Jason Cohen – Yes, but who said they'd actually BUY the damn thing?:

"In a big software project do you tackle the high-risk, ill-defined stuff first, or do you postpone that to the end? Obviously you address the unpredictable stuff first — most of the project risk is due to the unknown, so the earlier you can sort out uncertainty the more time you have to deal with the consequences.

I'm making the same argument, except the "high-risk unknown" is "everything that's not code." Your code will be good enough; it's the other stuff that will probably sink your ship — unable to find customers or unable to convince the target audience they should open their wallets."