Machine learning as anti-cheat: Valve's response to never ending cheating issues?

[laminutederire]

Hi, I haven't seen it being covered in this section of the forum, and it qualifies as news to me:

 

As PCgamers's article states, Valve is toying with machine learning as a way to overcome cheating in Counter Strike.

Their strong feelings about this approach being the best one stem from the unavoidable realization that fighting cheats the old fashioned way is equivalent to a never ending arms race, fixing anti cheat to detect cheats which will be made relevant again when programmers of said cheats finally find what have changed.

 

Cheating in online multi-player has always been a concern, especially in competitive games such as counter-strike. It is always welcomed to find new smart ways to fight against it, and it is a feature many competitive players ask for continuously.

 

Will this however be enough?

One can argue that rising the game price to throw off cheaters from the perspective to rebuy the game would be enough. Other can argue that machine learning algorithms could be fooled as well by someone knowing how they work.

 

From what I learned about the subject, the objection that it can be fooled is valid. It however remains that it is a complex enough algorithm, which may only be messed with by really talented programmers with strong mathematics background, which are less common than only really talented programmers. In addition, if the learning is unsupervised (as with neural networks or some other methods), it becomes harder to understand how it is choosing to flag players and therefore to fool it.

 

What place can it have in the future though?

Indeed, it is noteworthy to point out that those algorithms alone could ban legit players all the more easily. Should it be the only way to ban players? 

As of now it seems it will only be a support to the peer reviewed system.

 

It is a pioneer approach, which has its drawback for Valve at least. Their spokesperson pointed out that it will need huge computing power to account for enough matches to train it upon, and then on the ability to watch every game to flag cheaters. Enough to need additional datacenters, since the servers cannot cater to this need. Should they try to do the calculation on the client side as well to relieve them from part of those calculations?

 

I don't know if this fits the format, but I tried to speak of everything they said in the article and to add from my perspective to ask you guys what you think about this!

[ChineseChef]

While I think AI learning to prevent cheats is likely the future.  It will have to be careful to not flag people simply because they are "too good".  I forget which game it was, but they had anti-cheat that was based around player score/reaction times/accuracy vs the mean.  They found that it worked great until player got good enough that they started to get themselves flagged as cheating because they were simply "too good".  The devs ended up having to mod the anti-cheat, but it still created problems for a short time.  So that would be my only concern, false positives.

[tlink]

Machine learning is hard to get right, it almost always has unwanted behaviour. I think that its the lesser evil of the other options. Its a little bit like drm, just that this is supposed to actually support the consumer. It probably fucks over someone. They should make it easy to reach a person in case something bad happens with it.

[ZestyJalapeno]

Players stated very strongly there was absolutely no evidence that cheating affects the game. Just another scheme by losers Valve! Sad!

[Prysin]

44 minutes ago, ChineseChef said:

While I think AI learning to prevent cheats is likely the future.  It will have to be careful to not flag people simply because they are "too good".  I forget which game it was, but they had anti-cheat that was based around player score/reaction times/accuracy vs the mean.  They found that it worked great until player got good enough that they started to get themselves flagged as cheating because they were simply "too good".  The devs ended up having to mod the anti-cheat, but it still created problems for a short time.  So that would be my only concern, false positives.

All fun and game, until 4chan learns about this AI.

[BlueChinchillaEatingDorito]

1 hour ago, ChineseChef said:

While I think AI learning to prevent cheats is likely the future.  It will have to be careful to not flag people simply because they are "too good".  I forget which game it was, but they had anti-cheat that was based around player score/reaction times/accuracy vs the mean.  They found that it worked great until player got good enough that they started to get themselves flagged as cheating because they were simply "too good".  The devs ended up having to mod the anti-cheat, but it still created problems for a short time.  So that would be my only concern, false positives.

I remember back when I used to play Combat Arms (yea, that was thing), this was this guy originally called "aTastyCookie" and he was banned multiple times for being too good.

[2FA]

2 minutes ago, BlueChinchillaEatingDorito said:

I remember back when I used to play Combat Arms (yea, that was thing), this was this guy originally called "aTastyCookie" and he was banned multiple times for being too good.

Ravic gets banned from BF4 servers all the time because admins are idiots that think he is cheating. https://www.youtube.com/user/blackmesatech

[BlueChinchillaEatingDorito]

1 minute ago, DeadEyePsycho said:

Ravic gets banned from BF4 servers all the time because admins are idiots that think he is cheating. https://www.youtube.com/user/blackmesatech

 

Yea, I'm not as good and I've been reported multiple times in this game. Still not banned yet though.

[Doobeedoo]

Hmm well now that's something. Though just put them with other cheaters and let them all play like that.

[laminutederire]

1 hour ago, ChineseChef said:

While I think AI learning to prevent cheats is likely the future.  It will have to be careful to not flag people simply because they are "too good".  I forget which game it was, but they had anti-cheat that was based around player score/reaction times/accuracy vs the mean.  They found that it worked great until player got good enough that they started to get themselves flagged as cheating because they were simply "too good".  The devs ended up having to mod the anti-cheat, but it still created problems for a short time.  So that would be my only concern, false positives.

False positives are okay if they continue to peer review demos of matches before actually banning people, maybe?

[ChineseChef]

2 hours ago, laminutederire said:

False positives are okay if they continue to peer review demos of matches before actually banning people, maybe?

I would say false positives are ok if the admins are proactive in responding to disputes.  And err on the side of the player.

[laminutederire]

39 minutes ago, ChineseChef said:

I would say false positives are ok if the admins are proactive in responding to disputes.  And err on the side of the player.

At first it'll begin with spinbots so there shouldn't be that much false positives I guess

[cesrai]

I like how the article is based on a reddit comment exsdee .

[laminutederire]

51 minutes ago, cesrai said:

I like how the article is based on a reddit comment exsdee .

It seems legit enough to be put on counter strike steam page

[cesrai]

2 minutes ago, laminutederire said:

It seems legit enough to be put on counter strike steam page

I never said it wasn't legit 

[AnonymousGuy]

Valve is much behind the FairFight curve.  That shit makes cheating in BF pretty much impossible..

[laminutederire]

18 minutes ago, AnonymousGuy said:

Valve is much behind the FairFight curve.  That shit makes cheating in BF pretty much impossible..

How so? I don't play battlefield as often so I am genuinely interested!

 

22 minutes ago, cesrai said:

I never said it wasn't legit 

Sure, it's just that with politics and current launch hypes, I feel obligated to point out its legit potential compared to all the rumours our there

 

[W-L]

@laminutederire Please don't forget to include a quote from the article that related to the discussion.  Currently the thread has been moved to the PC Gaming Section until the revisions have been made, once done the thread may be moved back to the TNR section. 

 

https://linustechtips.com/main/topic/11724-posting-guidelines-read-before-posting/

Quote

• Your thread should also include quotes from the cited source(s). While you shouldn't just copy the entire article, your quote should give the reader a summary of the article in a way that gives the key details, but also leaves room for them to read the full article on the linked website. Please use quote tags (the speech bubble at the top of the editor, under the  :)) to show that you have copied this content from another site.

 

[AnonymousGuy]

4 hours ago, laminutederire said:

How so? I don't play battlefield as often so I am genuinely interested!

 

Sure, it's just that with politics and current launch hypes, I feel obligated to point out its legit potential compared to all the rumours our there

 

Valve is literally describing what FairFight does.  It looks at various player metrics and "impossible" actions to ban.  Examples include headshot % too high, shooting through walls constantly, tracking players through walls with your crosshair, snapping too accurately to someone, etc etc.  There's never been a legitimately god player banned from it, that I've heard of.

 

You basically end up spending all you cheating time pretending to play legit, which point you might as well just play legit.  Yeah you can turn your aimbot snap speed down, make it so it doesn't go for headshots, make it so it only aims at stuff within 15 degrees of the crosshair, etc etc but then you barely have any advantage.

 

It's all server side too so there's no getting around it.  Its biggest problem is its sometimes too slow to ban people rage-hacking (200-0).