in a FOSS game anybody can modify the game client all they want, so all the bullshit is out of the way from the start. You can’t hide behind make-believe notions such as “they can’t modify the client” – which is one of the major lies and fallacies of commercial close-source games.
Sometimes, just for practical performance reasons, with realtime games, the client is gonna need access to data that would permit one to cheat. You can’t do some game genres very well while keeping things on the server.
Consoles solve this by not letting you modify your computer. I think that if someone is set on playing a competitive game, that’s probably the best route, as unenthusiastic as I am about closed systems. The console is just better-aimed at providing a level playing field. Same hardware, same performance, same input devices, can’t modify the environment.
'Course, with single player games, all that goes out the window. If I want to modify the game however I want, I should be able to do so, as it doesn’t hurt anyone else. I should be able to have macros or run an FPS in wireframe mode or whatever.
For PC competitive multiplayer, in theory, you could have some kind of trusted component for PCs (a “gaming card” or something) that has some memory and compute capability and stores the stuff that the host can’t see. The host could put information that the untrusted code running on the host can’t see on the card. It also lets anti-cheat code run on the card in a trusted environment with high-bandwidth and low-latency access to the host, so you can get, for example, mouse motion data at the host sampling rate for analysis. That’d be a partial solution.