I'm doing some research on input-driven (example: "I've been pressing the forward button for 60ms and have held the fire button for 100ms") vs position-driven (example: "I'm at coordinate 123,456,78 facing 90,10 moving at a velocity 2,3, and I'm shooting player 8 in the head") client messages and trying to figure out the pro's and cons of each. This would be for a shooter game. I think most engines use the input system - anything based on Quake/id tech (Counter-Strike Source, Call of Duty, etc.) does. I think Unreal 3 uses this as well, but I haven't been able to find solid details. Halo 2 and up sound like they use the position-driven method. Maybe FEAR did as well?

Is anybody familiar with both of these methods and the drawbacks of each? What other games have used the position-driven system. Which system would you use if you were making a shooter?

The upside of the "input-driven" system is it's harder to hack. The downside is it requires more simulation on the server.

If you're on an open system like the internet, you probably want that.