The article diagnoses the problem well - Campbell's Law shows how any metric used for selection gets gamed. But randomness isn't the only solution.
The issue isn't meritocracy itself, but our implementation. Current systems fail because "merit" is cheap to fake. LinkedIn profiles, smooth talking, and connections matter more than actual performance.
What if merit claims required real stakes? If claiming expertise meant risking something you'd lose when proven wrong? If your surgical reputation couldn't boost your investment credibility? If gaming the system cost exponentially more than being honest?
Yes, KPIs fail for complex work. But a surgeon with 1,000 successful operations IS more qualified than a random person. That signal has value. Rather than abandon merit for randomness, we need merit systems that are expensive to fake and cheap to verify. Make the track record immutable, domain-specific, and consequential. The technical challenge is hard but solvable. Randomness might help for some positions (jury duty works!), but wherever specific expertise matters - engineering, medicine, research - verifiable performance still beats random selection.
I've been working on a system exploring these ideas [1], but the core insight stands regardless: the author's claim that only randomness can prevent meritocratic decay may be premature. We might just need better verification mechanisms.