[Week 33, 2025] Compounding and Competing

Welcome back to Price and Prejudice with a few musings from Week 33 of 2025.

The Longevity Problem

This article lays out a simple but destabilizing point: as life expectancy keeps climbing, the usual rules of retirement planning stop working. Target-date funds now hold over $4 trillion and follow a familiar path—heavy stocks early, then a slow shift to bonds. But that path was designed for 20-year retirements, not 35-year ones. If a healthy 65-year-old today could plausibly live to 100, then the idea of “decumulation” gets murky. The article even suggests staying mostly in equities after retirement.

Broadly, if you expect to live much longer, there are broadly two strategies: (1) change how you invest for retirement, or (2) change how long you work. The first means saving more, shifting portfolios, maybe having children who’ll support you (the old-school version of an annuity). The second means delaying retirement and working deeper into life. FIRE tries to retire early and still live forever, but it runs into the awkward constraint that compounding works better when you aren’t spending the returns.

From a social planner’s view, the trade-off looks different. Pushing people to save more today can suppress consumption and slow demand, especially if large cohorts do it at once. But asking people to work longer assumes a labor market flexible enough to absorb older workers and a healthcare system that keeps them employable. Both are plausible, but neither is free.

The Quant Temptation

The entry-level quant isn’t hard to describe: smart, competitive, and happiest when they’re solving hard problems that also pay well. Day to day, that usually means hunting for signals that survive out-of-sample tests or squeezing returns from data others overlook. It’s not glamorous, but it’s an unusually direct way to turn math into money.

Perhaps it’s no surprise, then, that this same profile is now the prime target for AI labs. As the Bloomberg article describes, firms like Anthropic and OpenAI are wooing quants with rooftop mixers and the pitch to “build the machine god,” offering salaries and equity that rival Wall Street. The work looks familiar, the compensation is comparable, and the story is better—at least until “build AGI” turns out to mean “optimize ad targeting.”