4 ms·

> CSGO's overwatch allows the community to self-moderate by replaying a player's gameplay. They literally record the player's input and rebuild what the player

> CSGO's overwatch allows the community to self-moderate by replaying a player's gameplay. They literally record the player's input and rebuild what the player could see in the reviewer's client. That's literally wrong, the gameplay you see in OW is not based on recorded inputs of the player's client. You are watching a demo recorded by the server, which does not match the PoV of the player precisely. The demos are also 32 tick, which makes it hard to tell if the aiming of the player is legit [1], unless they are very blatant about it. Due to interpolation and lag compensation the PoV that the player saw is not what you see in the server's demo. If you record the player's PoV using screen capture and compare it to the demo recorded by the server, there are subtle and sometimes relevant differences. tl;dr OW is only there to catch obvious cheaters. [1] Because the server runs at 64 tick, but the demo is only 32, the crosshair snaps every time a shot is taken. The aim of everyone looks way snappier and suspicious in a 32 tick demo compared to their PoV.

strbean 7y ago
Change "player's input" to "player's input to the server" and it checks out :)
oliwarner 7y ago
That's a rather aggressive rebuttal. Its conclusion is also incorrect. It was a slight simplification. You send your input to the server. Yes, filtered. Yes, not 1000Hz. But essentially what the client was doing. You could increase the resolution here but again, who's paying for it? I do agree that the rebuilt context is what the server saw, not what the user saw after network and screen latency, but we're talking about behavioural analysis. And that's where I think you're flat out wrong. One kill could be a suspiciously-tracked headshot. But systems like this can compile hundreds of kills over dozens of matches from a single user. Its volume builds an accurate picture about whether a user is a cheater or not. And it's a country mile better than games that have no repercussions for cheaters where the cheats haven't been detected by Gaming's answer to Norton AntiVirus 2003. Because zero-days still get by those. (Edit: I'll also use my edit window to add that many servers run a lot faster than 64 - just as you can record at any speed you like - by Overwatch may be locked lower for disk space reasons)
- blattimwind 7y ago
  Aggressive? > And that's where I think you're flat out wrong. One kill could be a suspiciously-tracked headshot. But systems like this can compile hundreds of kills over dozens of matches from a single user. Its volume builds an accurate picture about whether a user is a cheater or not. I don't know what OW does behind the scenes. But you and I as "OW investigators" only get to see ~eight rounds of gameplay with no rewind button and make a "beyond reasonable doubt? Yes/No" decision. I stand by what I said - OW is to catch obvious cheaters, not people who cheat subtly. > (Edit: I'll also use my edit window to add that many servers run a lot faster than 64 - just as you can record at any speed you like - by Overwatch may be locked lower for disk space reasons) Overwatch only applies to Valve MM (which is 64 tick); community servers and non-MM game modes are not subject to OW.

strbean 7y ago

Change "player's input" to "player's input to the server" and it checks out :)

oliwarner 7y ago

That's a rather aggressive rebuttal. Its conclusion is also incorrect. It was a slight simplification. You send your input to the server. Yes, filtered. Yes, not 1000Hz. But essentially what the client was doing. You could increase the resolution here but again, who's paying for it? I do agree that the rebuilt context is what the server saw, not what the user saw after network and screen latency, but we're talking about behavioural analysis. And that's where I think you're flat out wrong. One kill could be a suspiciously-tracked headshot. But systems like this can compile hundreds of kills over dozens of matches from a single user. Its volume builds an accurate picture about whether a user is a cheater or not. And it's a country mile better than games that have no repercussions for cheaters where the cheats haven't been detected by Gaming's answer to Norton AntiVirus 2003. Because zero-days still get by those. (Edit: I'll also use my edit window to add that many servers run a lot faster than 64 - just as you can record at any speed you like - by Overwatch may be locked lower for disk space reasons)

blattimwind 7y ago
Aggressive? > And that's where I think you're flat out wrong. One kill could be a suspiciously-tracked headshot. But systems like this can compile hundreds of kills over dozens of matches from a single user. Its volume builds an accurate picture about whether a user is a cheater or not. I don't know what OW does behind the scenes. But you and I as "OW investigators" only get to see ~eight rounds of gameplay with no rewind button and make a "beyond reasonable doubt? Yes/No" decision. I stand by what I said - OW is to catch obvious cheaters, not people who cheat subtly. > (Edit: I'll also use my edit window to add that many servers run a lot faster than 64 - just as you can record at any speed you like - by Overwatch may be locked lower for disk space reasons) Overwatch only applies to Valve MM (which is 64 tick); community servers and non-MM game modes are not subject to OW.