Currently fatigue indicates the percentage that certain attributes are lowered during a game. So those attributes would be lowered to 80% when the player's fatigue is at 80%. The amount a player fatigues during the game is dependent on his position and his Stamina. It's not a flat decline, so it may stay relatively flat for a period of time and then all of a sudden start to drop off. If you have very short drives on offense or defense, you might not see your players fatigue much at all during the game because any fatigue they might acquire might be wiped out as they rest between the drives. You generally will not see players get very fatigued unless they stay on the field for longer drives. So that's the pattern of how fatigue works currently in the engine.
The biggest problem isn't the rate at which players fatigue or how far down they go, but the general issue of randomness in the game right now. If a fresh player can have the same results as a fatigued player, and results are produced based on simple random chances, then fatigue may appear to be meaningless and not affecting the results at all. What's happening is that the fatigue, by way of lowering certain ratings, is lowering the chances of success for those players. But then the randomness can end up countering those hits to the chances and you end up not really affecting anything at all.
For the update, the major focus will be on minimizing the randomness. I don't know exactly what that will entail right now, but we've been working through some ideas that should help out. It might also be that I look directly at fatigue for some cases rather than just adjusting ratings, but I'm still looking at that. However, anything we do to minimize the randomness should help in making fatigue matter more. We might also look at the fatigue rate, but I'd rather check that after we get a good idea of how fatigue will affect the engine with the new changes.