The thing is, you're trying to create a metric that measures an essentially unmeasureable quantity and to predict future outcomes on this. Moreover you're tryi

The thing is, you're trying to create a metric that measures an essentially unmeasureable quantity and to predict future outcomes on this. Moreover you're trying to do this for generally very smart people, so obviously they adapt their behaviour to those metrics (they game the system is a less favourable way of saying this). So the only thing you have achieved that the academics change their work to maximize those metrics. We have seen this with grants as well, to be successful you essentially have to write lots of grant (a period without grants can be career ending), so academics spend up to 50% of their time on grant writing and reviewing. At this point I'm not convinced a lottery would not be a better way of distributing money (and I say that as someone who has been fairly successful at getting grants)