Quoting: RCLE.g. they flag stuff like an unfamiliar dynamic library being mapped to the game's address space or certain parts of the process memory not being the same as they expect - that's usually the extent of their "spying". They have no interest to "spy" on things that a user would be concerned about (like e.g. files or what not)
<...>
But I totally understand that this can all sound very abstract. And especially if there's no trust towards the developer, the very fact of allowing their software to hook up to the kernel can be a big no-no.

May be I should rephrase the above a bit. It's designed to enforce something as an expression of that lack trust. You can call it "spy" in a sense that this policing tool monitors something in your private space (up to the kernel) becasue it doesn't trust you. That fact alone (regardless of what exactly it's monitoring) I think is a good enough reason not to trust it in return and assume that it could do more than it declares it wants to do.

Which is basically what you expressed in that last sentence as well, so we don't really disagree.

View this comment - View article - View full comments

Quoting: RCLWell, this is a grim way to look at things, although I don't blame you for taking that POV.

Not so much grim as simply more security conscious approach. You wouldn't want to trust those who don't trust you. It's not like you know the other side is a crook, but if they treat you as an a priory potential crook, you should do the same to them, it's only fair.

Quoting: RCLHowever, IMHO the situation is more analogous with, say, airport security. Both sides assume the good will of the other, but it needs to be enforced. However, the enforcement isn't arbitrary - there are accepted limits what can be checked and what cannot, and everything is done in a respectful manner.

I think the core difference is that such kind of security is external to your private space. I.e. imagine you get agents in your home all the time, because anyone can be a potential threat. Digital space is more abstract, but there is still the sense of private space there too. Like your computer, your OS it runs, programs that run on your OS and etc. It's not that policing in general is a problem, it's policing your private space that is and that's the core of this issue above.

Quoting: RCLOf course among the Linux users, especially typical users, there's very little, if any, trust to closed source in general, and to closed source kernel components in particular. However, Linux gamers are crossing the boundaries - a lot of us do not mind running many closed source components on the system (the games themselves), including the drivers. Again, in that case anti-cheat doesn't stand out that much.

Lack of trust for blobs is not unreasonable, but I'd argue blobs in general can be more neutral due to their goals. I.e. some normal game doesn't have a goal of spying on you, even though it could since it's a blob and you don't really know fully what it could do. But anti-cheat? It's explicitly designed to spy on you (though formally just for the narrow purposes related to the game). I'd say in regards to trust it's like in a whole worse category than other blobs above due to that.

View this comment - View article - View full comments

Quoting: RCLWhile I agree with you, I want to stress that anti-cheat isn't a malicious software that has no boundaries. Anti-cheats are governed by EULAs which sets the limits of what they are allowed (by the user accepting the EULA) to do (e.g. https://www.easy.ac/en-us/support/cardlife/account/eula/ [External Link]). In principle, the situation with trusting the anti-cheat does not really differ from trusting the closed source kernel you're running on other platforms, or closed source binary drivers (or firmware) you might be running on Linux. In all these cases the trust between the user and the vendor is enforced via legal agreements that both sides accept as a reasonable compromise between the system's functionality and their control over the system. Different platforms (PC vs console) offer a different degree of that control, but in all cases they are based on the mutual agreement.

The problem with this (and same applies to DRM) is the very issue of trust that you bring. Users also need to accept the rules of the service, yet they still cheat, right? I.e. the company doesn't trust the user despite the EULA. And not only doesn't trust, but deploys spyware-like capabilities on user's system, meaning treating all users as suspects by default.

Now, why should the user trust the company in such situation? Trust goes both ways. That's the main pitfall of overreaching preemptive policing in general - those who use it don't have trust, but then why they themselves be trusted in return? Quite on the contrary, they also should never be trusted.

There was a neat illustration of this idea in this video:

https://www.youtube.com/watch?v=XgFbqSYdNK4 [External Link]

View this comment - View article - View full comments

Quoting: RCLbut I am not aware of any breakthroughs there.

I agree that making such solution is harder than it sounds, I don't doubt that. But I just don't see lack of current solution as a reason to erode users' privacy. Same as let's say lack of good self-driving AI isn't a reason to use some other poor solution that will endanger people's lives becasue self-driving itself is considered "cool" and someone says it has to be enjoyable already today.

This is actually a good comparison, because most people take such kind of security seriously and understand the implications of using inappropriate solutions. When it comes to information security - it's more abstract. And people more easily ignore issues with it like the above anti-cheat rootkits.

View this comment - View article - View full comments

Quoting: RCLThis is indeed an open area of research and work. However, as far as I am aware - and I'm not an anti-cheat specialist - there isn't enough robustness as of now yet, to avoid false positives or missing the cheaters, and more importantly, the current solutions have a fairly long lead time, which cheaters can beat by recycling the accounts faster (in a free-to-play game at least). So while a promising area, this is far from a solution that can be enabled "right here, right now" to combat today's cheaters, unfortunately.

I don't see it as an excuse to resort to unacceptable solutions. Rather it should be an incentive to invest more into such AI.

Also, the way this works is a slippery slope. Once these companies manage to portray this as "acceptable" and people start ignoring the issue, it becomes very hard to get rid of it even if there are better solutions. Because they don't want to give up power and control.

Same thing with other areas with similar issues.

I recommend reading the above Watchbird, it's very on point.

View this comment - View article - View full comments

Quoting: RCLYou seem to maintain that there's the "proper" solution that avoids both of those downsides and yet doesn't limit the user's freedom.

There is. Develop an AI that will analyze user behavior while running server side and will detect patterns that will be deemed as "illegal" by the game. Same as police would monitor people's behavior in the street, but wouldn't sit in people's homes because they "need to suspect everyone to maintain the level of trust".

How you define those patterns is totally up to developers. The more sophisticated ways they'll come up with and the more they train it, the better the AI will be detecting what they define as "cheating". I don't see any unsolvable problem there.

It's not a perfect solution, but it can be a sufficient solution for the level of trust you need. Same as policing in real life is. Point is - there is no perfect solution for it. But there can be good enough ones.

Basically, I don't see any huge philosophical problem here, it's the same old privacy vs policing idea that's been handled for a long time in some acceptable fashions that don't cross important boundaries.

View this comment - View article - View full comments

Quoting: RCLThis is a rather bold statement to make, without having the data or likely even an idea about the health of the game.

It is bold and based on a simple idea - messing up user's privacy and security becasue developers don't want to spend effort on proper solutions is unacceptable.

Quoting: RCLwhich poses a fundamental and philosophical issue - how to prevent bad behavior on a platform that trusts their users completely? People who are smarter than me are working on this and don't have a good (or economical at least) solution so far.

Simple - you don't (do it on people's systems). Same as you don't put police cameras in people's homes (wouldn't it help preventing more crime?). Same reason there is the Fourth Amendment and such.

No matter how "serious" this problem is, current idea of preemptive user side anti-cheat is not a solution. Not any more than let's say the Watchbird is a solution to crime.

For those who want to better get the reference, see:

* https://www.gutenberg.org/ebooks/29579 [External Link]
* https://www.gutenberg.org/files/29579/29579-h/29579-h.htm [External Link]

View this comment - View article - View full comments

Quoting: RCLBut it is more reactive than preventive.

No one said you need to have a preventive anti-cheat. Or to put it differently, the elephant in the room is that you don't need it to make the game good enough. But good reactive anti-cheat is a hard and expensive problem to solve. Not impossible.

View this comment - View article - View full comments

Quoting: Cyba.CowboyCouldn't they just implement anti-cheat software on the server instead?

Correct me if I am wrong, but I would imagine that this would be easier to maintain and it would be far more difficult for potential cheaters to bypass...

They could. But it's equivalent to investing into sophisticated enough AI. They don't want to spend on it. It's cheaper to push spyware junk on the user.

View this comment - View article - View full comments

Pretty cool trailer. I hope they'll make GOG release too.

This also reminds me to check how Everspace 2 progress is going.

View this comment - View article - View full comments