Listen: https://serve.buildtall.systems/1a024a595df66bb7d4b6038ed895ac826304048cdff1fcd201677ebdf2727342.mp3
Daily Briefing for November 10, 2026. Sixty-six articles entered the window.
The dominant story concerns Nvidia and Jensen Huang. Several pieces converged on it. "The Jensen Huang Story Gets Worse, Washington vs. New York, Silicon Valley" from Stratechery by Ben Thompson states that the story got worse for Huang on Monday, and argues that the real risk for Huang is governance, even more than national security. The piece ends by asking who benefits.
The fullest treatment is "Nvidia CEO Jensen Huang Had an Affair With a Chinese Spy" from Don't Worry About the Vase. The article reports that China's top spy agency recruited Alice Huang, and that she then had an affair with Jensen Huang. Even if Huang did not know she was a spy, he knew she was a major customer, and he sold her billions in GPUs. The article argues national security has been compromised and that Huang must resign. It notes Nvidia underperformed the Nasdaq by roughly seven percent on the news. The piece walks through several layered claims. In hindsight there were hints: a deleted tweet by Roon suggested rumors circulated in Silicon Valley for some time, and a report from Culper named a close relationship as early as May. A WIRED report said Huang began working with crisis communications firm Heller a year before admitting to the affair. The article reports that President Trump was informed in October 2025 that Alice Huang had a relationship with Huang and was presumed to be a spy, and that Trump said he had no regrets about not acting. Two months later, Trump agreed to allow H200 exports to China with a twenty-five percent government cut. The article argues this timeframe looks damaging. It also reports that Huang lied about the affair through company spokesman John Rizzo and directly to Bloomberg, saying interactions with her were typical of those with Chinese customers. When Huang apologized, the letter was formalized and stilted, because, the article says, he let an AI write the apology. The article highlights that the apology treats national security almost as an afterthought, framing the affair mostly as a family matter. It reports that Nvidia's board has initiated an independent investigation while announcing in advance that it has no reason to believe national security or Nvidia secrets were compromised; the author argues a real investigation would not be announced that way. Crucially, the article reports that Commerce Department investigators were deep into a months-long investigation of Megaspeed, the Singapore data center company led by Alice Huang, suspected of smuggling chips into China, when the case was abruptly transferred to the Justice Department after Nvidia's general counsel contacted Commerce. Multiple agencies proposed putting Megaspeed on the Entity List, and Commerce kept not doing so, reportedly at Huang's urging. Jensen Huang has been asked to appear before three House committees on November 18 over significant national security concerns. The article distinguishes two scandals: Huang's affair and sale of chips, and the government's failure to stop him, with a third synthesis: officials knew for a year and did nothing. It concludes that Huang must resign, and that government officials who knowingly let this continue should resign as well. The article also relays arguments about whether the CCP actually wanted Nvidia's chips, quoting Samuel Hammond that the earlier narrative of Chinese ambivalence was a bluff while the spy agency worked to acquire billions in GPUs through shell companies and smuggling. It ends by considering whether boards should proactively test executives against intelligence targets.
A shorter piece, "Nvidia CEO Jensen Huang's affair with alleged Chinese spy: a timeline" from TechCrunch, reports that the affair could have ramifications for national security as the U.S. and China race to develop the most advanced AI infrastructure.
A second theme is long-run AI timelines. "Q3.5 Timelines Update: Looks Like the Intelligence Explosion Will Happen in 2027 After All" from the AI Futures Project is a long and detailed forecast. The article reports that METR recently estimated that Anthropic is making AI progress about twice as fast as its historical rate, and guessed that around eighty percent of current acceleration is due to AI. The authors write that this implies an AI-driven acceleration roughly five months ahead of the AI 2027 scenario. They now forecast that Superhuman AI Researcher, or SAR, will likely arrive sometime in 2027, assuming no substantial deliberate slowdown, and give a thirty-five to sixty-five percent chance that what they call Top-Expert-Dominating AI will arrive before 2028. The article walks through several simple models: a linear extrapolation of research taste gives SAR by late 2027; labor uplift extrapolation gives mid-2028; extrapolating Anthropic automation levels gives late 2027; Ajeya Cotra's toy model gives December 2027; and Tom Davidson's toy model implies SAR sometime in 2027. The authors' filtered AI Futures Model gives a median SAR date of February 2027 when conditioned on the observed 1.7x November acceleration. They then explore hypotheses for why their prior model was wrong, including the possibility that near-automated-coder skills produce more research and development uplift than anticipated, that skills besides coding and research taste drive acceleration, or that their diminishing returns model is too conservative. The article emphasizes that a 1.7x acceleration of overall progress is a much bigger deal than a 1.7x labor uplift, because labor is only one input. The authors are explicit about the implications: if SAR is coming in 2027, an industry-wide pause on frontier AI research and development should be the top priority, because a misaligned SAR used for recursive self-improvement would put the world in an extremely bad position. They also warn that the algorithms, and maybe the weights, will probably leak to China without much better security.
A third theme is the crackdown on AI-powered reverse engineering. "'Reverse Engineer Anything': AI Companies Crack Down on the Coolest Thing People Were Doing With AI" from 404 Media reports that people using Anthropic's Claude and OpenAI's Codex for reverse engineering or decompiling are hitting guardrails or having accounts limited, while doing things like unbricking internet-of-things devices or removing surveillance ads from smart TVs. Users reported safety checks in Claude Code, a Reddit post titled "The Dream is dead," multiple reports in a Discord thread, and an X user banned by OpenAI for violating rules on cyber abuse. A project called Decomp.wtf shut down. The article explains that Reverse Engineer Anything made it easy to hack electronics people own, install new operating systems on smart TVs, turn baby monitors into AI assistants, and unbrick abandoned devices. The author argues this is one of the few LLM uses genuinely worth getting excited about, because it returns ownership and control over devices that manufacturers have bricked or turned into spyware. The article argues the policy tweaks show AI corporations protecting other hardware corporations, and notes that the future of the practice lies in open weight, local, and uncensored models. It quotes OpenAI engineer Tarun Gogineni, who tweeted that blocks are mostly accidental and that there is no prohibition against reverse engineering with Codex. It also quotes security researcher Matt Brown, who said his experience of blocks is nearly one hundred percent correlated with cyber-category blocks, and that he is investing in open weight models, many from China, because frontier guardrails change without warning. The article cites researchers at Cornell Tech, Columbia, and NYU who argue cheap AI-assisted reverse engineering challenges manufacturer dominance and existing intellectual property law. The piece closes with science fiction author Bruce Sterling's line that vibe-code hacking is only called reverse when somebody who lacks a trillion dollars is doing it.
A related newsletter piece, "Come and Take It" from 404 Media, covers the same reverse engineering story and adds several shorter items: a father in South Carolina who asked his six-year-old whether she had the gun before leaving her in line for Pokemon cards; the latest Nvidia lawsuit and the Times report that Commerce killed the Megaspeed investigation; a Department of Justice statement that it lacks enough attorneys to respond to every habeas corpus challenge; the rising price of top-end Raspberry Pi boards, attributed to AI datacenter memory demand; actors alarmed to learn they had been replaced by AI in a Return of the Living Dead sequel; an argument that U.S. government IT systems are compromised and require a total rebuild; an Atlantic piece on constant surveillance of children; and a printer that was Rickrolled.
A large set of articles concerned cybersecurity and government security. "New OpenSSH Vulnerability" from Schneier on Security reports a newly disclosed flaw involving OpenSSH and musl, dubbed asshole, that allows attackers to log in to certain Alpine Linux and other musl-based accounts configured to block password authentication, simply by entering an oversized password, meaning one longer than 256 bytes. The flaw has been in the code for a decade. "Russian hackers compromised government email servers to steal NATO, EU correspondence" from The Record reports that Russian state-linked hackers compromised government email servers in Romania and Kazakhstan to intercept messages exchanged with NATO and European Union officials. "Cybercriminals are increasingly infecting regular people with spyware" from The Record reports that cybercriminals are infecting devices on a large scale with spyware, increasingly compromising systems belonging to average people, according to a new report. "N-able takes its remote access service offline after researcher publishes dozens of zero-day bugs" from TechCrunch reports that the remote access company took its service offline after a researcher published dozens of zero-day bugs, and urged self-hosted customers to take down their own servers amid a critical security incident. "AI company Fireworks says hackers stole data, billing info during October breaches" from The Record reports Fireworks said the incident affected a limited number of customer accounts, all of whom were notified. "Microsoft's Patch Tuesday surpasses 1,000 bugs for second month in a row" from The Record reports an exploited Outlook vulnerability drew most attention, with the latest Patch Tuesday listing 1,093 bugs total. "US Army member arrested in UK over allegations of spying for Russia" from The Record reports an unnamed soldier was arrested by British counterterrorism police working closely with U.S. authorities, including the FBI and the military. "Palantir exec nominated to lead CISA; lame-duck Senate vote is uncertain" from The Record reports Michael LoSapio, Palantir's chief information officer, is officially President Trump's pick to lead the Cybersecurity and Infrastructure Security Agency. "Cyberattacks threaten to make open-source AI a little less open, Hugging Face exec warns" from TechCrunch reports Hugging Face executive Irene Solaiman warns that growing cyberattacks on open AI platforms could force tighter security restrictions, threatening accessibility. "Europe Must Rethink Security As Big Tech Power Grows, EU Says" from Slashdot reports European Commission Executive Vice-President Teresa Ribera said in Lisbon that the EU must review and deepen its concept of security, including cybersecurity, and ensure Big Tech companies are governed by democratic and transparent patterns.
Proton contributed two privacy and security pieces. "How Proton keeps your data secure in the age of AI" from The Proton Blog explains that Proton uses end-to-end and zero-access encryption, so it cannot read emails, files, or passwords, and no AI tool can either. The documented security principles have not changed since June 2024 and have not been loosened for AI. The company is ISO 27001 certified and completed a SOC 2 Type II audit in July 2025. For AI-written code, the same unified controls apply to all code: a person approves every change; security scanners are required; engineers remain accountable; and designs are reviewed before major features are coded. AI agents get only least-privilege access; the SIEM system records actions from people or AI agents alike; and physical tokens protect internal systems. The article argues AI is necessary for defense, because attackers need only find one flaw while defenders must close all of them. Proton runs AI reviewers alongside humans, uses non-public cyber models from OpenAI and Anthropic for threat analysis, and AI tooling helps its security operations center monitor a surface far beyond what unaided humans could watch. The company promises a disciplined approach with a person accountable for every production change.
from The Proton Blog compares the two payment apps. Venmo made transactions public by default for years; researcher Hang Do Thi Duc collected roughly 208 million public Venmo transactions from a single year, and reporters found President Biden's account in under ten minutes after a New York Times report. Venmo has since removed its global public feed but still has a social feed with Public, Friends Only, and Private settings. Cash App has no comparable social feed, but collects substantial information, including device details, IP addresses, and location, and in 2026 began using information to personalize advertising for other brands. Venmo is owned by PayPal and may share data with PayPal companies; it retains information for as long as the account exists plus ten years after closure. Cash App is owned by Block; a former employee downloaded Cash App Investing reports affecting about 8.2 million customers, and in January 2025 the CFPB ordered Block to pay up to 120 million dollars in redress and a 55 million dollar penalty over fraud handling. The article concludes that Cash App comes out ahead on privacy, that neither has a spotless record, and that physical cash remains the more private option.
Anthropic appeared in several business stories. "Anthropic sets IPO price range of $70 to $78 per share to reach an implied valuation of $1.404 trillion" from TechCrunch reports the company could end up worth 1.404 trillion dollars if shares list at the top of the range, on a fully diluted basis. "All 7 Anthropic founders will each cash out about $250M in IPO" from TechCrunch reports each of the seven co-founders plans to sell exactly the same number of shares, bringing each about 250 million dollars at the midpoint, while leaving the group in control. "Anthropic launches Claude Meetings to turn workplace discussions into plans and tasks" from TechCrunch reports that as participants talk, Claude can draft the plan being discussed, map a timeline on a whiteboard, answer questions in meeting chat, and access connected work apps. A Slashdot piece, "Anthropic Launches New Workplace Features To Take On Microsoft and Google," adds details: Claude Meetings can join video calls as a visible participant, initially supporting Google Meet; Claude Whiteboards is a collaborative visual workspace; Claude Sheets allows collaborative spreadsheets; and Claude Data, formerly dashboards, connects to CRMs and data warehouses. The launch is positioned against OpenAI's workplace tools, with Meta joining with Muse.
A broad group of TechCrunch product and business announcements followed. "Uber Is Launching Its Own Parking Reservation Service" from Slashdot reports Uber Parking integrates technology from its SpotHero acquisition to let users find and reserve parking through the Uber app. Users must provide make, model, and plate; lots use QR codes or plate recognition; most cancellations are free up to start time. Uber One subscribers get ten percent back in credits. Launch cities include Chicago, Nashville, San Francisco, Atlanta, Austin, Boston, Dallas, Denver, Los Angeles, Miami, New York, Philadelphia, San Diego, Seattle, and Washington DC. Uber also plans to experiment with lots as staging areas for autonomous vehicle fleets. "Ring, known for its doorbell cameras, launches a home security drone and an enterprise security robot" from TechCrunch reports Amazon-owned Ring is pushing into robotics with a home security drone and a rolling robot for patrolling businesses. "Spotify's AI DJ gets new voices, personalities, and an option for less chatter" from TechCrunch reports Premium users can customize the AI DJ with new voices and personalities, or turn off commentary. "Aura's newest e-ink photo frame is even better and more affordable than its last" from TechCrunch reports a ten-inch digital e-ink frame that blends with printed photos. "After challenging Google's Android with Cyanogen, its co-founder takes aim at Windows for AI era" from TechCrunch reports Kora is set to launch as a free-to-use alternative to Windows, targeting users who rely heavily on AI models and agents. "Adobe adds more AI to its products and launches an MCP" from TechCrunch reports Adobe is taking on Figma with a new prototype builder in Firefly. "Reddit is turning user conversations into AI-powered product reviews" from TechCrunch reports the new AI-powered reviews summarize what users like and dislike about 10,000 products, helping shoppers research purchases. A Slashdot version, "Reddit Is Turning User Conversations Into AI-Powered Product Reviews," adds that the feature has been tested since February and is rolling out to all U.S. users on iOS, Android, and web, with a breakdown of positive, negative, and mixed comments. "Meta's AI agent Muse lands on Apple Watch and Android Wear OS" from TechCrunch reports users can interact with Meta's assistant through voice calls, texts, or dictation on the watch. "Google broadens its AI hardware portfolio with two new specialized chips" from TechCrunch reports new TPUs broaden Google's portfolio for confidential model training and more memory capacity. "Apple TV makes sign language interpretation a built-in accessibility feature" from TechCrunch reports viewers can turn on sign language interpretation within supported titles. "Sundrop turns your endless baby photos into recaps, time-lapses and more — without AI" from TechCrunch reports an iPhone app using Apple's on-device AI to help parents rediscover favorite photos without uploading images to the cloud. The title's claim notwithstanding, the piece describes the app as using Apple's on-device AI. "Pixel Tag, Google's $30 Android-ready answer to Apple's AirTag, is now available" from TechCrunch reports the device tracks lost items with an Android device. "OpenAI rolls out new business profiles, making ChatGPT just a little bit more like Google" from TechCrunch reports OpenAI is positioning ChatGPT as a consumer discovery engine, with new features to help businesses reach users. "PSI is trying to make a business out of AI-driven scientific discovery" from TechCrunch reports PSI used an unreleased Anthropic model to answer a fundamental question about quantum computing that scientists pursued for nearly two decades. "Oura to acquire AI-native primary care startup Counsel Health for $625M" from TechCrunch reports the deal comes weeks after Oura postponed its up to 2.2 billion dollar IPO indefinitely, citing uncertainty in the IPO market. "OpenAI, Anthropic face new oversight as New York's AI safety law takes effect" from TechCrunch reports New York is preparing to enforce a sweeping AI safety law requiring major developers to register, disclose safety practices, and report critical incidents or face fines. "X-energy gets green light to build its first small modular nuclear reactors" from TechCrunch reports a construction permit from the Nuclear Regulatory Commission to build four reactors in Texas. "Proximal Ventures raises debut fund to back startups bringing AI to drug development" from TechCrunch reports former SignalFire investor Sahir Raoof raised 27.5 million dollars for a fund targeting AI in pharma. "OpenAI Acquires Odessia, travel startup created by Sonder founder" from TechCrunch reports OpenAI has not said what its plans are for the future of travel, but Sonder founder Francis Davidson is a well-known travel figure.
A smaller cluster addressed AI agents and development practices. "Long-horizon agents and durable execution" from Martin Fowler explains a pattern by Kiran Prakash for agents that automate tasks taking days or weeks. Durable execution keeps a record of each step that interacts with the outside world, so after a failure the program runs again, skips completed steps, and rebuilds state from the record.
Is BI Dead? Why Hex Hit Reset and Rewrote Everything for the AI Agent Era.
from Joe Reis reports that Hex spent a year rewriting its core architecture to be agent-native, after an earlier approach of bolting a chatbot onto legacy UI with hundreds of custom tools was not enough. A fireside chat with CEO Barry McCardel at Hex's Prompt conference covers the reset. "Fragments: November 10" from Martin Fowler surveys the November Thoughtworks Technology Radar. Chris Ford's macro trends article notes a team of passionate people can make almost anything work, but two hundred people cannot: shared context, token cost, review demands, agent reach, and platform responsibility appear. The piece also summarizes Ben Thompson's Stratechery post on data center opposition in Wisconsin campaign ads, noting anti-data-center sentiment is bipartisan, and that existing data center owners, other countries, China, and Apple's edge devices may benefit if new building is constrained. Fowler adds that incumbent companies may quietly lobby for regulation that blocks new competitors. Hadley Wickham's analysis of token costs concludes per-token API prices are not significantly subsidized, but subscription plans are. A paper by Takehi and colleagues on weight-space programming describes GPT-6 Astra reading raw weights to write small game-playing networks from scratch. Houck and colleagues propose a framework called CAFE(S) with five dimensions for context quality: clarity, actionability, fidelity, efficiency, and security, plus context smells. Cindy Sridharan notes that seventy percent of her daily token spend now goes into verifying patches, and only twenty to thirty percent into creating them.
Two long New Lines Magazine pieces concerned the Middle East. "Houthi Missiles Force Postponement of Saudi Asian Cup" reports that on October 12 Cristiano Ronaldo walked around Al-Awwal Park stadium with a Saudi flag scarf, and after the game posted that he is Portuguese but belongs to the Kingdom of Saudi Arabia. The start of that match was delayed one hour by a missile alert. On October 10 an attack on King Khalid International Airport killed at least thirteen people and injured over three hundred. The Asian Football Confederation postponed the 24-nation Asian Cup, due to begin January 7, by almost a year, now set for December 23, 2027. The article situates the postponement within Vision 2030 and Saudi attempts to become a sporting destination: Ronaldo's arrival, 957 million dollars in Pro League signings, the 2034 World Cup bid, and an estimated 8.5 billion dollar sporting industry with a 22 billion target. The postponement also affected the Future Investment Initiative, the Saudi International golf tournament, the Six Kings Slam tennis exhibition, the Urban Cycling World Championships, and a FIFA video game event. Governments including South Korea warned their citizens to leave Saudi Arabia, and the article argues no host can stage a tournament when players and teams are told not to travel. The piece reports Qatar was never an acceptable substitute this time, and possible East Asian hosts would face freezing January conditions. The article also reports the February attacks on Iran hit the Saudi economy hard, with GDP shrinking 4.7 and 3.8 percent year on year in the second and third quarters of 2026, and that LIV Golf filed for bankruptcy in September after PIF withdrew support. World Cup planning has fallen from fifteen stadiums to ten, and the NEOM stadium appears quietly dropped. Pro League summer spending fell to 407 million dollars, less than half the 2023 peak, with clubs seeking younger, cheaper players. The article concludes that the Asian Cup postponement may be a temporary setback or a turning point, and that geopolitics has tightened its grip on Saudi sporting ambitions.
"Syria's UNESCO-Listed Crafts Are Dying Out" from New Lines Magazine opens at the Azem Palace in Old Damascus, where wax figures watch a puppet show as a historical mirage. Shadow puppetry survives in the home of Shadi al-Hallaq in Hamoura. His father Abu Shadi was a famed hakawati, or storyteller. Shadi once counted over a thousand leather figurines; now a dozen remain, the rest sold to pay debts. He says UNESCO recognition in 2018 did nothing to save the art. The article reports Shadi fled Syria in 2015, then received a call saying the first lady wanted him back to practice his craft with promises of a house, a car, and a salary. He returned in a convoy that passed checkpoints on the authority of the first lady, but found a classroom with mattresses and an exit ban; the promises were lies. The article identifies the organization as the Syrian Trust for Development, founded by Asma al-Assad, which held a near monopoly on international humanitarian funding and was a primary recipient of over four billion dollars in U.N. funding during the civil war. The piece reports the trust evicted oud maker Antoun Tawil and dozens of craftspeople from the Sulaymaniya Takiya in 2022, the year after nominating Syrian oud-making for UNESCO. Antoun nearly quit the craft; a local cafe's fundraiser raised three thousand dollars toward a new shop. The trust has been rebranded as the Syrian Development Organization under new authorities, with seventy percent of employees kept, including the head of cultural heritage. The SDO says its priority is internal reform and governance, and secured just under one hundred thousand dollars from UNESCO for traditional glassblowing. The article then visits Abu Mahmoud, the last surviving handblown glass workshop in Syria, a family business of three hundred years. His furnace is cold because it must stay hot around the clock and there are no orders. His son has not taken up the trade. The piece reports that Antoun's daughter gave up oud-making after the eviction, and Shadi's son Jawad wants to carry on the puppetry even if he cannot live from it. The theme across practitioners is a generational tension: the new generation cannot afford to learn crafts whose rewards are small.
Two Marginal Revolution pieces touched economics and research.
reports on SaneCite's check of an anonymous AI audit of Emmanuel Saez and Gabriel Zucman's research. Of 147 problems the audit called serious enough to change a headline number, 131 had every checkable fact right and none had core facts wrong; 21 change a headline number; 98 depend on a disputed choice of method; 20 change no number; and 8 could not be checked. A second pass graded the 99 disputed-method flags: 11 standard, 70 defensible, 11 questionable, 6 ungradeable, and 1 contested, with one additional headline-number change. Saez and Zucman did not reply by November 6. "Are RCTs underpowered?" reports on a paper by Qingyuan Chai, Kevin Lang, and Liang Zhong using 745 principal-hypothesis tests from 190 articles. Actual t-statistics peak below the theoretical benchmark of 2.8, and the authors estimate approximately thirty percent of published rejections are false discoveries, with average directional power of forty percent. Doubling sample sizes lowers the false discovery rate by five points; tightening the threshold to 0.005 yields a larger reduction at the cost of discoveries. Under optimal replication policies, successful replications reduce aggregate false discovery rates to 1.8 to 7.8 percent. "Tuesday assorted links" points to Richard Hanania on AI as a normal technology, a free Cass Sunstein element on Hayek, a works-in-progress writing call, Matthew Lilley on Harvard graduate students, Ethan Mollick on Fable 5.2, John Cochrane on Brazil, and the new Godzilla release topping the box office in North America and Japan simultaneously.
A long essay on social intervention appeared. "We need more goat sacrifices" from Experimental History argues that before any major social policy or philanthropic program, people should sacrifice a goat as procedurally-induced humility, because an alarming number of well-intentioned interventions do not work. It lists failures: three billion dollars in school improvement grants with no effect; chief diversity officers who did not increase diversity; 280 million dollars for Afghan women's employment; one laptop per child with zero academic effects in Peru; and four vaccine persuasion messages, one of which made parents more likely to believe vaccines cause autism. It adds right-coded failures: Arkansas Medicaid work requirements reduced insurance without increasing employment; Building Strong Families counseling did nothing; and four abstinence-only sex ed programs had no effects. The piece recounts Peter Rossi's 1987 satirical Iron Law of Evaluation and his later walk-back, and the difficulty of finding reliably effective programs, noting that the Earned Income Tax Credit may not do anything, Australia's plain packaging may have increased smoking among some smokers, and an Indian reading and math program shrank when replicated elsewhere. The article argues the illusion of control explains misplaced confidence: the original 1975 Langer study was almost certainly false, yet people still blow on dice and buy The Secret. The piece reports that 76 percent of NSF-funded STEM education projects do not attempt experimental evaluation, that DHS has no goals or data for its fentanyl efforts, and that Kenneth Langone's hundred-million-dollar donation to NYU medical school did not change student specialties or demographics, merely increasing applications. The author describes three stages of grief when people see failed interventions: bargaining through common sense, denial through no-true-Scotsman objections, and acceptance that improving lives is impossible. The essay closes by arguing acceptance is the wrong lesson, compares the social sciences to alchemy, and urges patience and planting seeds, ending with a Seneca quotation and the observation that Experimental History is produced in the kidneys.
Several Nautilus pieces covered science and behavior. "This Photo of the Night Sky Reveals 14,000 Satellite Trails" reports an image so striking many people thought it was AI. "What the Measles Vaccine's "97% Effective" Rating Means for Pittsburgh Parents" reports demand for the MMR vaccine has risen in Pennsylvania as the state's measles outbreak has grown. "The Difference Between Nice and Kind" examines how people talk about good deeds and acts of altruism. "Descendants of Chornobyl's Pets Run Wild" reports their genomes suggest adaptation to toxic conditions. "Researchers Can Tell How Lonely You Are by Your Instagram Activity" asks, are you lonely posting.
A few pieces fell outside the main clusters. "Bad employees/bad bosses" from Seth Godin argues that when an organization is struggling it is tempting for the boss to blame the team, but the evidence does not support this: similar organizations draw from the same labor pool, and creating conditions for commitment, learning, and resilience is the boss's job. "In search of a bargain" from Seth Godin announces that the ebook of The Knot is on sale for less than five dollars in the US, and previews a three-book holiday bundle limited to the first 170 people. "The wholehearted life" from Aeon, by Ben Daitz, profiles Roshi Joan Halifax, an original, generous, gritty, and profound Buddhist teacher, asking how she looks back on her life. "Book Review: The Unfinished Harauld Hughes - Richard Ayoade ★★★⯪☆" from Terence Eden's Blog reviews Ayoade's absurdist novel about a semi-fictional TV presenter making a documentary about a probably fictional director. The review says it is wryly funny and mercilessly satirical, revels in mundane detail, and has moments of genuine profundity, though the drip-fed mystery is not enough to sustain the story. "Entitlement Isn't Attractive..." from The Next Step argues from the author's gap year on a student agricultural leadership team that no one is entitled to anything, ever, for any reason, and that joy comes not from material comparison but from gratitude and purpose. "llm-gemini 0.36" from Simon Willison notes Gemini 4 Carbon is now generally available at two dollars per million input and ten per million output, scheduled to double on December first, and includes several generated pelican images and a Three.js Yonkers animation test.
Two hardware and space items closed the window. "AMD EPYC 9006 Results Are Here and AMD Delivers a Knockout Part Deux" from ServeTheHome covers the launch of AMD's EPYC 9006 Venice platform and Zen 6 architecture, arguing Venice is primed to deliver another knockout blow. "Starlink Mini 2 Adds a Swappable Battery, Wi-Fi 6" from Slashdot reports SpaceX unveiled the Mini 2 with a removable 71 watt-hour battery, Wi-Fi 6, USB-C power delivery, a power button, foldable kickstand, and a 999 dollar preorder price with a year of Roam Unlimited. Shipments are expected mid-December. The battery weighs about 1.08 pounds and provides up to 4.5 hours of connectivity.
A few remaining items: "Blender 5.3 Released With Nvidia DLSS 4.5 Ray Reconstruction, Vulkan Default" from Slashdot reports support for 3D Gaussian splats, DLSS 4.5 ray reconstruction, RTX Spark devices, dispersion in the Principled BSDF, and Vulkan as the default graphics backend on Windows and Linux. "Nintendo Is Officially Investing In Game Preservation" from Slashdot reports Nintendo is hiring a digital archivist for a 12-month position to organize technical records, engineering documentation, and source materials. The piece notes the preservation community has criticized Nintendo for closing the Wii U and 3DS eShops and for funding lobbying through the Entertainment Software Association. That closes the window.
- Bad employees/bad bosses
- AI usage generates a fragile codebase
- Uber Is Launching Its Own Parking Reservation Service
- The wholehearted life
- The Jensen Huang Story Gets Worse, Washington vs. New York, Silicon Valley
- How Proton keeps your data secure in the age of AI
- Checking the AI audit of Saez and Zucman’s research.
- New OpenSSH Vulnerability
- AlphaProof Code
- Syria’s UNESCO-Listed Crafts Are Dying Out
- Book Review: The Unfinished Harauld Hughes - Richard Ayoade ★★★⯪☆
- Ableton Live 13
- Russian hackers compromised government email servers to steal NATO, EU correspondence
- This Photo of the Night Sky Reveals 14,000 Satellite Trails
- Cyberattacks threaten to make open-source AI a little less open, Hugging Face exec warns
- Nvidia CEO Jensen Huang Had an Affair With a Chinese Spy
- Don't buy AR glasses
- Ring, known for its doorbell cameras, launches a home security drone and an enterprise security robot
- Houthi Missiles Force Postponement of Saudi Asian Cup
- Spotify’s AI DJ gets new voices, personalities, and an option for less chatter
- Aura’s newest e-ink photo frame is even better and more affordable than its last
- After challenging Google’s Android with Cyanogen, its co-founder takes aim at Windows for AI era
- Anthropic sets IPO price range of $70 to $78 per share to reach an implied valuation of $1.404 trillion
- Adobe adds more AI to its products and launches an MCP
- Long-horizon agents and durable execution
- 'Reverse Engineer Anything': AI Companies Crack Down on the Coolest Thing People Were Doing With AI
- Fragments: November 10
- Palantir exec nominated to lead CISA; lame-duck Senate vote is uncertain
- Come and Take It
- What the Measles Vaccine’s “97% Effective” Rating Means for Pittsburgh Parents
- Reddit is turning user conversations into AI-powered product reviews
- Entitlement Isn’t Attractive...
- Meta’s AI agent Muse lands on Apple Watch and Android Wear OS
- Cybercriminals are increasingly infecting regular people with spyware
- N-able takes its remote access service offline after researcher publishes dozens of zero-day bugs
- Google broadens its AI hardware portfolio with two new specialized chips
- Europe Must Rethink Security As Big Tech Power Grows, EU Says
- Apple TV makes sign language interpretation a built-in accessibility feature
- In search of a bargain
- Gemini 4 Carbon is now available
- AI company Fireworks says hackers stole data, billing info during October breaches
- Q3.5 Timelines Update: Looks Like the Intelligence Explosion Will Happen in 2027 After All
- Blender 5.3 released with Gaussian Splat support
- .name Grandfathered
- Tuesday assorted links
- Sundrop turns your endless baby photos into recaps, time-lapses and more — without AI
- Reddit Is Turning User Conversations Into AI-Powered Product Reviews
- The Difference Between Nice and Kind
- AMD EPYC 9006 Results Are Here and AMD Delivers a Knockout Part Deux
- The newest iteration of the Great Script Restoration
- Nvidia CEO Jensen Huang’s affair with alleged Chinese spy: a timeline
- Venmo vs. Cash App: Which is more private?
- Is BI Dead? Why Hex Hit Reset and Rewrote Everything for the AI Agent Era.
- We need more goat sacrifices
- Anthropic launches Claude Meetings to turn workplace discussions into plans and tasks
- Val Town is winding down
- Blender 5.3 Released With Nvidia DLSS 4.5 Ray Reconstruction, Vulkan Default
- Synaptics patent claims stop work on DisplayLink drivers
- Starlink Mini 2
- Pixel Tag, Google’s $30 Android-ready answer to Apple’s AirTag, is now available
- OpenAI rolls out new business profiles, making ChatGPT just a little bit more like Google
- Nintendo Is Officially Investing In Game Preservation
- Descendants of Chornobyl’s Pets Run Wild
- PSI is trying to make a business out of AI-driven scientific discovery
- Oura to acquire AI-native primary care startup Counsel Health for $625M
- Anthropic IPO Financials Are Getting FUN (Money Stuff, Matt Levine)
- All 7 Anthropic founders will each cash out about $250M in IPO
- Are RCTs underpowered?
- llm-gemini 0.36
- Starlink Mini 2 Adds a Swappable Battery, Wi-Fi 6
- Stacker News Monthly: October 2026 NEW Territory Contributor Stats
- X-energy gets green light to build its first small modular nuclear reactors
- Proximal Ventures raises debut fund to back startups bringing AI to drug development
- OpenAI, Anthropic face new oversight as New York’s AI safety law takes effect
- Microsoft’s Patch Tuesday surpasses 1,000 bugs for second month in a row
- Anthropic Launches New Workplace Features To Take On Microsoft and Google
- Researchers Can Tell How Lonely You Are by Your Instagram Activity
- OpenAI Acquires Odessia, travel startup created by Sonder founder
- US Army member arrested in UK over allegations of spying for Russia

