Our first project was a mobile game we called Marco Polo. It was a fun idea but the project itself ran afoul of some issues:
1. Horizontal slices are a bad way to do games. Building out a system completely before building any of the things it attaches/talks to causes you to build on top of wrong assumptions that you don't realize until its too late to go back and change things.
2. Get your game in the hands of outside people as soon as it is feasible playable. There were some pretty glaring things that were caught after having some friends test Marco Polo, but this way down the line and realistically they could have been caught much earlier if we'd worked on getting something playable first and expanded from there.
Both of these things have at their core the same thing, assumptions. It's impossible to start a project without some base assumptions, or else you may get stuck in analysis paralysis trying to plan for every eventuality. The idea is to minimize the effect these assumptions will have if they are wrong.
Gamedevs, what ways to you reduce the impact of assumptions in your work?
Gamers, what developer assumptions have ruined or hampered a game experience for you?