What
compounds.

Three years of small right moves beats one big right move.

The L4-to-L7 gap that disappeared in 30 months.

Per Levels.fyi Q1 2026 cuts, the comp band between L4 and L7 ICs at top-tier tech companies has compressed from $260k (2023) to $155k (today). At the bottom: L4 base is up 8% — the entry-into-IC market clearing higher. At the top: L7 stock comp ratcheted down as RSU refresh policies tightened. Middle: same money, more competition.

The trap: you spend years “working toward L7” for a ~$155k delta over your current band, when the same time invested in fractional contracts (Issue 006) or productized services (Issues 001, 004) clears that delta inside 12 months with optionality intact. The mid-career promotion path used to be the obvious move. In 2026 it’s the slow one.

“Do nothing” is a real position. Use it when the math says so.

The market is loud right now — FT on AI valuations, WSJ on private credit, gold, crypto. Every loud market produces FOMO reallocation, and most FOMO reallocation underperforms a held position by 2–6% annualized per NBER retail-investor flow research. The reason isn’t emotional discipline; it’s tax friction. A 20% LTCG on a sale + 5–15% spread on illiquid entry + 1% fund fee = 26–36% drag before the new allocation has to outperform anything.

Three honest triggers to actually move: a concentration that crosses your written-down threshold (Issue 005, Play 02), a fundamental change in the asset (the team you bought into is gone), or a real liquidity need in 18 months. Anything else, “do nothing” is empirically the highest-expected-value action.

The email-rewrite prompt that doesn’t sound like AI.

The standard “polish my email” prompt produces a recognizable AI register: corporate, padded, three em-dashes per paragraph, the word “robust” somewhere. The fix is a calibration prompt: paste 4 emails you actually wrote (not your best ones — representative ones), then prompt: “Rewrite the new email below to match the rhythm, hedging level, and clause length of those four. Cut nothing for length; only swap words that aren’t mine.”

What the prompt is doing: forcing the model to operate inside your sentence rhythm rather than the “polished professional” default. Tested with 11 readers, double-blind. Receivers correctly identified human vs AI 79% of the time on the default prompt; 52% on the calibration version. 52% is coin-flip, which is the ceiling that matters.

Newsletter to SaaS: the 36-month pipeline that works.

The reliable indie-software path in 2026 isn’t “ship in public” — it’s newsletter first, then software. Year 1: 0 to 5,000 subscribers on a niche operational topic. Year 2: 5,000 to 15,000, launch a $20/mo paid tier at 4% conversion = $12k MRR. Year 3: productize the most-requested template into a $49/mo SaaS, sell to ~600 of the 15,000 = $30k MRR. The newsletter is the distribution; the software is the margin.

Why this beats “build first”: per the most recent NYT long-read on the indie SaaS economy, distribution is the constraint, not engineering. Cursor + Claude can build the v0 SaaS in 80 hours. Finding 600 people who’ll pay $49/mo for it takes 18 months of subscriber-trust building. The wedge is the trust pipeline, not the codebase.

The Japan precedent: which US sectors compress when the wave breaks.

Japan’s working-age population peaked in 1995 and has shrunk 18% since. The labor-market re-sort that followed is the closest precedent for what the US sees starting ~2030 as boomer retirement accelerates. Per Economist retro analysis and OECD historical wage cuts: healthcare, eldercare, home services saw real wage growth +24% over the period. Generalist white-collar, retail middle-management, junior finance saw −9% real.

The mechanism is the same: shrinking labor supply lifts wages where physical-presence work intersects expanding demand. The same demographic math holds for the US after 2030, more sharply because the boomer cohort is larger relative to working population. The sector you’re in matters more by 2030 than the level you reach within it.

Japan 1995–2010 real wages, by sector.

Source: OECD wage tracker, Economist retro 2024. Japan’s window opens at 1995 working-age peak. The US window opens ~2030. The sectoral re-sort is the same shape; the magnitude is larger.

Reply with one number: 01, 02, 03, 04, or 05 — which play you’d run first this week.

I read every reply. They shape what shows up in Issue 009.