5 ms·

If linux gaming requires you use a blessed immutable OS where everything you run is either unprivileged or signed by a central entity, it's not much better than

If linux gaming requires you use a blessed immutable OS where everything you run is either unprivileged or signed by a central entity, it's not much better than just using Windows at that point. Kernel level anti cheat is a failure. You need to control the hardware to do it properly. The only hope is to wait for less locally intrusive and more robust anti cheat solutions. Basically the only possible solutions at this point lie in the AI space.

advael 2y ago
Jokes I think there are three fundamental categories of cheating threat models that actually matter: State poisoning, Information leakage, and Input automation State poisoning means your game was poorly written, period. Either that's a vuln within the code itself or badly implemented netcode. A 2-player game can have total asynchronous client separation and still be peer-to-peer. A more-than-two-player game is almost always run on a server that serves as the single source of truth. In either case, a game that doesn't make the fundamental guarantee that the inputs available to a player and maybe some initial random seeds are the sole determinant of the gamestate have no hope, and rootkitting your computer because they wrote their game's statemachine or interfaces like shit is not the correct solution. If your answer to this is that big game studios shouldn't have to learn how to write more solid code, this means that the sanctity of their game isn't that important to them, not that they should get to root your computer Information leakage may be somewhat harder. Often you want the simulation to be running client-side, so a naive model of netplay would have the full state available to all clients from a technical perspective... but this doesn't have to be true. In most cases, you can do partial state with rollbacks to make it much harder to cheat from a technical perspective, even making no guarantees about the clients themselves. I think even when this is hard, the correct path here isn't rootkits, it's approaches that start to approximate zero-knowledge proofs. This also means there's a rich literature of zero-knowledge proofs to draw on Input automation, to be honest, is basically hopeless to prevent upfront regardless of what you do. If you can plug external hardware into your device at all, you can rig up something that automates your inputs. This can be hard to even verify in person, let alone through even a rootkit. I don't personally think it's worth worrying about that much, but if you care about macros and the like, it's really difficult to prevent. However, if there's money on the line or something, there are good analytic forensic techniques to detect this kind of cheating after the fact. Maybe this is where "AI" could actually help, as some kind of sequence-based anomaly detection that can run in real time might be able to detect unusual input clusters, but I worry that the false positive rate is going to be super high. Honestly seems like a lost cause. But crucially, not a lost cause that you get around via compromising the OS at a kernel level Anti-cheat that "needs to own your kernel" is more user-hostile corporate bullshit. Most games work fine on linux, but frankly no game is worth a rootkit, and no game needs one. The fact that some companies demand it should be viewed as those companies trying to scam you. That's not how the security of anything on the internet works. It's only how security of a bunch of mobile stuff works because Microsoft has trained generations of otherwise smart people to believe their total lies about security, and Google and Apple have taken advantage of this to secure a massive amount of control and surveillance over everyone who owns a smartphone (Which is increasingly required because they've also convinced people that fake 2FA that's just your phone as a single source of identity that can in fact often effectively be 1FA because it can override other authentication methods in most cases is somehow secure. The fact that everyone has a device with a bunch of proprietary backdoors that they don't have root on and that serves as a single lynchpin through which their life can be ruined is the most fundamental destruction of personal computer and identity security that's ever been realized - to say nothing of privacy, and that's a huge accomplishment given all that Microsoft, Facebook, and Amazon have done and still do to compete for the title) Giving a corp a backdoor to your computer doesn't secure anything except that corp's ability to fuck with you. Don't believe Microsoft, Apple, Epic Games, The NSA, or anyone else who tells you that the best way to secure something is to give them a backdoor. Fuck all those people. They have not only gotten their slimy tendrils in a ton of people's stuff through these lies, but have propagated bad information about how to do security to a ton of organizations. If someone who works at one of these scummy companies or agencies responds to this with some condescending corp-speak at me, I've got a bunch of work to do so I'll probably not get to you immediately, but I pre-emptively say that making this argument at all fundamentally undermines your credibility, and also I hate you on a personal level. You've been a spook too long and it's rotted your brain, hypothetical internet stranger who might not even exist, quit your job and fix your heart Basically, don't believe any of this "We have to own your computer for your own good" nonsense. That's a scam. Every time. Also, proprietary software should be assumed inherently insecure by default, not the other way around. A better world is possible
- Arch-TK 2y ago
  The reason companies seem to bother at this point is that, by implementing increasingly intrusive anti-cheat, they force cheaters to be increasingly subtle. With sufficiently intrusive anti-cheat you end up with gamers believing that the game they're playing has no cheaters. This currently happens in at least some of the games which utilise kernel level anti cheat, as demonstrated by numerous videos on the topic which also shed light on massive communities of cheaters who just end up buying or making their own hardware based cheats.
  - advael 2y ago
    Gamers will believe anything a company they like tells them. The fact that these companies can attract arguably the most insufferably delusional audience of frothing bootlickers to ever walk the earth to defend them doesn't justify their decisions on a technical or an ethical level If the explanation for why a company needs a rootkit is "they don't want to spend effort on a better solution" that means that solving cheating isn't a priority for them, and if we care about that we shouldn't buy their game. It especially doesn't mean you should accept a rootkit to buy their game
    - Arch-TK 2y ago
      I agree, but I am not the kind of person you need to persuade not to buy spyware. I am just explaining the kinds of reasoning I've heard first hand coming from "gamers".
      - advael 2y ago
        I view self-identified "gamers" as a cult at this point. If my goal were to persuade them of anything, I'd probably fail. I'm not in marketing for a reason. But the fact that a bunch of fools believe in something doesn't make it true
- DEADMINCE 2y ago
  > Basically, don't believe any of this "We have to own your computer for your own good" nonsense. That's a scam. Every time. Not really. It's all they can do to stop PC gamers cheating without having control of the hardware.
scheeseman486 2y ago
> If linux gaming requires you use a blessed immutable OS where everything you run is either unprivileged or signed by a central entity, it's not much better than just using Windows at that point. Yes it is, given that unlike Windows you still ultimately know what code being executed on your computer and have some degree of control over it. Should add, this isn't necessarily something I want, lucky for me I don't really play the games that require anti-cheat. But it is something that Valve could conceivably do.
- Arch-TK 2y ago
  You have as much control over it as you do over windows in this state: i.e. uninstalling the OS. Just because you know what the base is doing, doesn't mean you get to see what the proprietary kernel level drivers, loaded at runtime, are doing. Really it's effectively as good as having windows and running an open source web browser, or an open source kernel driver. It doesn't change the fact that your computer is being fundamentally controlled by components you can't change, some/many of which are also proprietary.
  - scheeseman486 2y ago
    This isn't a problem unique to what we're talking about, virtually every desktop PC on the planet has proprietary blobs running regardless of the OS used. I agree it sucks, but they're also not strictly necessary to run the containerized setup I proposed (any more than they're necessary to run the computer itself). It's possible to have attestation without anything proprietary/closed source.
    - Arch-TK 2y ago
      While it's possible to have attestation without anything proprietary/closed source. If you look at widevine you will quickly notice how that's very unlikely to happen. While it's possible to do this in containers, if you look at how DRM is built into the HDMI protocol, you will notice that it's unlikely you'll ever get to control a base while running the proprietary stuff in a container. Your idea is somewhat possible, but it's never going to happen in reality. I can already run windows in a VM for the exact same result.

advael 2y ago

Jokes I think there are three fundamental categories of cheating threat models that actually matter: State poisoning, Information leakage, and Input automation State poisoning means your game was poorly written, period. Either that's a vuln within the code itself or badly implemented netcode. A 2-player game can have total asynchronous client separation and still be peer-to-peer. A more-than-two-player game is almost always run on a server that serves as the single source of truth. In either case, a game that doesn't make the fundamental guarantee that the inputs available to a player and maybe some initial random seeds are the sole determinant of the gamestate have no hope, and rootkitting your computer because they wrote their game's statemachine or interfaces like shit is not the correct solution. If your answer to this is that big game studios shouldn't have to learn how to write more solid code, this means that the sanctity of their game isn't that important to them, not that they should get to root your computer Information leakage may be somewhat harder. Often you want the simulation to be running client-side, so a naive model of netplay would have the full state available to all clients from a technical perspective... but this doesn't have to be true. In most cases, you can do partial state with rollbacks to make it much harder to cheat from a technical perspective, even making no guarantees about the clients themselves. I think even when this is hard, the correct path here isn't rootkits, it's approaches that start to approximate zero-knowledge proofs. This also means there's a rich literature of zero-knowledge proofs to draw on Input automation, to be honest, is basically hopeless to prevent upfront regardless of what you do. If you can plug external hardware into your device at all, you can rig up something that automates your inputs. This can be hard to even verify in person, let alone through even a rootkit. I don't personally think it's worth worrying about that much, but if you care about macros and the like, it's really difficult to prevent. However, if there's money on the line or something, there are good analytic forensic techniques to detect this kind of cheating after the fact. Maybe this is where "AI" could actually help, as some kind of sequence-based anomaly detection that can run in real time might be able to detect unusual input clusters, but I worry that the false positive rate is going to be super high. Honestly seems like a lost cause. But crucially, not a lost cause that you get around via compromising the OS at a kernel level Anti-cheat that "needs to own your kernel" is more user-hostile corporate bullshit. Most games work fine on linux, but frankly no game is worth a rootkit, and no game needs one. The fact that some companies demand it should be viewed as those companies trying to scam you. That's not how the security of anything on the internet works. It's only how security of a bunch of mobile stuff works because Microsoft has trained generations of otherwise smart people to believe their total lies about security, and Google and Apple have taken advantage of this to secure a massive amount of control and surveillance over everyone who owns a smartphone (Which is increasingly required because they've also convinced people that fake 2FA that's just your phone as a single source of identity that can in fact often effectively be 1FA because it can override other authentication methods in most cases is somehow secure. The fact that everyone has a device with a bunch of proprietary backdoors that they don't have root on and that serves as a single lynchpin through which their life can be ruined is the most fundamental destruction of personal computer and identity security that's ever been realized - to say nothing of privacy, and that's a huge accomplishment given all that Microsoft, Facebook, and Amazon have done and still do to compete for the title) Giving a corp a backdoor to your computer doesn't secure anything except that corp's ability to fuck with you. Don't believe Microsoft, Apple, Epic Games, The NSA, or anyone else who tells you that the best way to secure something is to give them a backdoor. Fuck all those people. They have not only gotten their slimy tendrils in a ton of people's stuff through these lies, but have propagated bad information about how to do security to a ton of organizations. If someone who works at one of these scummy companies or agencies responds to this with some condescending corp-speak at me, I've got a bunch of work to do so I'll probably not get to you immediately, but I pre-emptively say that making this argument at all fundamentally undermines your credibility, and also I hate you on a personal level. You've been a spook too long and it's rotted your brain, hypothetical internet stranger who might not even exist, quit your job and fix your heart Basically, don't believe any of this "We have to own your computer for your own good" nonsense. That's a scam. Every time. Also, proprietary software should be assumed inherently insecure by default, not the other way around. A better world is possible

Arch-TK 2y ago
The reason companies seem to bother at this point is that, by implementing increasingly intrusive anti-cheat, they force cheaters to be increasingly subtle. With sufficiently intrusive anti-cheat you end up with gamers believing that the game they're playing has no cheaters. This currently happens in at least some of the games which utilise kernel level anti cheat, as demonstrated by numerous videos on the topic which also shed light on massive communities of cheaters who just end up buying or making their own hardware based cheats.
- advael 2y ago
  Gamers will believe anything a company they like tells them. The fact that these companies can attract arguably the most insufferably delusional audience of frothing bootlickers to ever walk the earth to defend them doesn't justify their decisions on a technical or an ethical level If the explanation for why a company needs a rootkit is "they don't want to spend effort on a better solution" that means that solving cheating isn't a priority for them, and if we care about that we shouldn't buy their game. It especially doesn't mean you should accept a rootkit to buy their game
  - Arch-TK 2y ago
    I agree, but I am not the kind of person you need to persuade not to buy spyware. I am just explaining the kinds of reasoning I've heard first hand coming from "gamers".
    - advael 2y ago
      I view self-identified "gamers" as a cult at this point. If my goal were to persuade them of anything, I'd probably fail. I'm not in marketing for a reason. But the fact that a bunch of fools believe in something doesn't make it true
DEADMINCE 2y ago
> Basically, don't believe any of this "We have to own your computer for your own good" nonsense. That's a scam. Every time. Not really. It's all they can do to stop PC gamers cheating without having control of the hardware.

scheeseman486 2y ago

> If linux gaming requires you use a blessed immutable OS where everything you run is either unprivileged or signed by a central entity, it's not much better than just using Windows at that point. Yes it is, given that unlike Windows you still ultimately know what code being executed on your computer and have some degree of control over it. Should add, this isn't necessarily something I want, lucky for me I don't really play the games that require anti-cheat. But it is something that Valve could conceivably do.

Arch-TK 2y ago
You have as much control over it as you do over windows in this state: i.e. uninstalling the OS. Just because you know what the base is doing, doesn't mean you get to see what the proprietary kernel level drivers, loaded at runtime, are doing. Really it's effectively as good as having windows and running an open source web browser, or an open source kernel driver. It doesn't change the fact that your computer is being fundamentally controlled by components you can't change, some/many of which are also proprietary.
- scheeseman486 2y ago
  This isn't a problem unique to what we're talking about, virtually every desktop PC on the planet has proprietary blobs running regardless of the OS used. I agree it sucks, but they're also not strictly necessary to run the containerized setup I proposed (any more than they're necessary to run the computer itself). It's possible to have attestation without anything proprietary/closed source.
  - Arch-TK 2y ago
    While it's possible to have attestation without anything proprietary/closed source. If you look at widevine you will quickly notice how that's very unlikely to happen. While it's possible to do this in containers, if you look at how DRM is built into the HDMI protocol, you will notice that it's unlikely you'll ever get to control a base while running the proprietary stuff in a container. Your idea is somewhat possible, but it's never going to happen in reality. I can already run windows in a VM for the exact same result.