My thinking is that YC's problem is mainly figuring out how likely people are to MSPW and how fun they will be to work with.

I am trying to think out how the co-founder problem differs from the YC problem, but I'm not sure which assumptions to use:

1) The productivity of the startup is mainly the sum of the traits of each individual. Plus, to a lesser extent, the synergies between them.

2) The productivity of the startup is mainly an emergent phenomenon that is difficult to predict from looking at either of the two co-founders alone. Plus, to a lessor extent, the traits of each individual. (For example, the output of someone working with Steve Jobs is different than their output in other circumstances.)

Whether it's more of case 1 or more of case 2 matters when figuring out in which direction to take the community, because in each case you are going to want a different systemic design to bring out the different salient factors. (Of course it could actually be case 1 for some startups and case 2 for other startups, which would make the problem even more difficult.)