Ten Days That Didn’t Shake the World: Is it 1905 in AI Time?
Will 2026 be one of those grand historical years that change the world — like 1917, 1789 or 1968? Not according to Keith Teare, publisher of That Was The Week newsletter and co-host of our weekly tech roundup. For Keith, the best historical analogy is 1905, the year of the first abortive Russian revolution. The year that didn’t change the world.
Keith’s latest tech newsletter asks “What Time Is It?” His answer is that we have “multiple clocks” — micro and macro, short, medium, and long term to make sense of our current AI moment. This week, for example, OpenAI and Anthropic both shipped work-focused products, and most of the world hasn’t noticed. Thus his allusion to 1905. We are on the brink of massive change. But nothing is going to change. Not quite yet. Until everything does.
Five Takeaways
• It's 1905 in the AI Economy. Keith's answer to the what-time-is-it question is the failed Russian revolution — the moment when the variables of transformation were all in motion but nothing was yet visible, and which took seventy years to fully play out. AI's radical change is real, he argues, but it is being experienced by a small number of people and is not yet generalized through the economy. The evidence of the week: OpenAI and Anthropic both shipped work-focused products — and most of the world shrugged.
• The Socialist Temptation of Slippery Sam. The Wall Street Journal frames Altman's offer of 5% of OpenAI to Washington as socialism creeping into Silicon Valley. Keith — who hated the word even when he was a communist — says the term has been Americanized into meaninglessness: it now just means the capitalist state doing more. What Altman is actually proposing is capitalism's end game — a sovereign wealth fund holding equity in the companies everybody wants to fund, so that private wealth creation reaches the point where everyone can imagine benefiting from it. The precise opposite of British Leyland.
• The Multiple Clocks. Keith's framework sorts the week's flood of AI news into micro and macro issues running on short, medium, and long-term timelines. At the micro-short corner sits deployment friction: Microsoft and Amazon spending billions on forward-deployed engineers, and Apple suing OpenAI. In the middle, work adapts — the human as the driver of AI rather than AI imposed on humans. At the top sits Arvind Narayanan's idea of AI as a “normal technology,” which deflates hysteria without deflating importance: electricity was a normal technology too, and it still changed everything — just slower than its loudest advocates expected.
• Abundance and Its Discontents. Matt Yglesias argues that saving capitalism requires radical land use reform, which reignites the show's longest-running argument. Keith's case: the Elizabeth Line and the congestion zone have redefined London, multiplying its effective land fifty-fold, and a house twenty minutes from the center can be had for a couple of hundred thousand pounds. Andrew's case: prices haven't fallen, London is more expensive than ever, and free is doing a lot of work as “a tendency, not an achievement.” The quarrel is adjourned until next week, with Keith cheerfully moonlighting as a real estate agent.
• Two Americas — and the Small Stuff. Ivan Krastev tells Yascha Mounk that American exceptionalism ran roughly from 1850 to Vietnam and has been replaced by defensive preservation — MAGA as a reaction to decline rather than a vision. Noah Smith's version: America can't build a passenger train, yet its AI industry is upending the world. And against John Battelle's worry that digital life has lost the plot,
