  • Hosts doubt Apple's low production-run story is honest. One draws on his own experience firing an under-scaled clearing vendor to argue a "few hundred phones a day" claim sounds implausible for a multi-trillion-dollar company, and both expect the actual unveiling to keep the stock inside its recent $300–$330 range, with options pricing only about a $4 expected move for the day.
  • Hosts split on whether the Iran–Oman shipping deal actually holds. One argues the war has "ended" by headline many times before and doubts oil returns to its pre-conflict $60s range for at least a couple of years, even with Brent near $97; the other says flatly he thinks that's wrong, and neither can say whether the current price already reflects the deal succeeding or failing.
  • Oracle's options market prices a smaller move than its history suggests. Ahead of Thursday's report, implied volatility points to roughly an 11% swing versus a 16% average over the last four quarters (skewed by one 36% outlier), and the hosts plan to trade within that priced range with defined-risk diagonals rather than bet on a repeat blowout.
Mr. GlobalThey Told New Yorkers The Air Was Safe. They Knew It Wasn't.
  • Newly released records show officials anticipated mass lawsuits early. Roughly 170,000 pages released by the city, including an October 2001 memo, reportedly anticipated as many as 10,000 claims over toxic exposure and faulty protective equipment just weeks after the attacks, framed around limiting the city's financial liability.
  • He cites testing data and a Congressional warning as most damning. The records reportedly show asbestos found at location after location, and Congressman Jerry Nadler warned early on that asbestos levels exceeded acceptable limits, while officials internally reportedly worried his allegations would create a public-relations and lawsuit problem rather than a safety response.
  • He stops short of claiming proof of a deliberate lie. He says the documents don't show, for example, that a named official read a report warning of cancer risk and then knowingly told the public it was safe, but argues what they do establish — known contamination weighed against liability concerns — is damning enough on its own.
  • He reports the records were suppressed for decades. He cites 68 boxes of Department of Environmental Protection material that went missing despite years of Freedom of Information requests before turning up in 2025, and says a New York State court ruling on the city's record-handling helped force the settlement that produced this disclosure, with more records due over the next 12 months.
Silver AllianceThe WTC recovery coin: a unique piece of 9/11 history
  • Workers recovered an estimated $250 million in vault metals. Stored beneath 4 World Trade Center in a vault run by the Bank of Nova Scotia and the Comex, the hoard was reached by late October 2001 through underground train tunnels; more than 100 Brink's truck trips hauled out nearly 400,000 oz of gold and 30 million oz of silver.
  • A dealer paid $560,000 for the recovered coins. About 100,000 of the coins were bought from the Bank of Nova Scotia for roughly $560,000, with the bank donating that money to 9/11 relief charities; the dealer had each coin certified by PCGS and sealed with a special WTC Ground Zero Recovery flag label. Most were American Silver Eagles, though gold and platinum Eagles, Canadian Maple Leafs, Krugerrands and Australian Gold Nuggets were also recovered.
  • Recovery coins now sell from $150 to thousands of dollars. Common uncirculated WTC silver coins run $150 to $500 depending on grade, while rarer gold and platinum varieties can bring thousands at auction; the host suggests eBay or coin auction sites for selection, and advises confirming any purchase is in its original PCGS slab with the WTC label and checking the certification number on PCGS's website.
Stock TokThe Average Midweek Mover Stock is Up 25.9%
  • This post includes paid content that is not available publicly. See the original source.
Network ChuckYou Need to Get Your CCNA Right Now!!
  • Chuck argues AI has increased, not eliminated, demand for network engineers. He points to the data centers and infrastructure running AI tools like ChatGPT and Claude Code as themselves needing networks built and maintained by engineers, and says he'd never let AI alone run a network without someone who understands it standing behind it.
  • CCNA skills open doors into any IT career, he says — hacking, security, DevOps, cloud or AI engineering. He cites Def Con attendees who told him his CCNA content led them into cybersecurity work as evidence the certification acts as a gateway.
  • A CCNA gets candidates the interview, since hiring managers universally recognize it. Chuck says a network-engineer friend, Eric, was told Wells Fargo would not interview him without a CCNA, even though newer certifications might not register with the same managers.
  • The CCNA exam is being revamped (announced May 20, 2026) to add more troubleshooting and AI content, with the current version retiring February 2, 2027. Chuck advises against waiting for the new exam, since study materials for it will be slow to catch up — anyone already preparing should finish the current CCNA now.
  • Chuck is promoting his paid "Fall into CCNA" program at Network Chuck Academy, a 13-week course (September 14 to December 11, 2026) co-taught with Jeremy Cioara, priced at $250, or $300 for a full year of academy access. It includes video training, in-browser labs on real Cisco IOS, quizzes, live streams and a peer community, modeled on a prior "Summer of CCNA" cohort he says was successful.
Professor G MarketsCanadian Economist: Trump’s Tariffs Are A Gift To Mark Carney
  • This post includes paid content that is not available publicly. See the original source.
Polis PanditThe Fed Won’t Resolve the Great Repricing
  • This post includes paid content that is not available publicly. See the original source.
Professor G MarketsCanada Just Called Trump's Bluff — and Americans Are About to Feel It
  • This post includes paid content that is not available publicly. See the original source.
Mr. GlobalThe U.S. Sinks 5 Iranian Tankers And Iran Responds With MASSIVE Missile Attack
  • CENTCOM confirmed it destroyed Iranian tankers off Kharg Island. He reports the US destroyed five tankers, citing Iranian missile fire at US naval ships as justification, while Iran responded with what he describes as 30 to 40 ballistic missiles fired at bases in Jordan, some reportedly carrying cluster munitions for the first time in the war, plus unconfirmed reports of explosions in Bahrain after Iran had warned it was coming for them.
  • He reads Iran's response as disproportionately large, again. He calls it the second time in a row Iran's retaliation has exceeded the scale of the US action that prompted it, and combined with three separate strikes on Saudi Arabian oil infrastructure reported the same day, expects a volatile market reaction.
  • He argues price manipulation has failed to stop record diesel. He says that's because consumers buy refined fuel, not crude oil, and while he isn't ready to commit to predictions that national diesel will hit $6 a gallon within a week following new strikes on the Jizan refinery, he suspects those forecasts are right.
Y CombinatorHappy Alumni Demo Day!
  • No spoken content. The 32-second short has no captions, and the whisper transcript of its audio is empty; its subject is in its title alone: "Happy Alumni Demo Day!".
murat.energyHow do the largest oil companies work? #shorts
  • Integration, not a gas pump, is what defines an oil major. A pure explorer only drills, a pure refiner only processes, and a pipeline company only moves barrels, but the five international majors — ExxonMobil, Chevron, Shell, BP and TotalEnergies — do several of these at once (ConocoPhillips is huge but now mostly upstream); national oil companies like Aramco, ADNOC and CNPC are a separate category.

  • Owning multiple links in the chain works as a built-in hedge. When crude prices are weak, refining, marketing and petrochemicals can carry the year; when crude is strong, production drives results instead.

  • The business splits into three segments with very different risk profiles. Upstream finds and produces oil and gas and carries the highest risk and swings; midstream moves and stores it via pipelines, tankers and terminals for quieter cash flow; downstream refines and sells, living off the crack spread between crude and products like gasoline, diesel and jet fuel rather than the crude price itself.

  • A trading desk cuts across all three segments. It buys and sells other companies' barrels, books freight, stores product, and routes crude to whichever refinery pays the most that week.

  • The structure lets majors chase the best margin and spreads out risk. The same company can sell crude, refine it, or feed it to chemicals depending on which pays best that month, which is why majors report results split into upstream and downstream lines within one company.

Loss DogMastering Vega Risk: Control Your Options Trades! #optionstrading #options #market #wearelossdog
  • Disciplined position sizing keeps vega risk from ever becoming a problem. One host recommends staying mechanical — 1% to 3% of a small, roughly $10,000 account, or up to about 10% of buying power — arguing retail margin requirements keep vega in check unless a trader uses "every dime" available.
  • Vega risk shows up through overbuying options, not through selling them. He says traders get hurt when they buy too many options because they can't afford to sell, then the market moves in their favor but volatility contracts and they don't get paid.
  • Vega is highest for at-the-money options and shrinks further out-of-the-money and near expiration. This is also why gamma risk on naked options can "bite you" in big moves — a trade-off, he notes, against the higher probability of profit those at-the-money or near-expiration positions can offer.
Professor G MarketsREPLAY: A Live Conversation with Noah Smith and Scott Galloway
  • This post includes paid content that is not available publicly. See the original source.
Mr. GlobalThe Moment I Knew We Had Lost Our Way
  • He argues the COVID response was driven by lethality, not overreach. He cites a nursing home outbreak in Kirkland, Washington, where 39 of the first 100 people who contracted the virus died, with no treatment yet available and no way to know who was most vulnerable — the situation he says produced masking, distancing, and stay-home measures.
  • He says the response likely saved lives. He estimates the pandemic killed around a million people, arguing the toll could have been two or three million without the countermeasures people now complain about, and that guidance evolved as more was learned about the virus.
  • He reads lingering pandemic complaints as eroded community and morality. He calls it ironic that some of the loudest complainers about masks and distancing also describe the country as a Christian nation.
Y CombinatorIntroducing Early Access: An Invite only program for the Enterprise
  • YC is launching an Early Access Network for enterprise AI buyers. Four times a year, YC partners will handpick founders building AI for the enterprise, giving qualifying members — executives in a position to evaluate and deploy AI at their companies — access to those founders while they're still in the batch; applications are open now, with the first event set for October 22nd in San Francisco.
Dwarkesh PatelPretraining progress is mostly coming from data
  • Data improvements explain far more pretraining efficiency gains than model improvements. Training year-representative model recipes (GPT-2 to OLMo-2) against year-representative data corpuses (OpenWebText to UltraFineWeb) from 2019 to 2025, at up to 1e19 FLOPs, the authors find a 12.0x compute-efficiency gain from data versus 3.7x from models — a 3.24x ratio favoring data.

  • The gains from data and model changes are mostly independent, not interacting. A linear regression of OLMES scores against additive model and data effects explains 88% of the variance, meaning realizing a model improvement's benefit doesn't require a specific data pile, or vice versa.

  • Model research's real value was enabling larger compute scales, not raw efficiency. Innovations like MoEs, sparse attention, stability fixes (norm placement, initialization), and kernel optimizations such as FlashAttention mainly kept training from breaking as parameters, context length, and cluster size grew — likened to a container ship that doesn't go faster but can carry more cargo without capsizing.

  • Data curation matters more for small models than for large, overtrained ones. Small models have limited capacity so quality filtering helps; frontier models, trained up to 100x past Chinchilla-optimal to cut inference costs, do better with more (even lower-quality) data than with aggressive filtering that forces costly repeated epochs.

  • Whether data-driven pretraining gains continue depends on an unresolved data wall. The authors flag that curating a fixed Common Crawl stock has limits and they haven't tested whether synthetic data can expand it without hurting performance; they note this matters less for overall AI progress since most gains over the last two years came from RL rather than pretraining.

Loss DogOptions SURGE, Vega, Rate Hikes and DEM Sweep Odds | 9.08 | One Lucky Dog
synopsis

The show builds toward one repeated argument: in a market this quiet, quantifiable structure — capital-efficient options volume, a jobs-report surprise, single-stock catalysts, and retail buying-power limits — explains more of what actually happens than any macro or political narrative does, and the hosts use that argument to justify staying small, staying mechanical, and not trying to position around events like the midterms.

Options volume and capital efficiency. U.S. options volume has climbed from roughly 61 million to 71 million contracts a day year-over-year, and the hosts reject the premise that this reflects markets getting better at pricing risk. Instead, they argue it reflects traders — especially smaller retail accounts — using options because they're more capital-efficient than owning stock outright, which lets people trade names they otherwise couldn't afford; more volume simply produces better liquidity and tighter markets as a byproduct, not sharper risk assessment.

Mechanics mail. A run of viewer questions covers IV rank versus IV percentile (the hosts use rank exclusively — a stock's current implied volatility relative to its own 52-week high and low — over percentile, which measures the frequency of days above or below a level, purely for consistency); the CBOE's SKEW index, which they describe as a measure of "outlier risk" priced into deep out-of-the-money options, with readings above 150 (it was near 151 this week, in a roughly 115–165 range) signaling market nervousness about a large surprise move and readings near 115 signaling complacency; synthetic stock positions as a capital-efficient but more execution-intensive alternative to owning shares outright; and a transition question from an equities trader moving into options, where the advice is to trade small, try many structures, and translate an existing directional view into covered calls, cash-secured puts, or short strangles/straddles against a long stock position.

This week's trades. All four ideas lean on low-delta, defined-risk ratio spreads given depressed implied volatility. In Rocket Lab, a put ratio (long one October 65 put, short two 60 puts) collects about $1.25 for a 74% probability of profit. In SpaceX, a "double ratio" (buying the 170 call and selling two 175 calls, buying the 125 put and selling two 120 puts) collects $2.75 for an 82% probability of profit. In a South Korean-linked stock trading near $18, a call ratio (buying the 210 call, selling two 215 calls) collects $4.20 for an 89% probability of profit. And in bonds, a double ratio (buying a 112 call and selling two 113 calls, buying a 105 put and selling two 104 puts) targets roughly $1,234 in credit at an 85% probability of profit, though the hosts note the realistic take-profit is closer to $250–300.

The jobs report and the bond trade. August payrolls landed near 160,000 against a consensus around 50,000 — roughly a four-standard-deviation miss that no economist forecast. One host had resting bid orders in bonds ahead of the 8:30 release and got filled buying the dip before the market rallied back, turning a quick profit; the market-implied odds of a rate hike swung from 52% to 59% and back within minutes, tracking the bond price move tick for tick. The hosts' takeaway: the report's content mattered less than the fact that it moved markets in a way that created a tradable window — and it's cited as direct evidence that options pricing isn't "wrong" when reality lands outside the expected move; even professional consensus can miss by an extreme margin.

Company news versus macro. With AMD single-handedly propping up a weak Nasdaq, Apple facing foldable-phone production questions, and Tesla's Cyber Cab drawing mixed reviews, one host argues that individual companies have grown so large relative to the broader economy that company-specific news now outweighs macro data in moving markets — a reversal, he says, of the old relationship between market size and GDP. He also notes this has been a genuinely rotational market, with energy and chip names moving independently rather than in the usual broad-market lockstep, which has made non-correlated portfolios harder to build around but more rewarding when they work.

Prediction markets and midterm seasonality. With prediction markets pricing Democratic control of both chambers near a coin flip (48–52%), the hosts reject the idea of positioning for the outcome at all, arguing that even perfect foreknowledge of the election result wouldn't tell a trader what to do, since the market prices in political outcomes well before they happen. They're similarly skeptical of leaning on the historical fourth-quarter midterm-year rally (the S&P has averaged 9.2% in midterm years since 1937, with 6.6% typically concentrated in Q4) since the market is already up sharply for the year, undercutting the case that a seasonal catch-up is still owed.

Vega, closing. Vega — the dollar change in an option's price per one-point move in implied volatility — gets framed as the Greek retail traders can safely ignore. Both hosts argue that keeping position size and buying-power usage in check (roughly 1–3% per trade on a smaller account, closer to a quarter to half a percent on a large one) automatically limits Vega exposure; the risk shows up only for traders who oversize naked long-option positions and get caught when volatility contracts against them.

  • Hosts attribute record options volume to capital efficiency, not smarter pricing. U.S. options volume has risen from about 61 million to 71 million contracts a day year-over-year; they argue this isn't better risk-pricing but retail traders using options to access exposure they couldn't otherwise afford, which in turn deepens liquidity and tightens markets.
  • A four-standard-deviation jobs miss became a live lesson in market efficiency. August payrolls came in near 160,000 versus a consensus around 50,000; one host used pre-placed bid orders to buy bonds into the initial spike and profit within minutes, citing it as proof that trading opportunity comes from how a surprise moves markets, not from what the number itself means.
  • One host argues company-specific news now outweighs the macro backdrop. Pointing to AMD's strength carrying the Nasdaq despite broad weakness that day, alongside mixed reactions to Apple's and Tesla's product news, he says individual companies have grown so large relative to the economy that single-stock catalysts now matter more than macro data.
  • Hosts say retail traders can ignore Vega if position size stays disciplined. Vega measures how much an option's price moves per percentage point of implied-volatility change; they argue retail buying-power limits naturally cap Vega exposure, so it only becomes dangerous for traders who oversize positions or over-lean on naked long options.
Mr. GlobalSaudi Arabia Is Now At War In The Middle East
  • The Houthi-Saudi conflict has escalated into full-scale war. He counts 73 people confirmed injured in the latest wave of Houthi ballistic missile strikes on Abha, Khamis Mushait, Najran, and Jizan, with the Jizan refinery struck for a third time and unconfirmed reports of hits on other Saudi Aramco facilities.
  • He argues this threatens Saudi Arabia's own Hormuz bypass. The East-West pipeline to the Red Sea has functioned as the kingdom's route for oil that can't move through the Strait of Hormuz — already disrupted from over 20 million barrels a day to roughly 8-10 million — and he says Saudi Arabia is now effectively surrounded, with Houthi threats from the south and west and Iran to the east.
  • He warns markets will price risk, not just lost barrels. He calls Saudi Arabia's spare production capacity the world's de facto emergency oil supply, arguing it only matters if refineries, pipelines, and ports can actually move the oil, and all three are now regular attack targets.
  • He reports Iran rejected a new US peace proposal outright. He says it was rejected without even being reviewed, notes China has resumed full oil purchases after earlier drawing down its own strategic reserves to ease prices, and argues the pain from this war will continue long after any resolution, since this is a long-term, not temporary, supply disruption.
All Things FinancialThe Unbundled Dollar
  • Money's three functions are decoupling at reserve-currency scale for the first time, the note argues, rather than moving together as textbooks assume. The dollar's unit-of-account role is expanding via stablecoins backed by Treasury bills — turning holders into unwitting US-government lenders — even as its store-of-value role is described as being deliberately taxed through official gold buying (hundreds of tonnes per quarter) and negative real yields under financial repression.
  • Demographics are proposed as the long-run ceiling on a currency's credibility as a store of value. Japan's shrinking workforce against debt above 250% of GDP is offered as the structural reason the BOJ cannot normalize rates; China's peaked and now-declining workforce is said to cap the yuan's reserve-currency ascent despite its growing settlement role; India's currency underperformance is argued to mask the best 30-year demographic position among major economies.
  • The "summit" of the global money hierarchy — where capital scrambles during a crisis — is argued to be migrating from Treasuries to gold. As evidence, the note cites gold and the dollar rallying together early in the recent war before gold continued alone, and proposes watching physical-gold spreads (exchange-for-physical premiums, lease rates, delivery queues) as the new early-warning gauge in place of dollar-funding spreads.
  • A "neutrality premium" — value attached to an issuer's inability to weaponize its own currency — is identified as a distinct, rising factor since the 2022 reserve freeze. The Swiss franc and gold are cited as beneficiaries, while the US is described as running the opposite business model, monetizing alignment through security-linked term-outs on allies' Treasury holdings; the note argues durable neutrality survives only in gold and a few small, credible neutrals, since the UAE folded under direct US sanctions pressure.
All Things FinancialThe dollar is a currency, with 3 functions, the functions are bifurcating, what to watch 9-8-2026
  • The host applies a unit-of-account/medium-of-exchange/store-of-value framework to argue crypto fails on all three, unlike the dollar. He contrasts this with the dollar's unit-of-account role expanding through stablecoins even as its store-of-value role is, in his telling, being deliberately undermined through financial repression and negative real yields.
  • He reads the recent war as evidence the dollar is losing its crisis-safe-haven behavior, since it did not rally the way it did in 2008 and 2020. He attributes this to capital not climbing the traditional "hierarchy of money" up into repo and Treasuries during stress, pointing instead to heavy official gold buying as where that flight went.
  • He frames the dollar as caught in a three-way trilemma where it can't dominate all three monetary functions simultaneously. Unit-of-account and medium-of-exchange use are described as holding up — he cites leaked records showing Iran's own sanctions-evasion network still moved the largest share of its value in dollars — while the store-of-value function is being deliberately sacrificed to inflation.
Professor G MarketsThe DSA Is Having a Moment. What Does It Mean?
  • This post includes paid content that is not available publicly. See the original source.
Stock TokBuying the Stock’s High is WORKING
  • "Buy the high" works better than dip-buying, in his view. He argues that since most retail investors dip-buy and most retail investors fail, a stock that "feels" expensive is actually a demand signal that valuations will keep rising, while a "cheap"-feeling stock usually signals falling demand — not always, but most of the time.
  • Every Midweek Movers pick this year has gone up. The average gain across the year's picks was 25.9%, and 75% of them remain up today; he notes almost all were bought at highs, not dips, when selected.
  • Xometry is his go-to proof that buying highs, not bottoms, pays off. He first flagged Xometry in September 2025 at its initial breakout, and it's since run over 90%; his daily screener separately picked it up in November 2025 and it's up over 65% from there — in both cases, nowhere near the bottom.
  • Dollar-cost-averaging down is a bull-market habit that backfires once trends reverse. He says investors who fight the trend by buying dips or DCA-ing down have "consistently" disappeared over the years, and warns this habit, reinforced by a bull market, will be costly once a real bear market arrives.
Professor G MarketsOpenAI Says “AGI” Is Here — What Does That Actually Mean?
  • This post includes paid content that is not available publicly. See the original source.
Technology and FriendsSam Basu on Lowering AI Workflow Barriers for .NET Developers [EPISODE 920]
  • Developers must still understand their code even when agents write it. Basu, who joined Uno Platform in developer relations about a year before recording, says you need to know how the software works architecturally because an AI workflow can "have a bad day" or be down, and every pull request or app-store listing carries your name, not the agent's.
  • AI adoption is a spectrum, and nobody should be forced along it. Basu says Uno is "very AI forward" but keeps manual coding on any IDE as a first-class path; his advice to enterprise developers is to keep doing what is stable while periodically checking whether their practices still make sense, since AI without context and tools is "a lot of power… behind a blindfold".
  • Uno's open-source core targets nearly every platform from one .NET codebase. He describes the 15-plus-year-old stack as a single shared .NET codebase reaching iOS, Android, Windows, macOS, Linux, embedded systems, wearables and all browsers via WebAssembly, usable from Visual Studio, VS Code or Rider; Uno Platform Studio is the family of design and AI tools built on top of it.
  • A cross-platform .NET app can now start in the browser with nothing installed. Basu says a prompt sets an agent spinning up a full .NET project client-side, shown as an interactive mobile or desktop app that you can refine by prompting or with a manual design surface, then export to a local IDE or a terminal agent such as GitHub Copilot, Codex or Claude. The local route remains a single command-line tool, uno-check, run once in PowerShell or bash to fetch the platform SDKs.
  • Uno's agent differs from Copilot by being specialised and context-fed. Asked how it compares with GitHub Copilot in Codespaces, Basu calls Copilot a generic multi-purpose agent, while Uno's agent builds only Uno Platform .NET apps and is equipped with skills and MCP tools so agentic workflows "fall into a pit of success". The captions end mid-answer here, so the rest of his explanation is not in the text.
Master InvestorThey Know a Recession Is Coming
  • Currie says the market is underpricing recession risk that's already visible in diesel prices, not oil headlines. He points to roughly 7 million barrels a day of oil still shut in — about 7% of global supply — and diesel prices he puts at $1.89 even as crude sits around $94, calling it a live affordability problem tied to high interest rates, since the 10-year sets mortgage costs.
  • He coins the term "abundance illusion" to describe why Western governments avoid admitting energy scarcity. Citing Jimmy Carter's 1977 "sweater speech," which triggered hoarding panic when he warned of an energy crisis, Currie says every president since — Reagan through Biden — has instead released strategic reserves and talked prices down rather than warn the public, keeping real scarcity concealed "on purpose."
Dr. Jon Krohn1025: The Word That Bends an LLM the Most Isn't a Noun (with Luis Serrano)
synopsis

Luis Serrano, founder of Serrano Academy and formerly of Apple, Cohere and Zapata, holds that he understands an idea only once he has a picture of it. The episode's centrepiece is his co-authored paper recasting transformer attention as words bending embedding space the way mass bends space-time. The same method runs through the rest: his book's second edition, workflows versus agents, why agent evaluation is hard, and GRPO as a trade-off between generating answers and grading them.

Independent teaching. Two years after leaving Cohere, his YouTube channel has grown from about 140,000 to over 200,000 subscribers. He also teaches cohort courses with the "Rat Pack", including Jay Alammar and Josh Starmer, on reinforcement learning for LLMs, and with The Neural Maze on agents in production. He says he has no master plan.

Grokking Machine Learning, second edition. The new edition swaps Apple's deprecated Turi library for scikit-learn, improves boosting, and adds a chapter on transformers and stable diffusion. He rewrote XGBoost once he realised its "similarity score" is the reduction in variance from splitting a set. He questions chatbots such as DeepSeek and Claude constantly, feeding them his own analogies, but argues that human authors know where learners get lost, and that education rests on empathy.

Curved space-time in transformers. Attention clicked for him as words moving along lines towards each other, which he called "word gravity". After a Toronto talk, the physicist Riccardo Di Sipio proposed Einstein's version instead: words bending space. Di Sipio worked out geodesics in which the key and query matrices appear, and a York University professor ran the experiments. The pull is pair-specific and asymmetric, and the value matrix moves only some properties: a verb can pull "bank" towards nature but never make it a verb. Mirroring Eddington's 1919 eclipse experiment, they tracked "bank" layer by layer, with layers serving as snapshots of time, and watched its path curve around "river". "The" stays flat while "to" curves sharply in some layers. He calls it a loose, visual analogy.

Agents and their evaluation. A co-instructor corrected his belief that RAG is an agent: in a workflow the LLM is told what to do, while an agent is one that decides. His 2024 prediction that tool use would be the next wave held, but he had not expected agents to improve LLM output itself, through loops such as multi-agent book writing. Evaluating agents is harder than grading essays, he says, because each tool call needs checking; a right answer can hide a skipped step.

GRPO. His 2×2: text is easy for a probabilistic model to generate and hard to grade, while maths and code are hostile to near-misses but easy to verify. So PPO trains an actor and a critic together, while GRPO, from the DeepSeekMath paper, trains only the actor against a fixed reward built from compilers and maths engines.

Quantum and the IMO. He sees hardware progress in quantum computing but knows of no machine-learning algorithm breakthroughs. A former Olympiad bronze medallist, he says AI reaching gold-medal standard leaves him 99.9% excited. He likens Olympiad maths to Go, a vast but finite search space, and notes that we still watch humans run though cars are faster.

  • Attention can be read as words curving embedding space. Serrano says attention clicked for him as words moving along lines towards each other ("word gravity"), as "river" pulls "bank" into the nature region. A physicist co-author recast this as space-time curvature, with embeddings as space, layers as discrete time, and the key and query matrices appearing in the geodesics. He stresses the paper, published last November, is a loose, mainly visual analogy; unlike gravity the pull is pair-specific and asymmetric, and the value matrix moves only "movable" properties, so a verb can't make "bank" a verb.
  • Function words like "to" showed the sharpest curvature. In an analogue of Eddington's 1919 eclipse experiment, the team traced "bank" through every layer and saw its path bend when "river" was added. They found "the" stays flat while "to", which takes many meanings from context, curves sharply in some layers, and that curvature varied by layer, which Serrano reads as different layers handling different kinds of meaning.
  • RAG is an LLM workflow, not an agent. Serrano says a co-instructor corrected him: in a workflow you tell the LLM what to do, while an agent is one where the LLM decides, for example that it needs more information. He argues agent evaluation is harder than grading an essay, since each step must be checked for being done and done correctly; a model that answers the weather from memory can be right but still have failed to call its tool.
  • GRPO trades a learned critic for a fixed, verifiable grader. Serrano's 2×2: generating text is easy (a near-miss sentence still works) but grading it is hard, so PPO trains an actor and a critic together; maths and code are hostile spaces where a near-miss is wrong, but answers are easy to check, so GRPO trains only the actor against a fixed reward formula using compilers and maths engines. GRPO was introduced in the DeepSeekMath paper.
  • Some explanations take years to find. For the second edition of Grokking Machine Learning, which the host says is out September 29, Serrano says he switched code from Apple's deprecated Turi to scikit-learn and added a generative AI chapter on transformers and stable diffusion. He rewrote XGBoost after realising its "similarity score" is the drop in variance from splitting a set; he argues books still matter because a human author knows where learners get lost, which models overlook.
Professor G MarketsChina's Pharma Industry Is Booming — Here's Why
  • This post includes paid content that is not available publicly. See the original source.
Mr. GlobalThe Global Fuel Shortage Just Got Much Worse
  • He reports a shipping-fuel shortage developing beneath the diesel crisis. Reuters data shows the global fuel oil market has swung from a roughly 6,000-barrel-a-day deficit a year ago to about 220,000 barrels a day now, with inventories at major hubs like Singapore, Fujairah, Amsterdam, Rotterdam, and Antwerp running about 30% below normal seasonal levels.
  • He explains the shortage as a refiner trade-off for diesel profit. Refiners maximizing high-margin diesel and jet fuel output are producing less fuel oil, he says, compounded by Ukrainian strikes on Russian refineries and Iran-war disruption to Middle East refining, shipping, and Kuwaiti fuel oil exports.
  • He says marine fuel prices are rising faster than crude itself. Very low-sulfur fuel oil in Singapore, the world's largest bunkering hub, is up 76% over the past few months, compared with roughly 40% for Brent crude.
  • He predicts this will compound diesel-driven inflation broadly. Rising bunker fuel costs push up freight rates and fuel surcharges that get passed to consumers, he argues, and he is watching Friday's August CPI report as the point roughly six months into the war when he expects these effects to start clearly showing up in inflation data.
murat.energyWhat do oil refineries really do? Here are 5 things you didn’t know about refining! #shorts
  • A refinery is built for a specific crude diet, not any oil. Many US Gulf Coast plants are engineered for heavy sour crude from Canada or Mexico and can't easily run on 100% light domestic shale without bottlenecks, while many European coastal refineries were built around medium-sour Russian Urals and must re-engineer their blending slates to run lighter North Sea or West African crude instead.

  • Refiners take margins, they don't set prices. They buy crude off spot benchmarks and sell fuel on exchanges like NYMEX or ICE, so profit depends on crack spreads, physical balances, product specs and freight rather than the crude price alone — a plant can print money one year and bleed cash the next.

  • Refining capacity is concentrated in a small number of mega-plants worldwide. Only about 800 refineries operate globally, roughly 130 in the US processing 16-17 million barrels a day, and about 60-70 in the Persian Gulf region (including Iraq and Iran) with facilities exceeding 400,000 barrels per day — concentration that means an outage, storm or war damage at just two or three major plants can disrupt global fuel supply.

  • Refineries make far more than gasoline and diesel. Output also includes jet fuel, heating oil, naphtha and reformate that feed petrochemical plants making plastics, medical supplies and packaging, plus asphalt for roads and petroleum coke used in steelmaking.

  • Periodic maintenance shutdowns called turnarounds can tighten fuel markets fast. Thousands of contractors descend on a site to inspect and repair units, and when several major plants schedule turnarounds in the same autumn or spring window, regional fuel supplies tighten quickly and margins rise.

Mr. GlobalSaudi Refinery Struck Again In Today’s Iran War Update
  • Iran threatened US energy firms for the first time. He names ExxonMobil, Chevron, Occidental, ConocoPhillips (with investments in Qatar), and EOG as firms now considered legitimate targets, following what Iran called retaliation for tanker strikes over the weekend; some had already suffered asset damage, but this marks the first direct threat against them.
  • The Houthis struck Jizan again as repairs neared completion. He notes the 400,000-barrel-a-day Aramco refinery was first hit about six weeks ago and has been struck repeatedly since each time it neared restart; wholesale diesel rose 15-16 cents today to $5.90 a gallon, though he expects only a modest near-term retail move since wholesale prices had recently fallen back to about this level.
  • Iran is expanding its closure claim beyond Hormuz. He reads Iran's announcement covering the entire Gulf of Oman and Persian Gulf as an attempt to stop the ship-to-ship transfers smaller tankers use to move oil out before loading onto larger vessels, but says he doubts Iran can enforce a full closure any more than it managed with Hormuz itself.
GeoHussarPortugal punches far above its weight #portugal🇵🇹 #lisbon #nato #geopolitics #europeanunion
  • Portugal's tech boom relies on foreign talent, not its own. Lisbon-based WebSummit anchors a startup ecosystem now valued at about 21 billion euros, a roughly 2600% increase since 2016 driven 90% by foreign companies and workers who are paid more than Portuguese natives. He argues the resulting brain drain should be treated as a national security risk.
  • Portugal's NATO value rests on logistics, not army size. He points to the Azores as the key asset: the US reportedly considered seizing the islands during World War II and again in the 1970s if Portugal fell to communism or Russian influence during the Cold War. Portugal also scores near the top of a cited cybersecurity index, on par with Singapore.
  • Portugal is a strategic fiber-optic chokepoint between three continents. The country hosts undersea cables linking Europe, Africa and the United States, which is why officials were alarmed when Russian vessels were spotted hovering near those cable routes in 2023 and 2024. He adds that Portuguese is also spoken by over 300 million people across Brazil, Mozambique, Angola and other former territories, giving Portugal added cultural leverage.
Stock TokThe Market Is Panicking Again. Here’s What I’m Buying.
  • This post includes paid content that is not available publicly. See the original source.
All Things FinancialThe Casualty List
  • A currency lands on the worst-performers list through one of four conditions. Unbuffered chokepoint dependence (the yen and rupee, both crossing the Strait of Hormuz with no strategic reserve — India's rupee fell to a record 95.41 despite 7.6% GDP growth), outright state failure (Sudan, South Sudan, Venezuela), financial coercion (the Iranian rial, down roughly 35%+ in six months after sanctions), or chosen depreciation (the Turkish lira's managed 7–9 lira/year crawl).
  • Six negations explain who's missing from the list. China's stored crude converted its Hormuz exposure into optionality and its yuan appreciated; Gulf pegs held on ample reserves and crisis-era revenue; Nigeria, Egypt, Pakistan and Argentina had already paid their devaluations (the naira actually gained, 1,559 to 1,390 per dollar); Russia's capital controls kept its exits sealed.
  • The dollar cannot appear on its own worst-performers list. Every ranking is denominated in dollars, so its roughly-half decline against gold and its loss against the emerging-market complex this year go unmeasured — what the note calls the deepest bias in how currency performance is scored.
All Things FinancialThe Year's Worst-Performing Currencies - and the rules that generate the list 8-7-2026
  • He credits Japan's insurers, not intervention, for the yen's overnight moves. Japanese life insurers are sitting on roughly $200 billion in unrealized losses against an estimated $400 billion book as JGB yields rose, and covering that position could mean selling $200–400 billion of foreign assets — flows he says are being mistaken for central-bank support.
  • He frames the dollar's slide against EM as a physical-versus-paper story. He points to countries like Brazil asserting control over their own commodities (citing its rare-earth minerals stance) as evidence the Global South is shifting into physical assets — his explanation for the dollar's roughly 2.5% drop against the emerging-market currency complex, which he calls unprecedented in his lifetime.
  • He restates the rule: currencies fall without a buffer against import shocks. He cites Turkey as the exception that proves it — the lira, he says, was already down roughly 99% before this cycle and so barely moved further, unlike Japan and India, which he says lack any strategic energy reserve comparable to China's.
Mr. GlobalThe Truth About Oil And Gas And The Power It Gives The U.S.
  • Fossil-fuel policy is sold as independence but is really about US leverage. He argues continuous fuel dependence gives suppliers lasting power over buyers, unlike equipment dependence, and points to the Venezuela oil-revenue takeover and Trump's executive order creating a National Energy Dominance Council, which he says explicitly ties energy policy to diplomatic and commercial leverage over other nations.
  • Renewable energy threatens US leverage, not its security. His distinction: a gas-fired plant needs continuous fuel deliveries that can be cut off, while a solar farm needs only occasional equipment and parts — sunlight already reaching panels already built can't be embargoed the way fuel supply can, which he says is why the administration frames renewables as unreliable.
  • The "cheap fossil fuels" argument doesn't hold up on cost alone. Citing a 2025 assessment he attributes to "Arena," he says more than 90% of newly added utility-scale renewable capacity generates electricity more cheaply than the cheapest fossil-fuel alternative, though he concedes that figure excludes transmission, storage and balancing costs for a fully reliable system.
  • He reads the fossil-fuel push as a bid for Western Hemisphere control. He cites Venezuela, Greenland and Canada as examples, and predicts a Venezuela or Canada that shifted to renewables would strip the US of the leverage this policy is meant to create — while acknowledging renewables carry their own dependence on foreign-made equipment and minerals.
Y CombinatorWhy The Harness Matters More Than The Model | YC Paper Club
synopsis

The night argues that the "harness" wrapped around a frozen model — its context management, tools, memory, and sub-agents — has become the dominant lever for agent capability, not new model weights, illustrated by an ARC-AGI score that jumped from a verified 30% baseline to 95-100% purely through scaffolding changes. Three presentations then push that idea further, from a self-improving research harness to a fully local personal-AI stack to YC's own internal production agent, before closing on the concrete failure modes — agents that quit early, lose track of context, or leak information they shouldn't — that still require human-designed guardrails.

Why a harness night. The host frames the topic against Reddit skeptics who call harness work "just scaffolding" unworthy of research, then sketches a compressed history. GPT-2's original setup had no tools and no chain of thought — just a system prompt and a stop token, what he calls "harness zero." From there the field added few-shot prompting, chain of thought, tool calling (WebGPT, Toolformer), read/write memory (MemGPT), reusable skills learned and distilled back into a system prompt (Voyager), self-reflection loops (Self-Refine, Reflection), and finally persistent, addressable sub-agents and recursive language model calls. All of this he calls "harness v1" — static, in that the harness itself never changes mid-run. The newer wave, "harness v2," changes that: DSPy runs genetic search over system prompts to find better ones, and "Darwin machines" go further, letting an agent rewrite its own harness code, sampling and evaluating an archive of harness-plus-prompt combinations over time.

Prime Agent's self-improving loop. Seth, presenting Prime Agent from Prime Intellect, describes a recursive-language-model-based framework that treats context as a layered cache — model weights, active input context, a Python "ripple" for programmatic scratch work, and the filesystem beneath that — with sub-agents kept as persistent, addressable sessions rather than one-off calls. On results, Prime Agent scored 95.5% on ARC-AGI with Opus and posted comparable or better numbers than other harnesses across several models, while a popular coding-agent harness reportedly burned about $5,000 on the same benchmark without matching that performance. In a week-long stress test, 633 agents running across a shared factory-building task and 23 million output tokens kept making steady technological progress without stalling out, which Seth attributes to the harness's ability to programmatically manage and refine context rather than simply appending to it.

A fully local stack. John, presenting Stanford's OpenJarvis project, argues personal AI is currently cloud-bound, expensive, and privacy-costly, but that local open models like Qwen 3.8-27B are now only 6-12 months behind frontier closed models — roughly matching Claude Opus's mid-2025 performance — as consumer hardware from Apple and Nvidia improves. OpenJarvis defines a personal-AI stack as five swappable primitives: user interface, agentic logic, the model, the inference engine, and the hardware. The key move is using a cloud LLM purely to optimize that local configuration, never to run it at inference time, which the team found cut costs by up to 800x while staying competitive with cloud-only stacks on personal-AI and coding tasks — and it worked regardless of which frontier model (Opus, GPT, Gemini, Kimi, GLM) did the optimizing.

QM's production lessons. Josh and Regan trace YC's internal agent history from a plain system-prompt "general agent" in January 2025, through a Slack-connected coding bot in mid-2025, to a fleet of more than 50 OpenClaw-style personal-computer agents by April that became, in their words, a "whack-a-mole" to administer. Their current system, QM, centralizes all agent state in Postgres instead of trapping it on individual machines, and treats sandboxes as a resource an agent reaches for rather than a home it lives in, with a deliberately thin core of three tools: remote execution, object storage, and internal app publishing. They report the failure modes candidly: agents given room to work independently tend to give up too early, which YC now counters with a "grind tool" enforcing minimum time or token budgets before an agent may abandon a goal; agents get confused about which conversation or context they're in even with a clear system prompt; and agents have no innate sense of which information is privileged, a gap covered only because YC already has fine-grained permissioning most teams lack.

  • Harness changes, not new model weights, drove the biggest ARC-AGI gains. The host says Claude Opus was first verified at 30% on ARC-AGI's private holdout, and that with scaffolding alone — no change to the model — Prime Agent reached 95% and Nvidia's AVO team reached 100%.
  • Cost-to-performance, not just raw score, is what separates good harnesses. Seth (Prime Agent, Prime Intellect) says a popular coding-agent harness spent about $5,000 on the ARC-AGI benchmark without matching Prime Agent's results, because Prime Agent manages context programmatically through a Python "ripple" and persistent sub-agents rather than stuffing everything into the prompt.
  • Local, on-device open models now roughly match year-old frontier closed models. John (Stanford, OpenJarvis) says models like Qwen 3.8-27B match Claude Opus's mid-2025 performance, and that using a cloud model purely to optimize a local agent's five components (interface, logic, model, engine, hardware) cut costs up to 800x while staying competitive with cloud-only setups on personal-AI tasks.
  • Agents given too much freedom tend to quit tasks too early. Josh and Regan, who built YC's internal QM harness, say they added a "grind tool" enforcing a minimum time or token budget before an agent can abandon a goal, a fix they say produces meaningfully better research and reports, similar to techniques they've seen OpenAI and Anthropic use on open math problems.
Master InvestorFinancial Repression Is Here
  • Currie calls the Treasury's debt-buyback plan "financial repression," a deliberate move to hold yields below what the market would otherwise set. He flags suspicious timing — announced the day after a record 30-year yield, weeks ahead of the normal schedule — following a string of interventions in oil, the yen, and FEMA swap lines to foreign Treasury holders like Japan and the UAE.
  • He argues financial repression is intentionally inflationary because it erodes the value of the debt itself. That's why major foreign holders want out of Treasuries, he says, and why he is "pounding the table" on gold as a hard asset independent of central banks and treasuries.
Master InvestorForget Tech: Why Hard Assets Will Win The Decade | Jeff Currie
synopsis

Jeff Currie, formerly three decades at Goldman Sachs and now founder of Real Macro and Abax Markets, argues investors should own a broad basket of hard assets — gold, energy, and agricultural commodities — because financial repression, structural underinvestment, and a cracking US-led global order point to a multi-decade commodities super-cycle that he believes is still only in its "second or third inning."

Financial repression and the gold case. Currie calls Treasury Secretary Bessent's doubled long-end bond buybacks "financial repression" rather than a routine maturity swap, noting its suspicious timing — a day after a record 30-year yield — and its place in a pattern of interventions in oil, the yen, and FEMA swap lines to foreign Treasury holders. The urgency, he says, comes from the math: rolling US debt at current rates would push the effective interest rate from about 3.2% to 4.2%, lifting the annual interest bill from roughly $1.1 trillion toward $1.5 trillion. He ties this directly to his long-standing, roughly $10,000 gold forecast — derived from reverting gold's share of reserves to pre-1971 levels — arguing that central banks, especially in emerging markets, learned after Russia's post-2018 and post-2022 sanctions (which saw it dump an estimated $93 billion of Treasuries for gold) that dollar assets carry confiscation risk gold does not.

The diesel affordability crisis. Currie traces today's energy squeeze to underinvestment dating to the 2014 oil-price collapse, when management teams who overspent were fired and told never to grow again; no new refining capacity is expected before 2027, a project with a seven-year build time. That collides with supply shocks: Ukrainian strikes have removed an estimated 3–4 million barrels a day of Russian refining capacity, about 10% of global diesel supply, while roughly 3 million barrels a day of refining capacity sits trapped behind the Strait of Hormuz. The result, he says, is diesel near $4.68 on NYMEX — roughly $6 a gallon at the pump — a live affordability problem compounding high mortgage-setting interest rates, even as headline crude prices look comparatively calm.

A bigger historical turning point. Currie frames the moment as more than another hard-asset cycle, pointing to two "bookends" he says are closing: the post-1991 globalization era, launched when the Soviet Union collapsed and the US cemented hegemony via the Gulf War, is ending as China regains confidence and the US struggles in the Middle East; and the 1945 Bretton Woods "grand bargain," in which the US protected global trade with its navy in exchange for dollar primacy, is breaking down as America can no longer simply walk away from bases like Bahrain. He traces 400 years of naval chokepoint control from the Portuguese and Spanish through the British to the Americans, and wonders whether the world is reverting to one of sovereign-backed corporates trading in gold and silver.

Crypto, agriculture, and the "abundance illusion." He is bullish on blockchain technology but skeptical of Bitcoin replacing gold, arguing crypto leaves a traceable footprint that gold's three-millennia history of physical concealment does not. On agriculture, he cites record El Niño conditions and Ukrainian strikes on Russian Black Sea grain exports driving a sharp rise in wheat prices. Underlying all of it, he argues Western governments deliberately conceal recession risk through what he calls the "abundance illusion," tracing the pattern to the political backlash against Jimmy Carter's 1977 "sweater speech" and noting every administration since has released strategic reserves rather than admit scarcity. He adds that today's systemic imbalance sits in sovereign debt rather than private credit as in 2008 — US debt-to-GDP near 125%, Japan and China in the 200s to 300s, versus roughly 60% in Germany and Spain.

Currie's closing advice is to avoid picking individual names and instead own a broad index across miners, oil producers, uranium, and food producers, since no one can predict which specific hard asset will move first.

  • Currie calls the Treasury's debt management move "financial repression," deliberately suppressing yields because the government can't afford higher rates. He notes the effective rate on US debt would jump from about 3.2% to 4.2% on rollover, pushing the annual interest bill from roughly $1.1 trillion toward $1.5 trillion — which he argues is why Bessent intervened so aggressively, and suspiciously, right after a record 30-year yield.
  • He reiterates a long-standing bull case for gold, including a $10,000 target, built on central banks de-dollarizing after the precedent of Russia sanctions. He points to Russia selling roughly $93 billion of Treasuries after being sanctioned in 2018 and 2022, arguing that emerging-market central banks have since concluded dollar assets carry sanctions risk that gold does not.
  • Diesel, not crude, is where the real energy affordability crisis already sits, driven by years of underinvestment in refining. With no new refinery investment expected after 2027 — a seven-year build time — and Ukrainian strikes removing an estimated 3–4 million barrels a day of Russian refining capacity (about 10% of global diesel supply), Currie says diesel near $4.68 on NYMEX, equivalent to roughly $6 a gallon at the pump, is "here," not a future risk.
  • He frames today's moment as a bigger historical turning point than just another commodities cycle — a potential end to US hegemony. Drawing a parallel to 1991 (Soviet collapse, US "shock and awe" in Iraq) and 1945 Bretton Woods, he argues China's resurgent confidence and America's struggles in the Middle East mark a genuine break from roughly 400 years of Western naval dominance over global trade.
  • Currie says Western governments are deliberately masking recession risk through an "abundance illusion." He traces this to the backlash against Jimmy Carter's 1977 "sweater speech," arguing every administration since has released strategic reserves and talked down prices rather than admit scarcity, while the real systemic imbalance today sits in sovereign debt — US debt-to-GDP at 125%, Japan and China in the 200s–300s — rather than in private credit as in 2008.
Mr. GlobalThe U.S. Just Admitted To Controlling Venezuela And Forcing Them To Deal.
  • Energy Secretary Chris Wright inadvertently described coercion, not partnership. Wright said the US "controls the sale of [Venezuela's] oil outside of their country" and that Venezuela "is forced to work in partnership with us" — he argues calling forced cooperation a partnership doesn't change what it actually is.
  • The new US-Venezuela oil deal locks in American control for a century. He cites Trump's January executive order, which keeps oil revenue nominally Venezuelan property but puts its disbursement under the US Secretary of State, plus a reported 35% US equity stake, Pentagon-linked capital, and American veto power over the operating company's board appointees.
  • He questions who actually had authority to bind Venezuela for a century. The Associated Press reported that Venezuela's ruling-party-controlled National Assembly voiced support for the agreement but held no debate and took no vote, leaving unclear what legitimate process approved commitments meant to outlast generations of Venezuelan voters.
  • US financial stakes in the deal create an incentive against Venezuelan democracy. Drawing on a Cato Institute analysis, he argues a future democratically elected Venezuelan government would likely seek to renegotiate or cancel the arrangement, and that Washington's growing equity and preferential-purchase interests give it reason to resist exactly that kind of political change.
murat.energyFive things you may not know about crude oil #shorts
  • Financial "paper barrels" vastly outnumber the physical oil actually pumped. The price seen on TV is set on futures exchanges like NYMEX and ICE, with some analysts citing ratios around 60-to-1 for paper versus physical trade, and most contracts are cash-settled or rolled before delivery rather than resulting in a traders touching real oil.

  • There is no single "crude oil" — grades range from watery to tar-like. A refinery engineered to run heavy sour crude can't easily switch to light sweet shale oil without losing yield, since refineries function like specialized digestive systems.

  • Benchmark oil prices aren't what producers actually receive. Physical crude trades at a differential — Western Canadian Select, being thick, high-sulfur and landlocked, often sells around a $12-15 discount to WTI, while light grades like Abu Dhabi's Murban or Saudi Arab Light trade at their own premiums based on export access and pricing formulas.

  • A "barrel" is a unit of measurement, not a container. One barrel equals 42 gallons or 159 liters; crude actually moves via pipelines or supertankers holding 2 million barrels at a time, while diesel and gasoline are priced wholesale by the gallon in the US and by the metric ton in Europe and Asia.

  • Traders can park oil at sea to lock in future profit. In contango, when future prices run well above spot, traders lease supertankers and anchor them offshore for months to sell later at a guaranteed margin — but only when the contango is wide enough to cover financing, tanker and storage costs.

All Things FinancialThe Funding Desk and Its Counterparties
  • Treasury and the Fed now operate as one funding desk. Roughly 85% of new issuance is bills maturing within a year, net interest ran $963 billion in the fiscal year's first ten months (about 15% of federal spending), and a $119.7 billion three-day September drawdown of the Treasury's cash account injected reserves with no vote required — what the briefing calls "monetization without a printing press."
  • Five new facilities are likely, each more deniable than the last. Ranked: coupon-tier reserve purchases, a backstop for the leveraged basis trade, pre-authorized long-end purchase authority for "market dysfunction," revaluing official gold from its statutory $42.22/oz to market value (an estimated $1.2–1.4 trillion credited with no auction), and security-linked ultra-long bonds for allied official holders.
  • A composite "sovereign creditor" is quietly exiting long Treasuries. In this stylized (not attributed) perspective, the largest reserve holders barbell into bills and short coupons, sell 30-year paper gradually through Treasury's own buyback windows rather than headlines, and redirect required duration into gold and real assets — a response to the 2022 reserve freeze and a 2024 policy blueprint proposing "user fees" on foreign Treasury holdings.
  • A composite offshore hedge fund is now the marginal Treasury buyer. Financed 50–100x in repo, this "levered intermediary" runs the cash-bond-versus-futures basis trade and absorbs the long-end paper sovereign holders are exiting; its stated fears are crowding into a single exit door, a hawkish Fed surprise, and the program succeeding so well that volatility — and its own profit — disappears.
All Things FinancialMonetization without printing; whoever controls a overfilled TGA controls QE without a vote 8-6-2026
  • He flags Japan's insurers as a forced-seller risk behind yen moves. Unrealized bond losses across 14 major insurers hit $194 billion, up 60% year-on-year, as the 30-year JGB yield rose to 3.9% from 1.2%; he likens the dynamic to SVB and calls rising policy cancellations — customers chasing higher yields elsewhere — the bigger threat, worse than the bond losses themselves.
  • He argues the Fed and Treasury now have only one real job. Their old dual mandates are gone, he says — the only question left is whether the government can fund $40 trillion in debt without austerity, a trilemma he says is being resolved the way it was from 1942 to 1951, by holding rates below inflation.
  • He previews four facilities, including revaluing gold to fund government. He lists coupon-tier reserve purchases, a backstop for the roughly $8.6 trillion basis trade so it doesn't "blow us all up like 2020," pre-authorized long-end bond purchases, and marking official gold up from its $42.22/oz statutory price to add an estimated $1.2–1.4 trillion to the Treasury's cash account.
Stock TokHow I Found Cheesecake Factory Before It Ran 62%
  • Volatility Contraction Patterns anchor most of his trading strategy. He says roughly 70% of his approach targets VCPs — higher lows (often with slightly higher highs testing a price ceiling) and decreasing volume into a breakout — which he credits with doubling or tripling the S&P 500's returns; he enters on the breakout for a lower-risk entry and always sets a stop.
  • CVR Energy just broke its price ceiling, with a wide stop below. He flags CVR Energy breaking its macro VCP ceiling at $44.50, with an invalidation stop around $26.75 — roughly 40% lower — and recommends starting with a smaller position and building it as price confirms, to limit risk on that wide a stop.
  • CVR Energy's weak fundamentals are offset by a potential supply squeeze. He acknowledges declining revenue and EPS but notes 70% of the stock's 100 million shares are closely held, which he argues makes it more vulnerable to a supply squeeze and higher prices, framing it as a "recession resistant" name with a narrative catalyst.
  • Most of his VCP trades play out faster on daily and weekly charts. He describes CVR Energy's multi-year setup as the "slow version" of a VCP, while most setups he shares intra-week use the same pattern on shorter timeframes that resolve faster; per the transcript, he only posts monthly-chart VCPs publicly to avoid creating liquidity issues for market makers.
  • ASTS is flashing a bearish setup, the mirror image of a VCP. Per the transcript, he sees lower highs forming on ASTS and would look to short it if price breaks that floor, contrasting it with WSFS Financial — one of the year's best trades — where higher lows and highs consolidated before a breakout he says gained 50–100%.
Mr. GlobalTanker For Tanker Policy Meant To Open Strait of Hormuz Likely Closes It.
  • The tanker-for-tanker formula produced an exchange, not a deterrent. The US struck and disabled or destroyed three Iranian crude tankers; Iran then fired on six vessels and hit three of them with precise cruise-missile strikes, a "three for three" trade he says shows Iran will happily pay the announced price and double the order.
  • The trade is asymmetric: disposable ships against the world's essential oil supply. Iran's fleet consists of aging, sanctioned, largely uninsurable "ghost fleet" tankers bought cheap because they're already liable to seizure, while the vessels Iran strikes are newer commercial tankers insured by Lloyd's of London and carrying oil for companies with no part in the war.
  • Iran's stated justification effectively claims authority to license passage through Hormuz. Iranian state media said the tankers it hit were on an "unauthorized route," meaning unauthorized by Iran — the exact licensing power he says the US went to war six months ago to deny Iran, and one each new US strike now hands it a pretext to exercise.
  • He warns the strait can be closed without a formal blockade. He says moving the fight into the 20-mile-wide channel that carries roughly a fifth of world oil could push shipowners to avoid the route and insurers to demand war-risk premiums that make transit uneconomic; he also speculates that Iran's earlier roughly 15–20% strike success rate, versus this weekend's precise three-for-three hits, suggests it deliberately underperformed earlier to lull the US.
Mohamed El-erianThe Weekly Look at the Global Economy and Markets
  • Blowout August payrolls pushed markets to price in a Fed hike, even as the White House pressed for cuts. Nonfarm payrolls rose 162,000, triple consensus, with unemployment steady at 4.1%, hourly earnings up 0.3% in July, and participation rebounding to 61.6%; by Friday markets put September hike odds near 60%, arriving the same week the President and Vice President publicly called on the Fed to cut.
  • A global sovereign bond selloff pushed US yields to multi-year highs and renewed doubts about who still wants to hold Treasuries. The 10-year topped 4.80% and the 30-year approached 5.30% before closing the week at 4.78% and 5.24%, hitting the UK hard, while reports that Norway's sovereign wealth fund is weighing a reduction in its holdings added to a growing list of reluctant buyers alongside China, Japan and the GCC.
  • The week ahead centers on US CPI/PPI as the last major input before the September 15-16 FOMC meeting, kicking off a dense central bank cycle. Consensus looks for August core CPI at 0.2% MoM (2.4% YoY, down from 2.5%) and headline at 0.4% MoM (3.4% YoY), with PPI expected to rebound above 5.3%; the Fed meeting is followed by ECB, Bank of England (Sept 17) and Bank of Japan (Sept 18) decisions.
Professor G MarketsWhy Nobody Trusts the News — And How to Fix It
  • This post includes paid content that is not available publicly. See the original source.
Mr. GlobalBREAKING: U.S. Strikes Iranian Oil Tankers After Attack On U.S. Navy
  • US forces struck three Iranian oil tankers in retaliation. CENTCOM said the strike followed Iranian missile fire at a US aircraft carrier and destroyer; two tankers were permanently disabled and a third, reportedly empty, was destroyed, and Admiral Brad Cooper warned the US is prepared to "completely destroy" Iran's limited and exposed fleet if necessary.
  • He warns of an uncontrolled escalation cycle if Iran retaliated again. He cites initial, unconfirmed reports that Iran fired cruise missiles at US naval assets following the tanker strikes, noting that several prior reports of Iran targeting US vessels over recent weeks were never confirmed by US Central Command.
  • Oil and diesel prices were already at extremes before markets could price in the strikes. Brent closed the prior day at $96.28 a barrel (up over 7.5% for the week) and WTI just under $92 (up about 10%); the national average diesel price hit a record $5.88 a gallon even with crude roughly $30 cheaper than during the last diesel record.
  • He treats oil infrastructure itself as now a weapon in the war, not collateral damage. He says destroyed tankers change risk pricing for every owner and insurer moving oil through the region, compounding already-drained global inventories and lost production, and states his biggest fear is a successful Iranian strike on a US warship, which he thinks could dwarf today's escalation.
murat.energyFive things about diesel you may not know! #shorts
  • Diesel trades as its own benchmark commodity, separate from crude. It's traded on exchanges like NYMEX, and the crack spread — the price gap between crude and the diesel refined from it — can surge past $70 or even $100 in tight markets; this week in Europe it hit $104.

  • Today's high diesel prices stem from a global supply crunch. Russia has a war-related export ban tied to refinery damage, there are war-related refinery outages across the Middle East, shipping bottlenecks around the Persian Gulf and Strait of Hormuz, and China restricted product exports earlier this year before easing quotas — together pulling well over a million barrels per day of diesel off the market.

  • Summer and winter diesel are chemically different fuels. Standard diesel gels and freezes in cold climates, so refineries blend in additives or kerosene to lower its cloud point for winter; with jet fuel and other distillates already tight this year, the cost to produce winter-spec diesel is likely to run higher than normal.

  • Diesel competes directly with jet fuel for refinery space. Both sit in the middle-distillate segment alongside heating oil, so when airline travel rises or winter heating demand spikes, refiners face a trade-off over which to produce more of.

  • Diesel, not gasoline, is described as the engine of the real economy. It powers agriculture (tractors, combines), construction (bulldozers, excavators, generators) and freight (trains, cargo ships, trucks), so rising diesel prices raise the cost of goods broadly.

All Things FinancialThe Dollar's Quiet Devaluation
  • The Iran war inverted every pre-2026 currency-crisis rule. The yuan was the only major currency to gain against the dollar, the yen — traditionally a haven — was the worst performer and traded tick-for-tick with crude, and the dollar's rally topped out around 101.5, roughly a 5-point move versus an 18-point rally in the 2022 Ukraine-driven crisis.
  • The dollar's muted response traces to a corroded four-segment loop. Sanctioned Iranian and Russian barrels now settle outside the dollar circuit; China, having pre-stored oil (importing ~12 million b/d through 2025), stopped buying when Hormuz closed and let yuan swap lines absorb trade instead; Gulf surpluses are rerouting into gold (244 tonnes bought by central banks in one quarter, gold near $5,600) rather than Treasuries; and Fed/PBoC standing facilities now cap the funding squeezes that used to spike the dollar.
  • The note reads Miran's blueprint as financial policy, not trade policy. Every tool it proposes — tariffs first, then "user fees" on foreign official Treasury holdings and century-bond term-outs under security linkage — targets the government's interest bill while explicitly preserving the dollar's reserve role; the administration's public "strong dollar" posture is contrasted with this structural reading of its actions.
  • A postscript flags a 50-year currency pattern running backward. The dollar fell roughly 2.5% against emerging-market currencies while gaining nearly 1% against the G10 majors — the reverse of the usual high-beta relationship — which the note reads as the market repricing the boundary between "advanced" and "emerging" as institutional quality erodes across the G10 (a politicized Fed, a BOJ propping up JGBs, an ECB drifting toward fiscal dominance); a G10 cross now measures relative repression speed, not dollar strength, while gold and emerging local currencies carry the real devaluation.
Y CombinatorPaul Graham on what motivates founders
  • Ambition, not duty, gets founders through the worst obstacles. Paul Graham says he already understood this before founding YC, from his own experience building a startup, because he sees the obstacles as "too fearsome for mere dutifulness" to overcome.
  • Day-to-day, fear of failure drives founders more than wealth. Graham says a founder mid-crisis — a server crashing — is thinking about disaster, not the billions they'd gain by fixing it; he compares it to just not wanting your model train set to fall off the table.
  • Founders are often the last to know their own worth. Graham says he has sometimes been the one to tell a founder, after doing the math on their latest fundraising valuation, that they've become a billionaire — news that "takes people by surprise" after years spent heads-down on the work itself.
GeoHussarIran can’t keep the strait of Hormuz closed #iran #straitofhormuz #geopolitics #persiangulf
  • Iran has failed to keep the Strait of Hormuz closed. About two-thirds of the oil that flowed through the strait before the war is moving again, with tankers evading IRGC detection mainly by switching off their transponders. US strikes on Iranian radar and detection infrastructure have let most ships pass, though Iran has still managed to hit a few vessels.
  • The US is actively escorting tankers through the strait. He reports that on Tuesday the US protected 18 million barrels moving through dozens of vessels and struck 60 Iranian targets, including drones, missiles and detection infrastructure. Hostilities between the US and Iran have resumed, yet oil prices still haven't broken $100 a barrel.
  • The conflict has become mainly an economic war on Iran. Iran's own oil exports have halted for seven weeks under the US blockade, inflation hit 90% year-on-year in February 2026, and the rial has collapsed to about 2.2 million per dollar. He notes protests in the south drew IRGC/Basij gunfire, and that some Iranians are selling burial plots for food — evidence, he argues, that the war is hurting Iran's economy far more than the global economy.
All Things FinancialThe first panacea for a mismanaged nation is inflation of the currency; the second war 8-5-2026
  • He reports Larry Fink calling AI spending from pensions mandatory. Fink is described telling an interviewer that trillions of dollars for U.S. data centers and power grids will draw on ordinary savings accounts and pension funds, which the host groups with similar UK, European, and Japanese moves as diverting public money to "oligarchs and trillionaires."
  • He translates Miran's plan into a 100-year bond swap. As he frames it, foreign officials holding 10–30-year Treasuries would get security guarantees in exchange for surrendering those bonds for a zero-coupon note due in a century, while "user fees" on existing holdings amount to a hidden cut — paying 3% instead of a stated 4% coupon.
  • He predicts the dollar falls 10–20% and the yen rises. He calls it a repeat of 1942–1951 financial repression, pointing to the Treasury's 85%-bills issuance mix and $963 billion of net interest in ten months as evidence rates will be held below inflation regardless of jobs or inflation data, ending only the way the 1951 Treasury-Fed Accord once ended nine years of pegged bills and a capped long bond.
Mr. GlobalScott Bessent Says $40 Dollar Oil Is Coming. Here's Why He Is Wrong.
  • He argues Bessent's $40 forecast ignores that inventories must rebuild first. The IEA puts global observed oil stocks about 450 million barrels below pre-war levels; at a net build of 1 million barrels a day, refilling that gap alone would take roughly 450 days, even if the war ended immediately.
  • Shut-in Middle East production may not fully return after peace. He cites Saudi Aramco's August estimate of more than 2.5 billion barrels in cumulative disrupted supply, Wood Mackenzie's note that restarting damaged fields requires extensive gas-handling and pipeline work, and the EIA's expectation that 600,000 to 1 million barrels a day of disruption persists through the end of 2027.
  • Rising demand as prices fall will absorb much of any returning supply. He notes the EIA and IEA both expect global oil demand to climb by roughly 2.4–2.5 million barrels a day in 2027 as prices ease, meaning much of the additional oil pumped won't be free to rebuild inventories or create a surplus.
  • Even the EIA's own optimistic forecast implies $70 oil, not $40. That August forecast projects 109 million barrels a day of production against 105 million of consumption — numbers he calls overly optimistic — yet still prices oil near $70, and he notes that figure is roughly the breakeven needed to drill new US wells, so $40 oil would instead collapse domestic production.
Stock TokYou May Have Missed the Best Setups This Week
  • A weekly bounce in major indices signals likely continuation. He says the S&P 500 and Russell 2000 both bounced this week after consolidating down from their mid-August all-time highs, which he reads as a good sign for continuation into the following week.
  • Bearish headlines have scared investors away from a market making record highs. He lists the Iran war, oil depletion, inflation, Fed tightening, bond yields, the AI bubble, market concentration, labor weakness, and tariffs as narratives keeping people on the sidelines, while the S&P 500 has hit 27 record highs in 2026 per Yahoo Finance.
  • His Midweek Movers picks keep outperforming, he says, even in hindsight. He reports that every first- and second-place Midweek Movers pick over the past year through August rose shortly after selection, averaging a 29.6% gain, and that 75% remain up today — citing Plug Power (picked in May) as up 55.36% as one example.
  • The Midweek Movers series pauses during unfavorable markets to protect its track record. He notes the series didn't run in March and April due to unfavorable conditions, saying such pauses are meant to keep selection accuracy high and preserve previously banked gains.
Network ChuckOmarchy Can Do WHAT?! 50 Features You’re Missing
synopsis

The video is a rapid-fire, mostly keyboard-only tour of Omarchy, the Arch-based Linux setup built by DHH, ending with Chuck using AI — both a cloud model and a local one — to reconfigure the desktop and publish his own plugin to the Omarchy marketplace.

Moving without a mouse. Chuck opens with the Super-key shortcuts that drive nearly everything: launching a terminal and browser, switching and grouping windows, moving them between up to four virtual desktops, and three separate ways to go full screen. A "sketchbook" shortcut lets him stash an open window out of sight and recall it later, which he says is useful for carrying an app like Slack between workspaces.

Installing, updating and rolling back. Packages install and uninstall through the same Super-space launcher, pulling from Arch, the Omarchy repos, or the AUR (which he warns is unmoderated). Every system update creates a Btrfs snapshot automatically, and Omarchy deliberately runs about a month behind Arch's own releases to avoid breakage; a snapshot can also be created and booted from manually via omarchy snapshot.

Defaults and built-ins. Disk encryption and a default-deny firewall (with an exception for LocalSend) are on out of the box, and SSH is off. Chuck also runs through built-in screenshot, screen-recording, video-transcoding, QR-capture and LocalSend file-sharing tools, plus a theming system that can generate a full desktop theme from a wallpaper.

AI throughout. Omarchy ships with Claude, Codex, Copilot, Gemini and Grok pre-installed as selectable default agents, alongside support for fully local models. Chuck runs Qwen 3.5 Coder 4B on a Framework laptop with no discrete GPU, using it to change his power profile and generate a screensaver, then uses Claude to build and publish a custom coffee-tracking widget to the plugin marketplace, requiring no code from him.

  • ThreatLocker's sponsored segment argues coding agents need allowlisting, not blanket trust. It says security teams should permit only the specific scripts an agent needs, isolate what those scripts can then reach, and use a learning mode to build that allowlist automatically — contrasted with the "dangerously" permissive setting Chuck says is fine for home experiments with Omarchy but not for a company.
  • Omarchy ships with several security defaults already turned on. Full disk encryption is mandatory, the firewall blocks all inbound traffic except LocalSend, and SSH is disabled out of the box; sudo can also be configured to use fingerprint authentication instead of a password.
  • Every system update automatically creates a rollback snapshot via Btrfs. Chuck says Omarchy intentionally trails Arch's own updates by about a month to avoid breakage, and a snapshot can also be created manually and booted from directly with the omarchy snapshot command if an update goes wrong.
  • Chuck ran a local 4B model on a laptop with no GPU. The model, Qwen 3.5 Coder 4B, has a 128,000-token context window and ran on his Framework laptop with no discrete graphics card; it checked and changed his power profile and generated a working ASCII screensaver, all without touching the cloud — which he calls "a victory" for a model that size on CPU-only hardware.
Technology and FriendsKaren Lopez on Vector Search in SQL Databases [EPISODE 919]
  • Vector search matches on meaning, and that extra power costs money. Lopez, an independent data management and governance consultant, explains that a model turns text into embeddings, numeric representations of its concepts, and SQL Server ranks rows by similarity; more dimensions give finer granularity at higher cost, 1,536 being the common starting point, and any one embedding must fit SQL Server's roughly 8,000-byte row limit. She says not to use it for exact-word lookups that regular search handles.
  • SQL Server's older search options each carry their own cost. She lists LIKE, where a leading wildcard defeats the index and forces a table scan; Soundex, a phonetic letter-plus-three-digits code whose algorithm was first patented in 1918 and supports only English; full-text search (from "2016 or something like that"), which needs its word catalog kept updated; and the fuzzy search new in 2025, which handles word stems and other languages using about 50 words of surrounding context but no LLM.
  • Store embeddings in a separate table, not a column on the source row. Lopez's setup is: enable preview features (which many organizations' governance won't yet allow), create a credential, connect to an embedding model in Azure Foundry or a local one, then use the new vector data type and a DiskANN nearest-neighbour index; a separate table, unlike Microsoft Learn's examples, keeps embedding metadata and allows several model variants side by side. She expects initial loads to be embedded outside production, and calls hybrid search, combining vector similarity with ordinary filters like region or date, "the key to the future" because filtering first makes queries more affordable.
  • Every row must be embedded by the same model, version and service. Lopez compares mixing models to encrypting rows with different keys: the embeddings stop being comparable, so any change means regenerating all of them. For new and changed rows she notes the SQL Server blog recommends triggers, which DBAs dislike, and offers Azure Functions, change data capture or change event streaming instead; a casual research set might refresh nightly, while a symptoms-to-diagnosis database would need it instantly.
  • Model choice matters most, and her small demo model shows the limits. In her roughly 120-row NASA abstracts demo, a small model maps "moon" to lunar and Artemis papers, treats astronaut, cosmonaut and taikonaut alike, crosses languages and absorbs typos, but returned poor results for Cyrillic and Chinese spellings, which she attributes to its grounding in a handful of mostly Romance languages. Very large models of up to about 30,000 dimensions require chunking long documents into roughly 500-word rows, and she cites recent changes to Copilot business models as the lesson to size the model to the need.
Mr. GlobalIran Has Fired On Tankers 3 Times Since We Implemented The Tanker For Tanker Policy
  • Iran has repeatedly tested the tanker-for-tanker policy without consequence. He reports Iran fired on tankers on three separate occasions and hit 13 commercial vessels in August alone, while the US went roughly 30 days without retaliating after a single night of response.
  • The administration denied an Iranian missile attack it later partly confirmed. He reports Iran launched missiles at Jordan, the UAE, and possibly Kuwait, and that Jordan confirmed firing interceptors to shoot them down, even as officials maintained the attack "didn't happen" because nothing was hit.
  • Trump's inaction reads as tied to the diesel price record, not deterrence. He reports AAA put diesel at $5.85 a gallon, 3 cents above the prior all-time high, and argues that once the record broke, Trump "doesn't have anything to protect anymore."
  • A huge oil futures bet ahead of a holiday weekend looks suspicious. He reports someone bought 3,000 contracts worth $275 million wagering oil would rise, calls it possible market manipulation, and predicts a weekend escalation followed by a declared peace before Monday's open so the position could be cashed out — worth roughly $3 million per $1 rise in oil.
Network ChuckLinux Doesn’t Have to Be Ugly
  • DHH built Omarchy to look like hacker-movie computer aesthetics. He cites films like The Matrix as inspiration, saying Linux's window arrangement and color coordination don't have to look haphazard, and that he wants someone running Omarchy in public — even just a terminal or screensaver — to make onlookers wonder if he's "hacking."
Y CombinatorHard Tech Startups Are Built One Milestone at a Time
  • An electric aircraft startup built its case with physical proof. The founder recalls arriving at YC Demo Day with only a 3D-printed model, then using letters of interest from SAS and other Nordic airlines to justify building a 400-kilowatt motor roughly the size of a small jet engine, which in turn attracted pre-orders from United Airlines and further capital.
  • United Airlines became a customer after its email landed in spam. A team member says she was cleaning out the spam folder when she spotted the inbound message and flagged it, which led to showing United the motor built to replace its turboprop engines.
  • The aircraft's design deliberately looks conventional rather than futuristic. Unlike most electric aircraft startups, the company says it designed its plane to resemble the turboprops it's replacing, describing the approach as "hiding its Superman cape under the hood."
Dr. Roman YampolskiyMain Character Strata of Reality
  • Computer scientists mistake their human-scale world for all of reality. The speaker, unnamed in the clip, calls this the "main character strata of reality": the level of people and hand-sized tools, which feels like "where reality happens". Bioengineers working with interacting organelles, or ecologists watching living systems adapt and evolve, see intelligent and emergent behavior on other planes that this view dismisses as "dumb stuff". The clip ends without saying what follows for AI.
Dr. Roman YampolskiyDaniel Faggella Explores the Loneliness of AI Ethics in Academia
  • Faggella finds academics outside AI more open to asking "to what end". He says the "cosmic" people he knows at Google DeepMind find it lonely there, because the question of AI's ultimate purpose is "extremely rare" inside the lab. Academics in philosophy, neuroscience and biology are "wildly more willing" to say things will have to change and to ask how that change is shaped. The clip ends just as he says he has figured out why, without giving the reason.
Stock TokToo Many Stock Landmines. Here’s Where I’m Putting My Money Instead
  • Silver miners offer bigger leveraged upside than gold or silver alone. Measuring from the 2022 bottom, gold is up roughly 220% (peaking locally in March 2026) while silver ran about 375% over the same stretch, and his top four silver-miner picks are up between 442% and 1,248% — the logic being that if gold rises, silver and then silver miners rise more.
  • History suggests silver can keep rising well after a bull market tops. He cites silver continuing to climb for 42 days after the brief technical bear market of early 2025 and for 154 days following the 2008 crash as precedent for why silver positions may still have room to run even if stocks turn down; the 2000 dot-com case differed, with silver still gaining 65% while the S&P kept rising.
  • Frequent single-stock blowups pushed him away from individual stock picks. He points to Lululemon and UiPath both crashing -17% and Guidewire Software falling -22% despite beating earnings estimates and nearly doubling EPS as evidence that too many names are "landmines" right now, so he's rotating toward index funds and "recession resistant" assets like silver instead.
  • Silver isn't bear-market-safe, but he says it buys more time than stocks. He's explicit that silver and silver miners are not safe holdings in an actual bear market, but argues, based on the historical pattern above, that they typically give more runway before a downturn catches up to them; he is not currently calling a bear market.
Professor G MarketsLittle Tobacco Moment
  • The author argues Meta's $17.1 billion settlement with 47 states and D.C. over child-safety and addictive-design claims is being mischaracterized as a "Big Tobacco moment" when it barely dents the company. Meta admitted no wrongdoing, and its stock rose 1% on the news — on a $1.4 trillion market cap, he calculates, enough to cover the entire penalty.
  • The settlement leaves Meta's core engagement-maximizing design untouched while making the company stronger relative to rivals, the author contends. He cites NYU psychologist Jonathan Haidt's point that the recommendation algorithm "engineered to maximize young people's engagement" remains unchanged, and argues Meta used the deal to weaponize compliance as a competitive moat, since $5 billion of the fine is conditioned on YouTube and TikTok agreeing to similar terms — letting Meta then run ads positioning itself as the safety leader against competitors it helped bind to its own rules.
  • Scaled to Meta's size, the author argues, the fine functions as a speeding ticket rather than a deterrent. He illustrates by scaling the $17 billion penalty down to his own media company's finances, finding the equivalent hit would be about $256,000 a year spread over a decade — far too small, he argues, to outweigh the financial upside of continuing current practices.
  • The author draws a direct parallel to the 1998 tobacco settlement, arguing it didn't end Big Tobacco and this one won't end social media's harms either. That $206 billion settlement (paid over 25 years) coincided with major declines in U.S. smoking rates, but he notes cigarette makers innovated around it with e-cigarettes and expanded into low- and middle-income countries, where 80% of the world's tobacco users now live.
  • The author proposes three concrete reforms to give regulation real teeth: reforming Section 230 to impose liability for algorithmic amplification, breaking up tech monopolies under antitrust law, and taxing targeted digital advertising revenue. He argues the current legal shield, written for 1990s bulletin boards, lets platforms avoid responsibility for what their algorithms choose to promote, and that a progressive ad-revenue tax could push companies toward subscription models that don't depend on engagement and surveillance.
Professor G MarketsThe Week: When AI Escapes, and Wars Don't End
  • This post includes paid content that is not available publicly. See the original source.
All Things FinancialThe Eastern Financial (Non-Dollar) Network
  • The verdict: China can replicate the dollar system's pipes, not its water. It is ready now to carry more settlement in its own corridors — rung 1 (trade settlement) and rung 2 (a SWIFT bypass via mBridge) — but reserve substitution, rung 4, stays blocked by non-convertibility; a gold side-door, the Shanghai Gold Exchange's yuan-to-gold swap window, builds trust without conferring reserve status since it lets partners exit the yuan rather than hold it.
  • CIPS carries the volume; mBridge is the real bypass. CIPS moved roughly $19 trillion in 2024 (a record ~$178.5B/day in March 2026) yet still routes over 80% of its messaging through SWIFT; mBridge settles atomically with no SWIFT messaging at all but has only $55.5 billion in cumulative volume across five central banks — China, Hong Kong, Thailand, UAE, Saudi Arabia — 95% of it in e-CNY.
  • A durability score of 2.65 hinges on one weak factor. Settlement capacity (3.8) and redundancy (4.0) score well, but convertibility/autonomy scores just 1.8 because the renminbi isn't freely convertible and China won't lift capital controls; the composite would rise to 3.03 if that single factor were fixed — meaning Iran-sanctions pressure alone can't substitute for it.
  • Sanctioning a tier-1 Chinese bank would trigger "financial MAD." In the scenario, the bank loses dollar clearing, China retaliates via rare earths and Treasuries, and flows migrate to mBridge and gold — mutual deterrence that leaves the US running "death by a thousand cuts" against teapots and mid-tier banks while China escalates rare-earth curbs (a Dec-2024 ban, an Apr-2025 magnet restriction, a Jun-2026 entity ban, and an Aug-2026 sanction on the Responsible Business Alliance and five other US supply-chain auditors) ahead of the Nov-2025 truce's expiry around November 2026.
All Things FinancialIran Economic D-Day sanctions. StableCoin, treasuries, e-CNY, gold & REM 84-2026
  • He reads Amazon's weak bond demand as a warning for CoreWeave. Amazon's $25 billion issuance got just 1.6 times coverage against buyers who typically want two-to-four times; if a company of Amazon's size can't clear that bar, he argues, CoreWeave — issuing next — will be "highly challenged" to hit even 1.5 times.
  • He flags a contradiction between rare-earth demands and Iran sanctions. A 100% tariff threat on any country trading with Iran covers China itself; Chinese rare-earth firms have already halted select US exports in response, and he says the sole open issue for Xi's September 24 meeting is a US demand for guaranteed rare-earth supply — set against recent US pressure on Venezuela's oil, a push for Venezuela to default on $10 billion owed to China, and a $2 billion Panama suit.
  • He flags a possible gold-backed e-CNY as the network's wildcard. He restates the underlying verdict — China can build the settlement pipes but not reserve-currency "water" — and notes its gold-vault buildout already lets partners swap yuan for gold without making the renminbi convertible; he asks whether an actual gold-backed digital yuan would break that barrier, calling it a five-year, not ten-year, question.
Mr. GlobalA Country That Can't Be Told The Truth Can't Be Saved By It.
  • Admitting a lie and backing it anyway is a loyalty test. He argues Trump supporters who say "I know he lies, I don't care" have stopped judging him as a truth-teller and started judging him as a weapon that "hurts the people I want to hurt," making the lie itself the appeal rather than a flaw.
  • Defending an obvious falsehood proves loyalty precisely because it costs something. He invokes Hannah Arendt's argument that propaganda's goal isn't belief in the lie but destruction of the public's ability to tell truth from falsehood, leaving people too exhausted and cynical to do anything but pick a side.
  • A government facing no penalty has no incentive to solve problems. He argues that once voters stop demanding truth, "the claim of a solution pays exactly the same as the solution," so officials get rewarded for announcements instead of results.
  • Tolerated lies will leave the public unable to spot real emergencies. He predicts that when a genuine crisis requires the government to tell the country something true and urgent, the message will fail because people have been trained to assume "the podium is just selling."
  • Tolerating dishonesty from your side sets a precedent for the office. He warns that future presidents of any party inherit a public that has already agreed dishonesty is acceptable, leaving today's defenders with no argument when the same lie is told by someone they oppose.
GeoHussarHow did Portugal get its name? #portugal🇵🇹 #History #geopolitics #porto #portuscale
  • Portugal's name comes from a Roman port town, not a native name. Before Rome, Greeks called the region Ophiussa and the local Lusitanian tribe called it Lusitânia; the name Portugal instead descends from Portus Cale, a Roman-era port town at the mouth of the Douro River.
  • The region changed hands repeatedly before Portugal's name stuck. Phoenicians settled the Tagus River around 800–900 BC, the Carthaginians took over from the 3rd to 5th century BC, and Rome conquered the area by roughly 190 BC–14 AD. After Rome fell, the Suebi and Visigoths took the peninsula and the name Portus Cale fell into disuse; Arab rulers later called the region Al-Andalus or Al-Gharb.
  • The Reconquista's completion cemented Portugal as a distinct kingdom. Starting around the 800s, roughly a century after Spain's Reconquista began, Portuguese forces retook the town of Portus Cale and renamed the surrounding county Portucale. Portugal finished its Reconquista before Spain did, and by 1139 the count of Portucale, Afonso Henriques, declared himself king of Portugal.
Y CombinatorOpen Models Change The Economics of AI
synopsis

Jeff Morgan, CEO of Ollama, argues that open models have crossed in 2026 from hobbyist tooling into an enterprise default — driven first by coding agents and now by OpenClaw-style autonomous work — and that the real bottleneck has moved up the stack, from the models themselves to the "curation" work of matching a model to a harness, an inference engine, and hardware fast enough to make day-zero releases usable. He ends holding that most enterprise tokens will eventually run on open models even though the biggest dollars still flow to closed frontier labs, with open and closed, and local and cloud, settling into a permanent hybrid rather than one replacing the other.

The shift to open. Ollama's own numbers frame the claim: 9 million developers, 178,000 GitHub stars, and use by 85% of the Fortune 500. Per-developer token usage on Ollama's cloud grew roughly 5x when open models became capable coding-agent backbones early in the year, then jumped again with OpenClaw-style agents in April as context windows expanded from 128K to over a million tokens — a combined roughly 150x growth in cloud usage since January. Morgan cites AT&T as a named example already routing 40% of its tokens to open models, mostly through coding agents, with cost as the entry point and customization the longer-term draw.

Packaging a model is now the hard part. Morgan compares Ollama's role to an operating system: getting a new model release working well requires an inference engine tuned for it, a harness or SDK built to use its new capabilities, and enough capacity across hardware providers — Nvidia, Apple Silicon — to handle a launch-day surge. Most of that work, he says, happens in the last 24 hours before release and is "generally a fire drill." He expects pieces of today's harnesses, such as memory and credential management, to unbundle into their own best-of-breed products as the stack matures, following a pattern he says already played out in cloud computing.

Where the money and the tokens diverge. Morgan's steady-state forecast is that 80-90% of an enterprise's tokens will run on open models, but that this won't mean 80-90% of the budget — falling per-token costs mean open models could take as little as 10-20% of spend even while carrying most of the traffic, with frontier closed models reserved for a shrinking set of the hardest problems, decided by a router or "coordination" layer sitting above the model layer. He extends the same logic to hardware: local models on next-generation Apple and Nvidia desktop chips, running 20-120-billion-parameter models, will handle easier, latency-sensitive tasks, while coding agents needing the largest models stay in the cloud for now — though he expects that workload to migrate back to local hardware as it catches up, much as it did in the earlier PC and cloud eras.

Origin, security, and Ollama's own slow start. On the geopolitics of Chinese-origin models dominating cloud usage, Morgan argues the operative question is security and where a model runs, not where it was trained, and says he knows of no confirmed case of a deliberately compromised, "Manchurian candidate" model, comparing the risk to the ordinary open-source supply-chain exposure enterprise security teams already manage. He closes by tracing Ollama's own path to this point: a security-tooling startup that spent two "lost" years failing to find its niche, a two-week pivot to build a local LLM runner timed to the Llama 2 release, product-market fit within weeks of launch, and a further two years before monetizing — a delay he frames as deliberate, waiting for open models to reach the point, arrived at only this year with coding agents, where they could finally take on the hardest problems that had been the exclusive domain of closed models.

  • Enterprises adopt open models for cost, but stay for control. Ollama CEO Jeff Morgan says cost is "by far the largest pain point" open models solve, but the deeper draw is customization; he cites a report that AT&T has already shifted 40% of its token consumption to open models, mostly for coding agents.
  • Agent-style products, not chat, drove open models' explosive growth this year. Morgan says per-developer token usage on Ollama's cloud jumped roughly 5x when OpenClaw-style agents launched in April, pushed context windows from 128K to over a million tokens, and grew Ollama's overall cloud usage 150x since the start of the year.
  • Most enterprise tokens will run on open models, he predicts. Morgan forecasts 80-90% of a business's tokens flowing through open models as their cost keeps falling, while frontier closed models still take the largest share of spend, reserved for the hardest tasks a router or "coordination layer" escalates to.
  • Security screening, not model origin, is what should decide adoption, he argues. Morgan says the real blocker to using Chinese-origin models like GLM or DeepSeek is security and where they run, not where they were trained, and that he isn't aware of any confirmed case of a "Manchurian candidate" model booby-trapped to cause harm.
Dr. Jon KrohnIn Case You Missed It in August 2026 (Ep. 1024 with Jon Krohn)
synopsis

August's compilation gathers four clips that share one argument: AI pays off when organisations pick measurable problems and keep humans, hard guarantees and shared definitions in the loop. Buying tools and counting tokens does not do it.

Picking measurable problems. MongoDB's Pete Johnson explains the gap between the 83% of organisations with AI committees and metrics and the 19% seeing matching ROI. The companies he heard at an NRF dinner in January reporting real ROI used employee-facing, human-in-the-loop agents, because existing performance metrics make the gains attributable. He says token-maxing rose and fell within about eight weeks, after Uber reportedly used its 2026 token budget in 13 weeks. His preferred measure for coding agents is the time from idea to production.

Agents plus solvers. A Gurobi strategist argues that LLMs can confidently ignore a hard constraint, while a mathematical optimisation model guarantees it, and can explain trade-offs such as siting a plant in Atlanta rather than Baltimore. In his view agents should help formulate the problem and then hand it to a solver.

Budgeting for habits. A former Microsoft AI transformation leader proposes spending 10% of the AI budget on tools, 20% on execution and 70% on upskilling. She cites a Stanford finding that only about 8–9% of use cases reach production, and expects a J-curve of six to eight months. Her Claude skills break blog writing into steps, with a human review in the middle.

Shared meaning. dbt Labs' Tristan Handy says the semantic layer, which stores agreed metric definitions, matters more now that agents ask the questions. Without it, agents spend tokens re-deriving each metric or get it wrong. He places ontologies one layer higher.

  • AI ROI comes from measurable, employee-facing, human-in-the-loop use cases. MongoDB's field CTO of AI, Pete Johnson, responds to the host's figures that 83% of organisations have AI steering committees and success metrics but only 19% see matching ROI. He says the first companies he heard reporting ROI from production agents, at an executive dinner during NRF in January, chose employee-facing, non-autonomous use cases, because the data-security bar is lower and existing performance metrics show the agent's effect. He argues coding agents should be judged by idea-to-production speed, not lines of code or tokens.
  • Token-maxing showed how fast AI's hype cycles burn out. Johnson says token-maxing peaked when Nvidia's CEO mentioned it at GTC in March and was dead about eight weeks later after the Uber story, which he says saw its 2026 token budget used in 13 weeks; Meta also ran a token scoreboard. He contrasts this with MCP, which Anthropic released in Thanksgiving week 2024 and competitors had adopted by March.
  • Agents should formulate optimisation problems, while solvers solve them. Gurobi's manager of decision intelligence strategy argues LLMs can confidently miss a hard constraint, such as an environmental rule carrying millions in fines, which a mathematical model guarantees it will not violate. The host adds that an author's experiments found frontier models skipping a mandated tool call a single-digit percentage of the time. The guest's proposed division of labour: agents help define the problem and write the formulation, then call a solver such as Gurobi, whose answers can be interrogated ("why Atlanta, not Baltimore?").
  • Spend 70% of the AI budget on upskilling. A former Microsoft AI transformation director proposes a 10-20-70 split: 10% on tools, 20% on execution and 70% on education. She cites a Stanford report that only about 8–9% of AI use cases are in real production, and says habits take six to eight months of a J-curve to form. She also describes structuring Claude skills as explicit human-style subtasks (research, a human review step, storytelling, format) to avoid slop.
  • A semantic layer stops agents re-deriving metrics at every turn. dbt Labs CEO Tristan Handy says semantic layers solve an organisational problem of big firms, such as Siemens with 2,000 data engineers serving 300,000 employees: storing agreed definitions of how to measure things. Without them, he says, agents either burn many tokens re-deriving a metric or get it wrong; he places ontologies, which model processes and causation, a layer above, suited to high-value domains such as genetics and defence.
Professor G MarketsTyler Cowen: The AI Bears Are Asking The Wrong Questions
  • This post includes paid content that is not available publicly. See the original source.
Mr. GlobalThe Price Of Diesel Just Set An All-Time High For Donald Trump
  • Diesel broke its all-time national price record. Citing Patrick De Haan of GasBuddy, he reports the national average reached $5.82 a gallon, beating the previous record of $5.81 set June 17, 2022, and argues Trump now "owns" it despite blaming Biden for high fuel prices before.
  • The annual average diesel record is also now at risk. He reports 2026 is running just below 2022's record annual average of $4.99 a gallon, but diesel would need to average $5.25 for the rest of the year to avoid breaking it — and today's $5.82 is already $0.57–0.58 above that pace.
  • Diesel costs feed into nearly every consumer price in the economy. He argues rising diesel raises trucking, freight, farming, and construction costs that work through the economy with a lag, and predicts the next inflation report will show a bigger increase than recent ones.
  • Record-low distillate inventories are the supply-side driver. Citing EIA data, he reports stocks of diesel, heating oil, kerosene, and jet fuel are at their lowest ever, pointing to reduced Ukrainian strikes on Russian refineries (some now reopening) offset by continued Houthi attacks preventing Saudi Arabia's Jizan refinery from restarting.
  • He expects the price rise to taper but calls the outlook uncertain. He notes wholesale diesel edged down today and guesses prices could climb for a few more days before leveling off, but says continued Iranian attacks make the path impossible to call.
Professor G MarketsWhy Trump’s Push for AI Data Centers is Backfiring
  • This post includes paid content that is not available publicly. See the original source.
Professor G MarketsThe AI Revolution Isn’t What Silicon Valley Promised — with Josh Tyrangiel
  • This post includes paid content that is not available publicly. See the original source.
Silver AllianceCopper and silver demand to skyrocket cuz of recent energy grid emergency order
  • A cybersecurity order, not market demand, is driving this metals thesis. An executive order issued last week requires foreign-made power grid equipment — the video names China — to be replaced with domestic equipment over the coming years, creating a policy-driven demand source for copper and silver.
  • Grid transformer replacement could consume over a million tons of copper yearly. Large power transformers use about 100 tons of copper each, and roughly 700 large units plus 7 million smaller neighborhood transformers need replacing within five years; the IEA figure cited is 1.3 million tons of copper a year, competing with AI data centers and EV manufacturing for supply.
  • Substation silver demand is modest per unit but adds up nationwide. Load tap changers inside transformers use silver-plated contacts, and installing digital grid-monitoring and cyberattack-detection hardware (described as a few ounces per control panel) across thousands of substations is put by the Silver Institute at 8 to 10 million new ounces of silver a year, on top of existing solar, AI server, automotive and medical demand.
murat.energyPhysical gas is not cheap gas! #shorts
  • US LNG export growth is accelerating and set to keep climbing. First-half shipments jumped 23% versus 2025 to an average of 17.4 billion cubic feet per day, driven mainly by Plaquemines and Corpus Christi Stage 3, with Golden Pass now started and adding more through 2027; total US capacity already exceeds 20 billion cubic feet per day, with Port Arthur and Rio Grande under construction bringing more than 38 billion cubic feet per day either running or being built, and exports are expected to step up toward 18-19 billion early next year.

  • Most of that new export growth went to Asia, not Europe. In the first half of the year Asia absorbed more than 2 billion cubic feet per day of the increase, while Europe saw almost no gain.

  • EU officials say there's no immediate risk to physical supply. The bloc's gas coordination group points to higher import capacity than in 2020-22, structurally lower demand, and independence from Russian pipeline gas as reasons more cargoes can be pulled in if needed — a claim the host calls "technically true."

  • The host's bottom line: physical supply is covered, cheap gas isn't. European storage sits at only 65% versus a normal 80% for this time of year, with Germany and the Netherlands barely half full, and because Qatari supply remains impaired, every cold snap will force European utilities onto the spot market bidding against Asian buyers for flexible US cargoes.

Loss DogBitcoin Strategy: Net Long & Anticipating Dips #bitcoin #bitcointrading #trading #wearelossdog
  • He trades Bitcoin purely on price, treating it as an uncorrelated asset with no macro view. He says he sold around the "70-something" level anticipating a drop back into the 60s, and cut his position from about 3% to 1% of his book while remaining net long Bitcoin, Ethereum, Solana and some Ripple.
  • The hosts tease a forthcoming dollar-pegged stablecoin, distinct from a standard cryptocurrency. It's described as pegged one-to-one to the US dollar and backed by Treasury holdings, with the claim it carries no principal risk — a claim about an unreleased product, not a demonstrated fact.
Loss DogInverse Relationships, Corrections, Crypto & Calendar Spreads 9.03 One Lucky Dog
synopsis

Tony and Scott open a broadly green tape — stocks, gold, oil, silver and bonds all higher — and read it as a market unworried about Friday's jobs report; once Tom joins, the theme becomes that nothing on the tape is concerning, but the near-total absence of volatility is. The show moves between viewer mail on trading mechanics, Tony's trade ideas, caller questions, and a genuine disagreement over whether calendar spreads carry any structural edge, closing on mortgage rates and crypto as further evidence that most market moves are just noise.

Stocks, bonds and correlation. A viewer asks whether a correction is needed to restore the historic inverse relationship between stocks and bonds, last seen decades ago before low rates pushed the two together. Tony expects the old inverse pattern — flight to quality in a selloff — to return now that rates aren't artificially suppressed, though he won't say when. Scott disagrees, arguing risk-on/risk-off behavior now runs through crowded trades like chip stocks rather than bonds, so reversion isn't guaranteed.

Mechanics mail. The hosts endorse the long call / short put "synthetic stock" substitute (both 50 delta, 100 net long deltas) as capital-efficient, especially in an IRA, though it needs active rolling unlike actual shares. On skew, IV rank is the simplest tell — high IVR with call skew on individual names, put skew on ETFs — and a call trading roughly twice the equidistant put catches Scott's eye. Both insist position size stay consistent regardless of conviction, though Scott admits breaking that rule when a credit improved intraday. They marvel that a single Meta fine ($16–17 billion) reads as one month's revenue, and expect IPO valuations to keep resetting higher. Tony calls trading a game he enjoys, unlike gambling sites that sell speculation with odds stacked against the player. On startups, both say the founder matters as much as the concept, citing a health-tech founder "going to kill himself to make it work."

Tony's trades of the week. Hood (IVR 51) and Tesla both moved too far intraday to fit the setups he'd prepared. In Apple he keeps a 295/355 strangle roughly unchanged, IVR near 44–46. The featured trade is Boeing, "a dog with fleas": a back-ratio spread (long two 210 in-the-money calls at 70 delta, short one 200 call at 50 delta) for about a $1,950 debit — a near defined-risk stock substitute. Selling a further 230 call would cap upside but cut the debit to roughly $1,550. Tony flags this as his least favorite of the four, since it's only a 50/50 shot.

Caller questions. Errol asks whether sizing should track buying-power reduction or max loss in a margin account; both answer buying power, unambiguously. A caller from San Diego asks whether a short put should be rolled exactly at the mechanical day-count or opportunistically on a down day; the hosts say a day or two early or late is immaterial so long as it isn't dramatically off schedule. Brian asks why NASDAQ (NQ) options quote such wide spreads; the hosts explain market makers widen quotes deliberately because the underlying can move 50–100 handles in seconds, making tight autoquoted markets too easy to pick off — patience and working the mid is the workaround.

Tom's trades and product notes. Tom describes shorting crude oil futures near 92.24 and rolling bond and note positions to October ahead of 21 days to expiration, cutting risk roughly 30% ahead of the jobs report. He calls a proposed Nvidia LEAPS bull call spread (225/285, September 2027) reasonable for a bull; says SPY, XSP and MES are functionally identical products; and argues iron condors work best at market extremes rather than in quiet, range-bound stretches. Asked about shorting Snow after a 22% earnings pop, both say wait two to three days rather than fade the move. A caller's question about the show's agentic AI tools becomes a product pitch: tools meant to catch mechanical mistakes before an order routes — illiquid strikes, mismatched durations, selling premium when IVR is too low.

Jobs report and complacency. With Friday's payrolls forecast at just 50,000 — a bar Tony notes has fallen from roughly 200,000–250,000 in past years for "zero growth" — a poll shows 54% expecting a beat. Tom argues the report itself isn't the worry; today's joint stock-and-bond rally already signals the market isn't scared. What concerns him is how little the VIX has moved on recent down days — "a little scary" — and upside surprises in bonds, yen and gold worry him more than a soft print. He frames the Fed's next move as mechanical: bonds below 108 argues for a hike, a rally back to 109–110 gives room to pause.

Side arguments. On congressional stock trading, Tony argues politicians are too unsophisticated to meaningfully exploit inside information and should be free to trade; Scott counters that officials with committee access to material information should be barred, though he agrees the optics, not the substance, are the real problem. On Bitcoin and Ethereum, both call the day's rebound essentially random, driven by price rather than geopolitics or oil; Tom treats his crypto holdings as a permanently uncorrelated asset class traded on level alone. The 30-year mortgage near 6.75% prompts a similar read on housing: Tony calls today's over-asking bidding wars a possible warning sign, and argues from decades in Chicago real estate that stocks have consistently outperformed housing for him.

Calendar and diagonal spreads. The show closes on a genuine disagreement: Tony argues calendars carry no structural edge because retail traders can only buy them (selling ties up too much capital), leaving them "priced to perfection"; Tom flatly rejects this, insisting there's a real edge when front-month implied volatility sits at or above back-month volatility. Both agree calendars are useful mainly as a learning tool, hard to convert into other structures beyond skipping a month, and that sizing them in expensive underlyings like SPX — targeting a 10–20% profit — beats a spread so cheap a "win" is a few cents.

  • One host says stocks and bonds will "categorically" resume their historic inverse relationship. He argues today's positive correlation came from artificially low rates the Fed is now unwinding, and a real correction would trigger a flight to quality into bonds. The other calls that "completely random," arguing risk-on/risk-off flows now go elsewhere, like chip stocks, since there hasn't been a sustained downturn to test it.
  • Both hosts call Bitcoin and Ethereum's moves essentially random, unrelated to geopolitics or macro events. They describe crypto as a non-correlated asset class traded purely on price — one host trimmed a position from roughly 3% to 1% of holdings while staying net long Bitcoin, Ethereum, Solana and Ripple, which traded near $81,000 and $2,460 respectively during the show.
  • Rising 30-year mortgage rates, near 6.75%, are framed as a bond-market repricing rather than a housing-specific problem. One host argues buyers who've known only near-zero rates for two decades wrongly see 5–7% as abnormal when it's closer to the historical norm; he also says 50 years of Chicago real estate never made him money versus stronger stock-market returns.
  • The hosts split sharply on whether calendar spreads carry a genuine structural edge. One insists there is one, "150%," when front-month implied volatility is equal to or higher than back-month; the other argues there is none, since retail traders can only buy (not sell) calendars for capital reasons, pricing them "to perfection" — with any profit coming from execution, not edge.
Mr. GlobalIran Escalates As Reports Circulate That Trump Wants Out
  • The Wall Street Journal reported Trump may declare victory and leave Iran. He notes he is skeptical because this White House sometimes plants false leaks that are later debunked, but says Iran's immediate large-scale missile and drone attack on Kuwait makes this report harder to dismiss.
  • Withdrawing now could teach Iran that gas prices force American exits. He argues that if Iran widened the war, raised oil and diesel prices, and made Americans suffer only to watch the US leave, it would create "a pricing mechanism for American resolve" that adversaries could exploit more efficiently in future conflicts.
  • Iran again tested the tanker-for-tanker policy with cruise missile strikes. He reports this came alongside the Kuwait attack, with cruise missiles fired at tankers in the Strait of Hormuz, and reads it as Iran probing whether the US will actually follow through on a retaliation policy it has not enforced so far.
  • Declaring victory doesn't end a war only Iran can end. He argues Gulf Coast states and Israel, which joined the US campaign against Iran, would remain exposed to continued attacks even after an American withdrawal.
  • He ties a possible US exit to avoiding a Biden-topping diesel record. Even so, he concludes that despite the risk of rewarding Iran's escalation, withdrawal may still be the least-bad option given how bad the overall situation in the Middle East has become.
All Things FinancialCoreWeave (CRWV) — A Financial Accident In The Making
  • CoreWeave's interest expense already exceeds its operating income. Debt grew from $2B in 2023 to $35B by Q2 2026 at a 9.0% weighted-average rate, with quarterly interest hitting $536M ($2.14B annualized, climbing toward $3B) against guided 2026 operating income of only $960M-$1.15B — a coverage ratio under 1.0x — while free cash flow has been negative every year since 2022, funding interest with new debt.
  • The real trigger is the bond demand ratio, not a calendar date. Bond yields widened from 9.0% (Jul 2025) to 9.75% (Apr 2026) to 10.99% secondary (Aug 2026) and a reported-but-unverified 13.1% in September; the author says a next-deal demand ratio below 1.5x oversubscription would signal dealers absorbing the slack, with the $8.5B DDTL 4.0 facility's June 2027 termination (and a claimed, unverified $4.1B December 2026 maturity) as the actual forcing dates.
  • Insider selling by Magnetar and the CEO reads as a credit-trajectory signal. Magnetar Financial, a 10% owner and pre-IPO backer known for structured-credit expertise, sold ~$478.8M across three sessions in August 2026 just before a drawdown, and CEO Michael Intrator sold $32.87M; the author treats this, alongside the widening bond yields, as evidence insiders are cutting exposure ahead of stress, while Nvidia's dual role as supplier, shareholder ($2B stake), and financier (via the DDTL facility) creates a floor but not an unlimited one.
  • A distressed-debt exchange, not Chapter 11, is the most likely 2026-2027 outcome. The BBC framework assigns ~45% probability to a creditor-negotiated exchange (70-90% equity dilution, no filing), ~35% to a stalking-horse acquisition by Microsoft, Meta, or Amazon at 30-50 cents on the dollar, and ~20% to an Nvidia-engineered bridge; recovery is expected to bifurcate within the same estate, 70-90 cents for the durable civil/power layer versus 20-40 cents for depreciating GPU compute.
All Things FinancialCoreWeave grow now pay later strategy runs out of free cash flow to pay interest 8-4-2026
  • CoreWeave's debt has ballooned faster than its ability to service it. Total debt rose from $2B in 2023 to $35B by Q2 2026 (roughly $30B long-term, over $10B raised in Q2 alone) at a 9% average coupon, with quarterly interest near $536M against still-negative free cash flow; he adds that the GPU collateral backing this debt depreciates roughly 30%/year, so within three years liabilities could run ~15x current assets.
  • Two dates could force the crisis: December 2026 or June 2027. He flags a "plausible" $4.1B maturity in December 2026 and the $8.5B DDTL facility's June 2027 draw deadline as the two forcing events, but says the real leading indicator is the next CoreWeave bond's demand ratio — a print below 1.5x oversubscription would confirm the company can no longer roll debt to cover interest.
  • Magnetar's ~$500M exit reads as the more telling insider signal. He waves off CEO Michael Intrator's sale as routine but flags Magnetar Financial — CoreWeave's original financial backer — offloading roughly $500 million of the debt as meaningful; he ranks a distressed exchange (~45% probability) as the likeliest bankruptcy-phase outcome, ahead of a Microsoft/Meta/Amazon stalking-horse acquisition (~35%) or an Nvidia-engineered rescue (~20%), and notes his stress-timeline call from four to five months ago is "so far" holding up.
Y CombinatorPaul Graham On Startups, Ambition, and Great Founders
  • Today's YC startups are more ambitious than before. Graham dismisses claims that YC has "jumped the shark," noting critics have said this since 2008; he contrasts old examples like Reddit with current-batch startups building literal orbital cargo delivery and on-demand cancer research — the latter founded by the same team that helped a past YC founder fight his own cancer.
  • New incumbents win by waiting, not attacking head-on. Graham says his 2012 "frighteningly ambitious" essay predicted this is how a new Google would emerge, and argues OpenAI did exactly that — once he stopped using web search for information, the search business model was already obsolete.
  • Fear of disaster drives founders day to day, not wealth. Graham says the immediate motivator is avoiding failure — the server crashing, looking like a fool — and that he defines a "formidable" founder as one who reliably gets what they want, which is exactly why it aligns investor and founder interests.
  • AI arrived backwards from what he expected. Graham says he expected AI to build up from insect-level competence to human intelligence, but instead got a "bullshitting undergraduate" that was human-like yet unreliable from the start, producing a "jagged frontier" where it solves hard math but fumbles simple facts; he adds that shipping pace remains the best predictor of a startup's success, with AI's real effect being that compute is now a startup's biggest cost.
Network ChuckDEFCON Is Not What You Think
  • DEF CON runs a public "Wall of Sheep" shaming insecure Wi-Fi use. At Packet Hacking Village, it displays a live scoreboard of intercepted usernames, domains and passwords from attendees who connect carelessly; Chuck's advice is to skip the conference Wi-Fi entirely and turn off Bluetooth before going in.
  • DEF CON has grown into the world's largest hacker conference. It began as a small hacker gathering in the 1990s; this year, DEF CON 34, drew roughly 30,000 attendees — researchers, red teamers, and hardware hackers — to Las Vegas, spread across themed "villages" for lockpicking, cars and biohacking, among others.
  • Chuck streamed his DEF CON footage home without opening a single port. The segment is sponsored by TwinGate; he set up a Mac Studio inside Packet Hacking Village and used TwinGate's zero-trust access to relay recordings to his home studio over the public DEF CON network — a tool he says he uses daily for remote access.
  • Chuck says DEF CON's real value is its people, not the talks. He points to conversations with hackers he met there, like Vega, a principal consultant at Sophos who pentests networks, web apps, APIs and medical devices — the latter, he says, often built fast with security bolted on afterward, leaving gaps — and says much of the real networking happens at unofficial post-conference parties, not the scheduled talks.
Stock TokThe Market Pulled Back. I’m Buying These 2 Next.
  • This post includes paid content that is not available publicly. See the original source.
Master InvestorThe Fed's Next Move Could Decide the Market
  • Sonders expects the Fed under Kevin Warsh to hike rates 25 basis points. Knowing Warsh for 23 years, she argues his Jackson Hole speech signaled that fighting inflation will take precedence over the Treasury's more growth-friendly agenda.
  • She says what matters most for stocks is the speed of any Fed move, not the move itself. Slow tightening cycles have produced over 10% equity returns in the following year historically, versus roughly -4% after fast hiking cycles, against a baseline average of about 4.5%.
Professor G MarketsBonds Are Warning Of A Global Inflation Crisis
  • This post includes paid content that is not available publicly. See the original source.
Technology and FriendsAshley Janelle on UX and AI [EPISODE 918]
  • Hiring managers now expect UX designers to work with AI. Janelle, who founded and runs a UX boot camp, says companies want to know which AI tools a candidate uses in research, strategy and design to work efficiently, and that designers who have been inside one job for three or four years will find the market has changed.
  • Every UX solution should be checked for AI that improves it. She has students write "AI" with "better, smarter, faster" beneath it and test each against their finished design, citing the AI summary TikTok shows on a live stream before you open it, and search suggestions that complete a partial query.
  • Prompt-to-app tools suit early concepts, not final design or learning. Figma, free and used in her program for seven years, remains the main tool; she steers students away from Lovable, Figma Make and chatbot design tools such as ChatGPT's and Claude's, because UX rests on research into human psychology that a chatbot cannot connect even when fed it. Giard draws his own parallel to generated code, which he uses but does not fully trust without understanding it.
  • AI's real gain is showing users rough concepts much earlier. Janelle says these tools let her generate early wireframes from a set of parameters before research is done, so users have something to react to at the start rather than only after strategy and design; she notes this was possible before, but is now much easier.
  • Leaner teams in 2026 reward generalists over specialists. In her view AI is an enhancement rather than a replacement, but because it speeds work, companies now expect one designer to cover research, management, design and content roles that used to be separate; she urges designers to learn every part of UX and to include AI in the solutions in their portfolios.
Master InvestorTwo Stocks Drive One Third of S&P Earnings Growth
  • Sonders highlights how narrow S&P 500 earnings growth has become. Nvidia alone accounts for 18% of expected 2026 S&P earnings growth, and together with Micron that reaches 32%, while the top 10 contributors are responsible for two-thirds of total growth.
  • She argues the market cares more about the direction of travel than the absolute growth rate. A deceleration from 60% to 30% earnings growth is still "phenomenal," she says, but it's the inflection point and rate of change that can move sentiment — and she believes that effect is already starting to show up.
Silver AllianceJoin the Alliance!
  • The clip carries no market content, only a call to action. The full caption text is "Okay, let's go," consistent with a short promotional teaser titled "Join the Alliance!" rather than a piece of precious-metals commentary.
American PowerDid Trump Get Played on Venezuela’s Oil Deal?
synopsis

This panel episode treats Trump's new Venezuela oil deal as an arrangement that cannot deliver on its stated purpose and is unlikely to survive contact with Venezuelan politics, while a parallel flare-up with Iran exposes how depleted US missile stockpiles and military leadership have become. Both stories, the hosts argue, reflect a White House making high-stakes decisions without expertise, realistic planning, or an accurate read on its counterparts.

Why the deal can't refill the SPR. Matt explains oil stored in the Strategic Petroleum Reserve's salt caverns must fall between 30 and 45 degrees API gravity, while Orinoco belt crude sits around 8 — four times heavier than the reserve's own limit — and cannot be diluted for long-term storage the way it can for refining. Converting it would require Venezuela to build an upgrading plant costing $10 to 12 billion and taking five to seven years, something it currently lacks the capacity to do, making Trump's stated purpose for the deal effectively impossible. Matt adds that Venezuela's widely cited 303-billion-barrel reserve figure traces back to Hugo Chavez inflating his "recovery factor" assumptions from an accurate 6% to a claimed 20%, a number the USGS accepted and outlets from Reuters to NPR still repeat, even though Venezuela's own 10-year production target is only 1.5 million barrels a day.

A Chavista in the driver's seat. Chad argues Trump misjudges Venezuela's acting president, Delcy Rodriguez, as someone he can control, when her 30-year record — her father founded the Socialist League, she sat in Chavez's inner circle from 2006, and Maduro called her his "tigress" — makes her anything but a Western partner. He likens the current deal to the 1990s Apertura Petrolera opening, which collapsed in 2007 when Chavez forced foreign oil companies to hand over majority stakes; ExxonMobil and ConocoPhillips won international arbitration worth $13 to 15 billion but have recovered only about $700 million of it. Chad predicts Rodriguez will let American companies build the infrastructure, then reclaim it once Trump is gone or weakened, much as Chavez once did.

Numbers that make the deal "uninvestable." Matt lays out royalty terms above 30%, compared with the 12.5% paid on US federal land, pushing the government's total take toward 40% of revenue against an estimated $80-to-$90-a-barrel break-even cost for new Venezuelan production — roughly $100 billion just to start and $180 to $250 billion for full development. He notes ExxonMobil's CEO has already called the terms uninvestable, which is why the deal runs through an obscure vehicle and a Pentagon defense-loan program rather than a major oil company, leaving both hosts expecting the arrangement to eventually collapse and Venezuela to seize whatever gets built.

Iran tests American resolve as munitions run thin. Chad reports Iran answered a limited US strike on two rocket launchers on Loraq Island with an estimated 30 to 35 ballistic missiles fired from nine locations and drone strikes on the UAE — a response he reads as deliberate, since Iran had gone 30 days without retaliating for tanker attacks before choosing this moment to escalate. He connects this to a leaked Pentagon report describing US precision-strike and interceptor inventories as critically low, and to a wave of resignations, including Army Secretary Dan Driscoll and the Southcom commander over strikes on boats, alongside 98 of 196 US ambassador posts sitting vacant — a "brain drain" he compares to the one that preceded the decline of Russia's own military.

  • Venezuela's heavy crude cannot go into the Strategic Petroleum Reserve. Matt Randolph says oil stored in the SPR's salt caverns must fall between 30 and 45 degrees API gravity, while Orinoco belt crude is around 8, far outside that range, and Venezuela lacks the $10 to 12 billion, five-to-seven-year upgrading plant it would take to convert it, making Trump's claim that the deal will refill the reserve effectively false.
  • Delcy Rodriguez is a lifelong Chavista, not a moderate. Chad Scott says Trump believes he can control Venezuela's acting president, but traces her 30-year record inside Chavez and Maduro's inner circle — her father founded Venezuela's Socialist League, and Maduro called her his "tigress" — and argues Trump is repeating the 1990s Apertura Petrolera opening, which ended in 2007 when Chavez nationalized foreign oil firms owed roughly $13 to 15 billion in arbitration, of which only about $700 million was ever recovered.
  • The Venezuela oil deal's financial terms make it "uninvestable." Matt Randolph says the arrangement carries a royalty over 30%, versus the 12.5% paid on US federal land, pushing the government's total take toward 40% of revenue against an $80-to-$90-a-barrel break-even cost for new production, which is why Exxon's CEO called it uninvestable and why the deal instead runs through a Pentagon defense-loan program and an obscure business vehicle rather than major oil companies.
  • Iran's massively disproportionate strike was a deliberate test of US resolve. Chad Scott says Iran answered a limited US strike on two rocket launchers with an estimated 30 to 35 ballistic missiles fired from nine locations plus drone strikes on the UAE, after 30 days of not retaliating for tanker attacks, at a moment when a leaked Pentagon report shows US precision-strike and interceptor stockpiles running critically low.
Professor G MarketsREPLAY: Is the USPS Helping Trump Rig the Election?
synopsis

Jessica Tarlov and Aaron Parnas move through a week where, in their telling, pressure is visibly cracking several fronts at once — the war, the administration's internal ranks, House Democratic discipline, the voting system, and public tolerance for violence — closing on real alarm about the integrity of the coming midterms.

War fallout and a fraying administration. Renewed U.S.-Iran strikes send the Dow down nearly 500 points and oil back toward $100 a barrel, while August's private-sector hiring badly misses forecasts. Speaker Mike Johnson is caught on camera disclaiming any responsibility for the war's cost, which the hosts call absurd given Congress's oversight role. They connect this to visible churn inside the administration: Army Secretary Dan Driscoll's resignation fits a pattern of Vance allies exiting as Secretary of State Rubio gains ground, with one host predicting Rubio exits around April 2027 to launch a 2028 bid.

A commerce secretary under fire. CNBC's Joe Kernen presses Howard Lutnick on air about a financing deal tying his sons' firm, Cantor Fitzgerald, to Commerce Department decisions. Lutnick dismisses it as "minor," but the hosts see him, after a string of earlier controversies, as the administration's likeliest scapegoat.

Democratic discipline frays too. Leadership threatens to strip committee assignments from two members who broke ranks on a procedural vote without warning Hakeem Jeffries first, a episode the hosts read as exposing weak caucus communication rather than real disloyalty, compounded by Nancy Pelosi's continued, implicitly critical presence.

The real alarm: the mail. A USPS whistleblower describes a rushed, "fundamentally flawed" ballot-verification system that could kick back entire batches of mail ballots over a single scanning mismatch, just as voting begins; a court has frozen the mandatory piece, but the Justice Department is appealing.

A culture war over violence. A deadly Times Square knife attack on a young bank executive draws online justifications tied to her corporate job, which the hosts link to polling showing younger generations more willing to excuse political violence.

  • Renewed U.S.-Iran military strikes rattled markets and exposed Republican leadership trying to dodge accountability, the hosts argue, amid visible churn inside the administration. The Dow fell nearly 500 points, oil neared $100 a barrel, and the U.S. added just 38,000 private-sector jobs in August; Speaker Mike Johnson denied any responsibility for the war's economic toll when pressed on air, which one host calls absurd given Congress's constitutional war-oversight role. The hosts also flag Army Secretary Dan Driscoll's resignation as part of a pattern of Vance allies departing as Secretary of State Marco Rubio consolidates influence, and predict Rubio will likely leave his post around April 2027 to mount a 2028 presidential run.
  • Commerce Secretary Howard Lutnick was pressed on CNBC over an apparent conflict of interest involving his former firm, Cantor Fitzgerald, now run by his sons. A PIPE financing deal reportedly used Cantor to raise equity while Lutnick's Commerce Department controlled related federal financing decisions; Lutnick called it "minor stuff" and said he had no role in the decision, which the hosts frame as the latest in a string of controversies (Epstein-related issues, the Kennedy Center, the Canada trade war) that make him a likely fall guy for the administration.
  • House Democratic leadership's threat to strip committee assignments from Reps. Jared Golden and Marie Gluesenkamp Perez for breaking ranks on a procedural vote reveals a communication breakdown more than a betrayal, the hosts suggest. Golden's own co-sponsored fishing-industry bill was bundled into the GOP measure; one host argues Hakeem Jeffries needs to tighten caucus discipline given a possible narrow majority ahead, while both say Nancy Pelosi remaining visible and commenting on Jeffries's choices (including his meeting with Jared Kushner) continues to undercut him by comparison.
  • A USPS whistleblower's account of a "secretive, rushed, chaotic, and fundamentally flawed" mail-ballot verification system raises serious election-integrity concerns weeks before voting begins. Under the plan, a single scanner mismatch in a batch of barcoded ballot envelopes could cause the entire batch — potentially tens of thousands of ballots — to be rejected and returned to the state instead of delivered; a federal judge has frozen the mandatory portion of the rule, but the Justice Department is appealing, and one host calls the underlying Supreme Court ruling a bigger green light for the administration than it's been given credit for.
  • A deadly Times Square knife attack on a young Bank of America executive drew troubling online reactions excusing the violence because of her corporate role, the hosts note. Victim Erin Piacenti was killed on her first day back from maternity leave; the hosts contrast public celebration of the attack in some online corners with the Luigi Mangione case, citing polling showing younger generations are more likely than older ones to see such violence as justified.