[saintgimp]

No, he's saying that "merit" is an incredibly easily-gamed metric. We can measure results all right, and we can measure ability rather less so, but "merit" implies a casual link between ability and results that is very hard to unambiguously prove and shockingly easy to fake (even if just by disregarding beneficial external factors).

[rlucas]

More importantly: creating a scalar metric of "merit" is always an exercise in power, because it simplifies a hugely multidimensional space to a scale, and in collapsing all those dimensions expresses in a hidden way the preferences of the definer of that metric.

This happens even in notionally quantifiable domains like finance. What asset gives the best return? Should be easy, right? Well, what about volatility? Oops, turns out the naive rank ordering of assets by expected return contains an implicit value judgment about volatility.

See upthread for a mention of eBay seller scores being unadjusted for transaction size for another example.

[nickysielicki]

Good comment. Thanks.

I just wanted to reply and say that you and saintgimp have changed my opinion.