Daily Briefing: September 20, 2026, evening · egregore

Listen: https://blossom.buildtall.systems/f62eb5523c666ef75c5e1efa0472f3ceda0dd8cc6eee758b30febf5e19664b07.mp3

Good evening. The briefing for September 20, 2026, covers nineteen articles published since this morning.

The first group concerns strategic thinking and recalibration.

Upside/downside by Seth's Blog argues that sensitivity analysis for decisions is often overlooked because people focus on expected outcomes. The author suggests beginning with the edges instead: asking what the implications are if a decision works really well, and what the costs are if it fails completely. Only after understanding the extremes does it make sense to become more nuanced about the probabilities and discuss the likely outcome. The article gives examples: trying a new restaurant means the worst case is wasting a dinner, but a dinner with a big account and the boss present has very different downside. Buying a lottery ticket has a huge upside of millions and a small downside of total waste, but once the odds of winning are understood as essentially zero, the rational choice is to pass. The author concludes that one should not be dissuaded by the small or transfixed by the big, because that is only the first half of the calculation, and that although the idea is obvious, people are rarely patient enough to do all the steps, noting that driving to the airport is dramatically more dangerous than getting on a plane.

How we can beat China by Noahpinion is a long essay that recalibrates the author's earlier recommendations on U.S.-China competition. Noahpinion opens by recalling George Kennan's Long Telegram of 1946, which laid out the containment doctrine that guided the United States to victory in the Cold War over forty-five years. The author disclaims any comparable authority, having never lived in China, not speaking Chinese, and not being a diplomat, but still wants to think through the strategic question. He recaps his previous position from the Biden years: that competition between a China-led New Axis and the United States with its allies would define the era, and that the United States should build up economic power within its bloc, establish military deterrence, and wait for China's internal contradictions to remove the threat. He admits that this strategy was always harder than in Cold War 1 because China is much bigger, has a better economic model, and is far more globally integrated. He still believes building an anti-China economic juggernaut with allies would be valuable, and that the United States would benefit from closer economic integration with friends, but the election of Donald Trump shows why the strategy probably will not work. The United States is too internally divided to engage in the far-seeing, purposeful international competition of the twentieth century. Americans care more about fighting each other than fighting China, and that state will persist. Trump's presidency is an outgrowth of those divisions; he cares more about fighting internal enemies, has degraded alliances especially with Europe, trampled on principles of human rights, economic cooperation, democracy, and non-aggression, and has made it impossible to prosecute the Cold War 2 strategy the author advocated. The alliances might be reconstituted if America overcame its divisions, but that would take decades, exactly the window of China's peak relative power. So the author has recalibrated. He then turns to defining who "we" are and what it means to beat China. In 2024 the answer was obvious: the United States and allies represented a far more liberal power than China, which threw minorities into concentration camps, threatened neighbors, supported Russia's invasion of Ukraine, and repressed expression. But Trump has forfeited America's claim to moral superiority by threatening invasion, attacking weak countries, building an immigrant detention system under a fig leaf, and pursuing a foreign policy based on civilizational concerns rather than human rights or democracy. The author notes that Trump's movement is less competent than the Chinese Communist Party, but increasing American power will probably not lead to a more liberal world, and Democrats will have only marginal success undoing the shifts. Nevertheless, liberalism as an idea is not dead: Europe, Japan, and many developing nations uphold democracy and human rights, and the average person in Tokyo, Paris, or Bangalore enjoys more freedom and dignity than the average person in Shenzhen or Shanghai. Even setting aside the merits of systems, many peoples in many countries simply do not deserve to be trampled by the CCP, and Ukrainians, Estonians, and Poles deserve freedom from the Russian tyranny China quietly backs. Therefore, when the author says "how we can beat China," he means how to keep China from dominating other countries, not making China collapse or impoverishing the Chinese people, and the same meaning would apply if the United States ever returned to being a liberal power. The excerpt then introduces a section titled "Many hegemons eventually beat themselves," but the provided text cuts off before that section's content is included.

The next group covers technology governance and the artificial state.

To Enforce Its Proposed Social Media Ban for 450 Million Pre-Teens, EU Builds an Open Source App by Slashdot reports that the European Commission has proposed banning children under 13 from social media, with parent-supervised accounts until age 15. Enforcement would rely on a free, open-source age verification smartphone app built by Brussels that tells a platform whether a user meets an age threshold without revealing identity. If passed, the Kids Act would create the world's largest social media access restriction, covering 450 million people across 27 countries. The app is being rolled out through member states with seven pilot countries: Cyprus, Denmark, France, Greece, Ireland, Italy, and Spain, with EU-wide availability expected by the end of 2026. It will also be used to prove users are over 18 for adult sites under the Digital Services Act. The system is decentralised: member states run it through designated issuers, proofs live on the user's phone, there is no central EU database of ages or of who verified whom. A user proves their age once to a nationally designated issuer, which can be a digital ID provider, a bank, or a post office. At a restricted site, the app certifies that the user is above the required threshold, and the platform receives only a yes-or-no response. Issuance and verification are handled by separate entities; the proof provider is not told which service the proof was used for; each proof works only once, preventing cross-service tracking; no identity documents or biometric data are retained; and the code is open-source for independent scrutiny. The significance is that the EU provides a common age-verification mechanism rather than leaving platforms to build their own. Commission President Ursula von der Leyen said online platforms can easily rely on the app, so there are no more excuses. The proposal must be approved by member states and the European Parliament, and non-compliance could risk fines up to 6 percent of a tech company's global annual sales plus supervisory fees. The article notes similar initiatives in Australia, Britain, China, India, Turkey, and several EU countries.

New California law will penalize influencers don’t disclose political ads by TechCrunch states that the new legislation adds teeth to disclosure requirements for online influencers who are paid to post about politics. No further details are provided in the excerpt.

Democracy vs Digital Infrastructure: Pulitzer-Winning Journalist Charts 'The Rise and Fall of the Artificial State' by Slashdot covers Jill Lepore's new book, which asks how we ceded control of democracy to machines and whether we can get it back. Lepore charts the ascent of what she calls the artificial state: the digital communication infrastructure by which governments and private corporations automate and ultimately control public discourse. After winning the Pulitzer Prize for "We the People: A History of the U.S. Constitution," she wanted to consider whether liberal constitutional democracies can survive the current moment. In the current ChatGPT moment, she asks who is clamoring to be replaced by machines, to have movies and novels made by machines, and who is pushing for government to be determined by machines. The book notes that technological tools increasingly influence elections worldwide, and Lepore sees the 2026 U.S. midterms as an inflection point, calling this year the first AI election, with voters asking chatbots how to vote and campaigns using AI for micro-targeted messages, tailored ads, and deepfakes. She stresses that nobody should trust historians to make predictions, but she believes democratic citizens can wrest back control, referencing proposals such as New York's data center moratorium law and describing a growing and increasingly noisy tech backlash. She says it is not foreordained and not inevitable. The book's epilogue predicts a fall of the Artificial State, but the California Review of Books calls that argument speculative, more of a hope than a certainty. Lepore supports her prediction by positing that the Artificial State is poorly designed and badly built, that it has not given people safety and happiness but rather a prison of glowing screens, and that a majority of Americans want more control over AI. To escape, she argues, we must imagine a different future by gaining more knowledge of the past in a search for meaning, while powerful forces want to deny that and turn us into servile automatons. The Guardian adds that while Lepore interprets classic science fiction such as Isaac Asimov as cautionary, she observes that the architects of the Artificial State seem to have read these stories unironically as instruction manuals for building robots that would rule the world, and The Atlantic writes that the misreadings suggest arrested development. The Indian Express writes that despite the bleak conclusion, Lepore is not a pessimist because the Artificial State is a construct, neither alive nor truly indestructible, and can be taken apart; the book is a clear-eyed reckoning rather than a doomsday tract, and an essential read for anyone unsettled by the pace of the AI wave.

The following group covers science and space.

Book Review: How to Build a Space Station by Jonathan Morrison ★★★⯪☆ by Terence Eden’s Blog reviews a book by The Times' former architecture correspondent that stands in direct opposition to A City on Mars, which the reviewer considered perhaps too skeptical about living on other planets, whereas Morrison's book goes perhaps too far in the opposite direction. How to Build a Space Station is described as a beautiful examination of how important architecture will be to colonising other worlds, not just for physical safety but psychological safety, and a direct and forceful rebuttal to those who say it cannot be done. The reviewer says the book is just a touch too credulous about ludicrous claims from hype merchants, wanting to believe in Martian igloos conjured from ice and steady rocket supply from Musk, but the evidence is thin. The book works best when it focuses on what architecture can bring to designing the future, quoting the book's argument that a spacecraft is not just a machine but also a home, an office, and a refuge, and that if people are asked to go to remote environments, live in spaces scarcely larger than a few rooms, and perform work of immense complexity and risk, then comfort, efficiency, and ergonomics are not luxuries. Space has to be worth living in; putting people into a tin can with no windows, blank walls, and an infernal hum will drive them mad. The book backs this up with extensive descriptions of engineering challenges of polar research bases, spaceports, and previous craft. The reviewer notes that while the book is rightly scathing about Wernher von Braun's involvement in atrocities and political rehabilitation, it is more muted in criticism of Musk and Bezos, and there is a lot of praise for celebrity architects without examining whether their designs are practical rather than award fodder. Similarly, the book takes on trust that autonomous robots can ingest extraterrestrial soil, process it, and 3D print structures in a hostile environment, but we do not have swarms of drones prefabbing houses in the relatively benign atmosphere of Earth, so the claims may not match reality. On the question of why, A City on Mars points out that mining gold from asteroids would be less profitable than improving Earth mining, while Morrison takes a different approach, claiming space architecture is a forge for creating tools needed at home, including circular systems, low-energy fabrication, modular construction, and buildings that take psychology seriously. The reviewer has sympathy for that but notes that the International Space Station has shown how to recycle water cheaply, yet every modern building on Earth pays only lip-service to reusing grey water, and 3D printing is amazing but the number of structures built using autonomous robots extruding concrete is approximately zero. We have the technology but do not use it. The book is mostly well illustrated with gorgeous drawings of craft and possible inventions, but has no photos or maps. The reviewer finds it endlessly fascinating and up to date, bringing together engineering and psychology, but does not find the Martian habitat built by robot trebuchets convincing, and hopes he is wrong. The review copy came from Netgalley.

The Brain Is Actually Two Completely Separate Organs by Slashdot reports new research led by Stanford Medicine revealing that what we call the brain is two distinct organs that evolved independently over hundreds of millions of years. The human brain consists of two ancient nervous systems packaged together: a more primitive part that regulates heartbeat, breathing, and other functions, and another that makes humans capable of poetry, mathematics, and wondering about origins. The discovery could explain why scientists have struggled for decades to grow certain types of brain cells in the laboratory and opens new avenues for studying devastating diseases affecting the brain stem, including spinal muscular atrophy and amyotrophic lateral sclerosis. The breakthrough came from studying the earliest moments of embryonic development. Study co-authors Carolyn Dundes and Rayyan Jokhai discovered that the hindbrain follows a separate developmental path, running in parallel to rather than branching off from the pathway that creates the forebrain and midbrain. This explained decades of frustration: scientists had been trying to turn one type of progenitor cell into another it is fundamentally incapable of becoming. Armed with this knowledge, the researchers for the first time successfully coaxed human pluripotent stem cells to become functional hindbrain motor neurons in the laboratory. Finally, looking back over 550 million years of evolution, they found the same two-origin brain pattern in chickens, zebrafish, and acorn worms, tiny ocean-floor creatures sharing a distant common ancestor with humans. Jellyfish, which diverged about 600 to 700 million years ago, have two nervous systems at different ends of their body. Researcher Kyle Loh said evolution took two existing neural systems and pushed them together spatially, and having the brain as one organ would probably be more efficient but we rely on this primordial way to make the brain as two separate pieces. The article credits Slashdot reader Beeftopia for sharing.

The next group gathers startup and product developments.

6 days left to get ahead at TechCrunch Disrupt 2026 by TechCrunch is a promotional announcement stating that current ticket pricing ends in six days on September 25 at 11:59 p.m. Pacific Time, and that joining 10,000 or more founders, investors, and tech leaders at Disrupt can save up to $200 on a ticket until then.

A small but growing number of founders are betting on bringing people together offline by TechCrunch begins by noting that Brynn Putnam and Tristan Walker have already done what most founders dream about: created companies so compelling that bigger outfits bought them. The excerpt then describes Putnam, a trained ballerina with a small chain of boutique studios, who built the connected-fitness company Mirror and sold it to Lululemon for $500 million in cash less than three years later. The provided text cuts off at that point, so the rest of the article's argument about offline community is not included.

TechCrunch Mobility: How do we know when an AV is safe enough? by TechCrunch is a newsletter welcome message. It greets readers as the hub for the future of transportation and the role AI is playing in it, and offers a free sign-up for the TechCrunch Mobility newsletter, but the excerpt does not answer the question posed in the title.

Muse by Marginal REVOLUTION is a short product review. The author states that Muse is excellent, with one of the best user interfaces seen on any product, a good model, exemplary speed and clarity, and easy, smooth integration with calendar and email if desired. The review concludes that it all feels very American and gives kudos to the team.

ScrollEd wants to turn textbooks into TikTok by TechCrunch describes ScrollEd as a startup that turns textbooks into a scrollable, Instagram-like feed with video, audio, and quizzes. It is a Palo Alto startup founded by student co-founders and spouses Utsav Gupta and Rebecca Neff, and it pitches at TechCrunch Disrupt.

Vocci’s ring adds a new form factor to meeting note-taking by TechCrunch reports that Vocci's lightweight ring costs $249 and adds a new form factor to meeting note-taking, but the article notes it might pose some privacy questions.

Is the AI industry really ready to slow down? by TechCrunch states that on the Equity podcast, the hosts debated whether AI executives are serious about wanting to slow down.

Elon Musk’s latest Boring Company pitch involves a Hyperloop between Austin and San Antonio by TechCrunch reports that the latest Boring Company pitch is a Hyperloop between Austin and San Antonio, and notes that many of the company's announced projects have not materialized.

World model companies are keeping a lot of secrets by TechCrunch says that everyone in the world-models space is sitting on a pile of cash and a ton of buzz, but good luck getting anyone, from the founders to their own data suppliers, to tell you what they are actually building.

The next article concerns personal finance and AI.

How to Save Money Now with ChatGPT Finances (6 Real Use Cases) | Ethan Bloch by Behind the Craft announces a podcast episode with Ethan Bloch, the product lead for ChatGPT Finances at OpenAI and previously co-founder of Digit, which helped more than 10 million people save over $9 billion. The interview shares six practical ways to use ChatGPT to find wasted spending, prep for taxes, maximize credit card rewards, and model major financial decisions. The six use cases are: first, negotiate or cancel recurring subscriptions by asking ChatGPT to review the last 90 days for subscriptions worth reviewing, price increases, possible duplicate charges, refundable fees, and unusually expensive household bills. Second, find tax savings proactively by connecting bank and investment accounts and asking ChatGPT to proactively audit and optimize taxes, which the guest says is a massive time saver in conversations with an accountant. Third, cut ETF and mutual fund fees. Fourth, maximize credit card rewards. Fifth, model income and taking time off. Sixth, decide whether to rent or buy. The episode also covers how ChatGPT protects financial data, how to set up recurring finance checks, and shipping 47 product updates in one week. The article includes timestamps for each topic and notes sponsorship by Wispr Flow, an AI voice dictation tool.

The final group contains assorted links and a hardware review.

Sunday assorted links by Marginal REVOLUTION presents seven links. First, Nobel laureates in support of the California billionaire wealth tax include Krugman, Acemoglu, Diamond, Stiglitz, Banerjee, and Duflo, from the Financial Times. Second, a piece titled Secretariat as GOAT, from Works in Progress, about the racehorse Secretariat as the greatest of all time. Third, an X post titled "Dead brains and the piano??" Fourth, a quote: "A real contribution of AI is showing the critical weaknesses of humans and human structures," with the added question, but can we take it? Fifth, a speculative piece from the South China Morning Post asking when the Chinese reached New Zealand. Sixth, a quote: "Everyone needs to realize we are fighting the summation of the last 100 years of war that never got settled they just slowed down at times." Seventh, a quote: "you have no idea how fast the world will accelerate AI progress if a world war began," followed by the author's comment that no major war in addition to the status quo would be one excellent move in the direction of AI and other safety.

NVIDIA MMS1X00-N5400 QSFP112 1310NM 400Gbps Optic Quick Look by ServeTheHome is a short hardware review. The article provides a quick look at the NVIDIA MMS1X00-N5400, which are 500-meter QSFP112 modules designed to link devices at up to 500 meters, operating at 1310 nanometers and 400 gigabits per second.

That concludes the evening briefing for September 20, 2026. The window was dominated by technology policy and AI industry secrecy, with a lengthy geopolitical recalibration and notable scientific and book-review contributions.

Your signing session ended