Unequal splits are an invitation to resentment and the degraded communication and problem solving that comes with resentment. If you've done a bunch of startups, you've probably had the experience of having a harrowing argument with cofounders, and the realization of how easily your company can disintegrate based on nothing more than silly interpersonal conflicts. Some people, myself included, take from this the lesson that you should rig your startup to minimize these issues as much as possible.

Another reason to avoid unequal splits is that they often presume you can see into the future. Who's to say that whatever you did to "earn" an extra 10% of the company isn't going to be more than offset by something your cofounder does later?

"Obviously want to optimize for success."

No, you're optimizing for "fairness" as you see it, for yourself. This split will do nothing good going forward ... unless, of course, you will internally burn at the idea of your cofounder getting more than what you view as his fair share. Which should tell you something, and brings me to:

The pitfalls are inside people's heads and hearts, so I think you'll find that strategy to be particularly difficult to execute.

The cofounder should be obvious; others who, especially down the road when the starting details are lost in time, will ideally see you two working just as hard as each other, and wonder why the split is so uneven. I don't see anything good that can come from that.

This data somewhat contradicts that however, as it says unequal shares are correlated with higher initial funding valuations (obviously that's not a perfect metric) http://siepr.stanford.edu/system/files/shared/pubs/hellmann%...

I'd really like to see the data which suggests equal shares lead to better success outcomes (maybe YC has this?).

My guess is that the numbers would swing sharply in the other direction if the study accounted for survivorship bias.