A 50/50 split can work for the (very) small subset of cases where two people of comparable skill and commitment start building something from scratch (nothin' but a half-baked idea) together, with no prior investment/work/IP, no domain expertise, no key contacts, no customer channels, nor any major capital infusions. Oh, and both parties have a clear record of making good decisions together and resolve disputes effectively.
But for most other cases in the real world, these two resources offer a more rational and open/honest approach:
http://founderresearch.blogspot.com/2006/01/splitting-pie-fo...
http://www.andrew.cmu.edu/user/fd0n/35%20Founders%27%20Pie%2...