Remix.run Logo
lwansbrough 4 days ago

Biggest hurdle for me to do this is just multiplayer games. I wish Linux would offer a solution to that. No idea what it would look like though.

Contrary to most Linux advocates I’m a big believer in giving studios the tools they need to defeat cheaters and I don’t care much about system integrity if it means fairer games.

ajvs 4 days ago | parent | next [-]

The anti-cheat creators other than Valve aren't bothered to invest into making a Linux kernel anti-cheat, and most Linux users would be unwilling to allow one to be installed either.

paulbgd 4 days ago | parent | prev | next [-]

Totally agree, but it seems like competitive games have solved it. CS2 (VAC), The Finals (EAC), and Overwatch 2 (Warden) all run flawlessly on Linux.

nopcode 4 days ago | parent [-]

Those all have poor reputations. There's a reason why ESEA/FACEIT has been around all this time.

zamalek 4 days ago | parent | prev | next [-]

The only sure-fire way to defeat cheats is with something like Counter Strike's overwatch system: have humans vet replays. Cheats are a ludicrous business, there is simply far too much incentive to defeat software-based systems.

wiredpancake 4 days ago | parent [-]

[dead]

charcircuit 4 days ago | parent | prev | next [-]

>No idea what it would look like though.

It looks like attestation. Linux needs to be able to assure game developers that the kernel their game is running on is actually protecting the security of their game.

bee_rider 4 days ago | parent | prev | next [-]

Well, the beauty of Linux is that anyone can go implement whatever features they want. But, I’m very happy that folks aren’t very interested in supporting this kernel level DRM stuff.

srjek 4 days ago | parent | prev | next [-]

> I wish Linux would offer a solution to that. No idea what it would look like though.

It probably would have to be an isolated environment to run in. Something like the Secure VM efforts adopted for desktops, perhaps with a small trusted hypervisor instead of CPU vendor extensions. Anything else I can think of starts to restrain what software you can run on your machine, or becomes highly invasive in ways similar to Anti-Cheats on Windows, both of which would be rejected by the general Linux community. (Through, it's not like anyone was asking Microsoft either before implementing anti-cheat and trampling on system integrity, at least until Microsoft started requiring signed drivers)

However, given that a generic blackbox implementation enables DRM and binary encryption there will probably still be opposition. It gets particularly nasty if it's given access to something like a full TPM to unlock application data in the same way a TPM can unlock an encrypted drive for your OS. That would make it the penultimate closed source application, which is really anti-ethical to a number of communities. (open source, modding, game/app preservation...)

coppsilgold 4 days ago | parent | prev | next [-]

Even on Windows they are losing despite the invasive anticheats.

I suspect the answer to cheating will ultimately be big brother and hiding information from the client.

The server should stop sending positions of undetected enemies - this requires rethinking game engines due to the predictions they perform.

The server should log every single action by every single player (full replays) in perpetuity, train models on it to detect outliers, classify some outliers as cheaters and start grouping them all together in lobbies.

Another idea would be to conduct automated experiments on players at random. Such as manifesting "fake" entities behind cover and measure player reactions - of which there should be none. Spawn bots (from the beginning of the game) that a compromised client (cheats) cannot discriminate from players and have them always remain in cover and gauge player behavior relative to them, despawn them if a [presumably real] player is about to detect them.

It all requires work and imagination which is in short supply in the industry. But given how cheaters kill certain types of games maybe someone will eventually do it.

zamalek 4 days ago | parent [-]

> I suspect the answer to cheating will ultimately be big brother and hiding information from the client.

The speed of light makes this _marginally_ problematic to do. It is possible that a unit might move out of the fog of war, or out of cover, during the latency to the client (or between server ticks). You'd effectively have pop-in during some scenarios - but it would be minor and the net benefit would probably make it worth it.

I recall one of the MOBAs adding this during its lifecycle, HoN I think?

ThatPlayer 4 days ago | parent [-]

Riot has a pretty good article about why it is hard: https://technology.riotgames.com/news/demolishing-wallhacks-... . Anything more than simple geometry makes server side calculations costly, especially when you're targeting 128 tick/s.

Their settled solution is still not perfect, hence still the need for client-side anti-cheat. The final video clip is definitely done to look effective than it actually is. Those positions are transmitted based on space, not time, and in a real game you'd be moving slower.

Mobas generally have a lower tickrate and simpler vision setups

coppsilgold 3 days ago | parent [-]

Looks like they put in some effort. You will get better results by having this type of anticheat baked into the engine from the very beginning and requiring a GPU on the server.

What I had in mind is having a primitive physics simulation with point clouds (+ velocity smearing) for entities and geometries for surfaces. You will be able to to do many more checks this way.

umanwizard 4 days ago | parent | prev | next [-]

Who exactly is "Linux"? What entity, specifically, would work on kernel anti-cheat? The only realistic company who would care about this is Valve. So really you should say Valve, not Linux.

That's the biggest problem with Linux on the desktop: outside of Red Hat and Canonical (neither of whose business has anything to do with gaming), there is basically no well-funded company that cares about it at all. Linux already works great for the use cases that matter to the people who develop Linux, who mostly are not trying to compete with Microsoft or Apple.

broodbucket 4 days ago | parent | prev | next [-]

The anticheats themselves typically do support Linux, it's the devs that don't choose to use them

lwansbrough 4 days ago | parent | next [-]

Well EAC for example is user space only because it has to be, which some games decide is not an acceptable level of security.

ThatPlayer 4 days ago | parent | prev [-]

Those are generally not the same anticheats with the same levels of functionality. As an analogy it's like saying Excel supports iPad. Or a gaming example that used to be way more common: Tony Hawk Pro Skater 2 is supported on Game Boy Advance.

It's a game and it is Tony Hawk, but it's not really comparable as Tony Hawk on PS1.

chrneu 4 days ago | parent | prev | next [-]

they have the tools they need to defeat cheaters, they just choose to go about it in very invasive and lazy ways because people still buy their product.

then people complain when the product sucks and is invasive.

mvdtnz 4 days ago | parent [-]

This is just factually incorrect.

aaomidi 4 days ago | parent | prev [-]

Plenty of competitive multiplayer games run on Linux fwiw.