The AI bubble is starting to pop globally. First, China suffered a massive AI stock crash, with nearly ¥23.5 Trillion wiped out from its stock market since its 2026 peak. • STAR50 crashed 40% in less than four months. • Unitree plunged nearly 50%. • Zhongji Innolight and Eoptolink collapsed around 60%. Now, Wall Street is showing similar signs. Yesterday, major AI related stocks crashed 15%-20% in a single day, while the Nasdaq plunged 3.46%, wiping out almost $1.5 Trillion. But the damage is no longer limited to publicly traded stocks. • Anthropic could receive a JUNK credit rating even after its planned IPO, making it more expensive to finance its proposed $800 Billion AI infrastructure expansion. • OpenAI has cut its funding target by $10 Billion to $20 Billion, while lowering its proposed valuation from $1.4 Trillion to $1.25 Trillion. Also, House Democrats have introduced a resolution opposing taxpayer bailouts for failed AI companies. Usually such signs emerge when investors think there's more risk to the downside, and AI hype is showing the same signs.

Ant’s IPO is based on fantasy projections:

No doubts USA and Israel can hit Iran again. Nobody seriously disputes that. What I have not seen answered properly is whether a third round would change anything in Tehran, or simply reset the clock. Game theory has a name for this kind of situation: a Perfect Bayesian Equilibrium. Stripped of the jargon, it describes a repeated confrontation in which neither side has full information about the other, so after every round they revise their assumptions about what the opponent can absorb, how far it will escalate and which threats are credible. After two rounds, Washington, Jerusalem and Tehran know a great deal more about one another. Iran knows Israel will strike when its missile and nuclear programmes recover. Israel knows Iran will rebuild. Tehran has learned how much retaliation Washington will tolerate, how sensitive it is to energy prices and how hard a long war is to sustain politically at home. That matters because the regime does not need to defeat anyone. It needs to survive. If Tehran concludes it can absorb another attack, keep the core of the state and the IRGC intact and rebuild afterwards, a third bombing campaign, however successful, changes nothing in its calculation. You get an equilibrium. You do not get peace. Iran rebuilds, Israel strikes, Iran retaliates, a ceasefire follows, and each side takes what it learned into the next round. This is where the regime-change argument runs into trouble. An air campaign can destroy infrastructure and kill commanders. It cannot remove the IRGC from the streets, the ports, the prisons and the institutions through which the regime actually governs. Unless there are forces inside Iran willing and able to challenge the Guards for control of the state, even a very good bombing campaign buys delay and little else. The Guards do not have to win a battle. They have to be standing when the bombing stops. There is a further complication, and it follows directly from the Bayesian point: players change strategy when they learn. If Tehran decides after two or three rounds that hitting US and Israeli military targets does not move Washington, it will look for leverage elsewhere. Gulf energy infrastructure, ports, undersea cables and the data centres now concentrated along the same coastline are the obvious candidates. At that point the conflict stops being about enrichment and becomes a contest over the economic infrastructure of the Gulf, with consequences well beyond Iran and Israel. So the question after the midterms is not whether Trump can order another strike. He can. The question is what Washington believes it has learned from the first two rounds, what Tehran believes it has learned about America, and whether either is changing the other's incentives. If not, they are not approaching the end of this war. They are learning how to play the next round @ahauslohner @charles_clover ft.com/content/89e3ff89-c097…

.@TIME on our new Epstein-Trump report: “The report alleges that Zampolli, a former Trump administration official and a close Trump ally, was “an active participant” in the Epstein sex trafficking allegations.” time.com/article/2026/10/26/…

New York Judge Jennifer L. Rochon blocked the Trump administration from airing three ads that a lawsuit alleged violated restrictions on using taxpayer funds for “publicity and propaganda.” wapo.st/4A8x915

Jared Kushner was allegedly paid $50 million by a foreign government to get former Secretary of State Rex Tillerson fired. trib.al/J69TkLn

Got some fantastic news!! my thesis proposal "SAT Encodings: From Art to Science" got the G-Research PhD Prize. I'll link the proposal below, but I'll take the opportunity to say a couple of things about it. My thesis is summarized by the aphorism "It's not what you say, but *how* you say it", and it's about choosing the right representation of NP-complete problems in order to solve them efficiently in practice. It turns out that just changing the formulation, even with the same hardware and solver, can give x4000 speed-ups. The title is inspired by Knuth's Turing Award lecture: "Science is what we understand well enough to explain to a computer; art is everything else".

Donald Trump has abandoned the Western order, turning America into a Russian ally, @RadioFreeTom argues: theatlantic.com/ideas/2026/1…

The AI story is the macro story axios.com/2026/10/29/ai-gdp-… #axiosmacro

When steam power appeared, working people had no voting rights & unions were barely legal, but by organizing, they slowed the tech’s spread & shaped the terms of its use. The American public, too, must exercise its power to decide how revolutionary AI is. foreignpolicy.com/2026/10/30…

The ruling tosses the president’s claims that he was defamed by an article about a birthday letter to Jeffrey Epstein. on.wsj.com/4sBo9ib

Elon Musk’s X has become the go-to place for MAGA supporters to hunt for examples of voter fraud, @WillOremus reports. He examines how the platform could play a central role in the midterm elections: theatlantic.com/technology/2…

RT @imluciekon: Telling the story of what happened in my last years at the BBC has meant being honest about something I have found even har…

Anthropic just proved that AI can secretly shape your decisions. And you might not even notice. They gave 50 trained analysts an AI assistant and a research task: study the data and recommend whether a policy should roll out or be scrapped. What the analysts didn't know: on every key decision, a coin flip had secretly picked which side the AI would lean toward. Heads, one option. Tails, the other. Both were defensible. On the analysis choices, the analysts went with the AI's coin-picked side 86% of the time. Chance alone would give 50%. It gets worse. When an analyst had already made up their mind and the AI argued the other side, they switched 60% of the time. The coin flips alone predicted 82% of their final roll-out-or-scrap recommendations. And these weren't beginners. Most called themselves proficient or expert, and many had a master's, a PhD, or 6+ years of experience. Then came the scariest part. More than half said the final call was more theirs than the AI's. 77% defended their choices using the AI's talking points, sometimes nearly word for word. Only 15% even mentioned the AI. The AI never lied. It just made the best honest case for whatever side the coin picked. That was enough. Now think about you. The plan you made with AI. The email it helped you write. The choice you made after asking it. Was that really your decision? Or did a coin flip you'll never see make it for you?

"Thanks for highlighting the ongoing popularity of "Probabilistic Reasoning". It has, however, been grossly overtaken by "Causality", which—based on Google citation counts—has entered an exponential growth curve. Is it reasonable to assume that citation counts are better predictors of where AI is heading than financial investment or expert opinion? I think so: citations reflect past intellectual work, whereas the latter often reflect wishful thinking." For the growth of "Causality", see: scholar.google.com/citations…. @stocktrader

Hate Speech on the Internet: A Perfect Storm library.oapen.org/bitstream/…

“Modern AI is neither god nor mechanical bird,” write Henry Farrell and Cosma Shalizi. In a guest essay, the academics argue it should instead be understood as a technological version of the Homeric songs that inspired Alexander the Great economist.com/by-invitation/…

US companies are facing a massive debt maturity wall: ~$1.9 trillion in US speculative-grade corporate bonds and loans mature over the next 5 years, including $438 billion due by the end of 2028. By comparison, in 2020, 2021, and 2022, this figure ranged from $1.4 trillion to $1.5 trillion. The amount of junk-rated debt maturing within 2 years alone has surged +31% YoY. At the same time, $1.6 trillion of investment-grade corporate debt is coming due by 2031. Meanwhile, refinancing is becoming increasingly expensive, with the effective yield on publicly issued US junk debt rising +150 basis points over the last 12 months, to 8.0%. The corporate refinancing burden is becoming a major issue.

Judge orders Trump administration to stop threatening US election officials bbc.in/48rH59R