[evanwarfel]

Ah, perhaps. Our systems will only improve to the extent that someone takes the inherent issues on as personal problems.

And in some sense, what we are saying is that systems with metrics that are only shallow proxies for actual success (citations, valuations) often fall into multi-polar-trap type situations [1].

I fully believe it's possible for new(ish) models to outperform Academia and some of the sillier SV practices. It's a question of getting things right in the beginning.

[1] http://slatestarcodex.com/2014/07/30/meditations-on-moloch/