Week 10 · 2025-03-03 → 2025-03-09 · 10 newsletters

Regime Change and Red Pens

markets-regime-change · ai-and-knowledge-work · writing-and-effort · aggression-and-inner-work

The first full week of March, and the inbox was thin: nine substantive emails across seven days, no dominant news story, no breaking thread. What signal there was clustered around four through-lines: a markets regime change in U.S. equities, a turn in the AI conversation toward how to measure progress in domains without right answers, two writers arguing in different ways that compression and effort are the actual craft, and one essay on aggression as a contagion that can only be broken from the inside. A profile, a vintage Bill Gates email, and a short investing roundup filled out the rest.

Markets: A Regime Change in U.S. Equities

The week's one piece of hard news writing was The Last Bear Standing on what looks like a real shift in the U.S. equity tape. The setup: after two red weeks, the reflex rally has stopped showing up. Trump's tariff announcements on Sunday soured the open, the subsequent reversals failed to rebuild confidence, ADP reported the weakest payroll gains in seven months, the Challenger report logged the largest monthly job cuts since COVID as DOGE moved through the federal workforce, and the Atlanta Fed's Q1 GDP nowcast sat at negative 2.4%. The technical read is the part worth holding onto: the S&P 500 closed on its 200-day moving average for the first time in eighteen months, and the Nasdaq 100 broke its 200-day for the first time in two years. The framing of daily tariff flip-flops as "less 3D chess and more flailing ad-lib" is the kind of line that ages well.

The companion read on the investing side came from Investing Insights, a short roundup anchored on Morgan Housel's "Pure Independence" essay and a piece on buying at low valuations. The Housel line worth carrying: "a lot of financial mistakes come from decisions that would be right for someone else but wrong for you." The Charlie Munger three-rule frame ("don't sell something you wouldn't buy, work for people you admire, partner with people you enjoy") is the version most operators forget when markets are doing what they did this week.

The take: the bear case is no longer a contrarian position, and the technical breakdown plus the macro deterioration plus the policy noise is the kind of three-way confirmation that usually marks a regime change rather than a dip. If you only track one piece this week to revisit a quarter from now, this is it.

AI and Knowledge Work: How Do You Grade an LLM Doing History?

Mark Humphries at Generative History wrote the week's most useful AI piece, and it is a piece almost no one else was writing in early March: how do you measure LLM performance in the humanities and social sciences, where there are no right answers and the existing benchmarks have run out of room? The framing is that math, science, and coding benchmarks are getting saturated (models are all becoming A+ students on the standardized tests), but there is no equivalent measurement for qualitative analysis, argument, use of evidence, or prose style. Without that, the public AI discourse splits into two unfalsifiable camps: pundits reading plateauing benchmarks as proof of a bubble, and accelerationists reading the same data as proof of imminent AGI. Humphries flags Ezra Klein's recent "The Government Knows AGI is Coming" column as the cleanest summary of the serious-people-inside-the-labs view.

The piece does not resolve the question, which is the point. The argument is that the absence of good qualitative benchmarks is itself the story: it makes thoughtful disagreement impossible, because nobody can show their work.

The take: this is the right frame to carry through 2025. The interesting AI debate is not "is it a bubble or AGI" but "what would it take to actually measure progress on the tasks knowledge workers care about." Until someone builds that benchmark, the discourse stays stuck.

Writing and Effort: Compression Is the Craft

Two pieces from very different writers landed on the same argument from different angles. Ami Vora at The Hard Parts of Growth wrote about a former manager's "red pen trick": he printed her work and crossed out anything unnecessary, no inline questions, no smiley faces, just brutal cuts. The lesson she pulls is that compression forces commitment. "When my ideas were surrounded with filler, it was hard for the reader to know what the real point was. But when I distilled my message down to just a few, carefully chosen words, the ideas cut like a sharpened knife." The deeper note is that hedging language ("I think X happened" when X clearly happened) is fear of being wrong dressed up as humility, and that cutting it is the work.

Rob Thomas at The Mentor ran a shorter post on the same instinct from the talent-versus-effort side. The Tim Notke line at the top ("hard work beats talent when talent doesn't work hard") is the setup; the payoff is that sustained effort produces what looks from the outside like talent. The frame that stuck: "what looks like natural talent is often just the visible result of invisible work."

The take: Vora and Thomas are running the same play. Vora is arguing that the visible work of writing is mostly the invisible work of cutting, and Thomas is arguing that the visible performance is mostly the invisible practice. Both pieces are about what the audience never sees, and both are arguments against the genre default of dressing up the surface to hide the absence of work underneath.

Aggression and Inner Work: The Hardest Practice

Abby Falik at Taking Flight wrote the week's most reflective piece, "Aggression is contagious. Courage is, too." The frame is that the world has become "a stage for unprocessed fear," with leaders from Moscow to Mar-a-Lago performing dominance as a substitute for leadership, and that we are still mistaking the theater for the thing. The hinge of the essay is the Tibetan Buddhist concept of shenpa: the chain reaction when something pushes a button, chest tightens, heat rises, breath thins, and we start swinging before we have noticed we have been hooked. Falik's argument is that breaking the cycle is "an inside job," and that the practice of pausing before the reaction is the only intervention that actually scales. The Brené Brown distinction she carries through ("getting it right, not being right") is the operational version.

The take: this is a sermon, and it knows it is a sermon, and that is fine. The piece earns its register because the framing of strongman politics as collective shenpa is genuinely useful, and the suggestion that the only intervention is at the level of the individual nervous system is a harder argument than the usual political-essay defaults.

Grace Notes

A few smaller pieces worth a mention. Tanya Windman ran a profile of Kerry, the founder of Taka and Taka Table, describing her as a "cultural chameleon" who moves between writing, design, and cooking with what Windman calls a "steady and sustainable beat." The Taka origin story (a fashion-industry veteran building well-crafted essentials as a response to overproduction, then extending the same philosophy to food through Taka Table) is the kind of founder profile that reads like a manifesto without trying to.

Zarik Khan at Fintech Compliance Chronicles closed his 2025 fintech preview series with a deep dive on mBridge, the BIS-spearheaded CBDC effort connecting the People's Bank of China, the Hong Kong Monetary Authority, the Bank of Thailand, the UAE central bank, and the Saudi central bank for real-time cross-border payments. The piece is dense and the implications for SWIFT are real, even if the timeline is long.

Internal Tech Emails surfaced a November 2003 email from Bill Gates to Will Poole, written less than two years after the iPod launched, in which Gates asks "where the scenario thinking is in Windows" and notes that "because we are going to be so late with a music service we are going to be behind others almost forever it seems like." It is a small artifact, but the self-awareness about being permanently late to a category is the kind of thing operators should reread when they feel themselves rationalizing a slow start.

Three Takeaways from the Week

The U.S. equity tape changed character this week, and The Last Bear Standing is the cleanest read on why. The convergence of a technical breakdown (S&P on the 200-day, Nasdaq through it), a macro deterioration (weak payrolls, Q1 GDP nowcast negative), and policy noise (tariff flip-flops, DOGE cuts) is the kind of three-way confirmation that usually marks a regime, not a dip. If the tape recovers next week, this read still ages well as a marker.

The AI conversation has a measurement problem, and Mark Humphries named it before most other writers. The fact that we have no benchmarks for the qualitative tasks knowledge workers actually do is why the discourse keeps splitting into bubble-versus-AGI camps that cannot argue with each other. The interesting work for the rest of 2025 will be on the benchmark side, not the model-capability side.

If you only revisit three pieces from the week, I would suggest The Last Bear Standing on "Regime Change" for the markets read, Mark Humphries on "Stochastic Canaries in the Coalmine" for the cleanest framing of where the AI debate is actually stuck, and Ami Vora's "red pen trick" for the piece on writing that is really a piece on commitment. A thin week, but the signal that was there was unusually concentrated.