So, if I understand the system correctly:

In the top 10, I see projects in Hardware, AI, Psychology, and Economics.

So, the psychology guy goes up the leaderboard by producing results that will impress the people in the other fields. The surest way to do this is to jump to conclusions and ignore everything about sound experiment design.

Similarly, the AI guy is incentivized to train the same neural net on a different game every week, and not to do anything new and risky.

Pardon me, but this does not sound like a good system.