Spec Cheat Detector via inverse mouse movement relationship

Spec Cheat Detector via inverse mouse movement relationship

Requests for features in the spring code.

Moderator: Moderators

Post Reply
NeOmega
Posts: 62
Joined: 15 Dec 2010, 11:06

Spec Cheat Detector via inverse mouse movement relationship

Post by NeOmega » 14 Jan 2019, 04:44

So, the more I think of this tool, the more I think it makes sense to be implemented at the engine multiplayer level.

Basically, have a program that detects (or reads, if in a save file) movement of the player, and pairs it against a spectator.
If the spectator only moves their mouse when the player does not, and visa-versa, then there is a very high likely hood of twin-box spec cheating.

The first check would just scan for mouse movement time. If (gameTime - mouseMovementTime = spec mouseMovementtime), twin-box spec cheat likely.

A deeper map could then be done, to check if the mouse movements were occurring at inverse times of each other.
I imagine a pair of lists, containing a series of time stamp starts and stops, finding the intersection time intervals, then add the intersectional time spans together. If the amount of time where both mouses were moving is close to 0% of gametime, definite pattern of twin-box spec cheating.
A black/white/gray line readout would be nice too... but I don't know if this data can easily be displayed graphically.

Recheck other save games where the suspect player and spec were present. If the pattern is re-producable, you have weeded out a spec cheater.

Even after 1 game, the likely hood that two people were moving and not moving (inversely) simultaneously, is near impossible.
After 5 games, it is impossible. The player and spec are spec cheating.

The interesting thing about this scanner, is it could check retroactively. >D

It of course could not check for friend-pair spec cheating, since that does actually require 2 people.
1 x

User avatar
Jools
XTA Developer
Posts: 2804
Joined: 23 Feb 2009, 16:29

Re: Spec Cheat Detector via inverse mouse movement relationship

Post by Jools » 18 Jan 2019, 16:54

NeOmega wrote:
14 Jan 2019, 04:44
If the spectator only moves their mouse when the player does not, and visa-versa, then there is a very high likely hood of twin-box spec cheating.
---
Even after 1 game, the likely hood that two people were moving and not moving (inversely) simultaneously, is near impossible.
After 5 games, it is impossible. The player and spec are spec cheating.
How likely? Would you mind developing your theory a bit...

What is the likelyhood of type I/II errors? That means false positives and false negatives and what is the strategy to minimise both these?
0 x

raaar
Metal Factions Developer
Posts: 815
Joined: 20 Feb 2010, 12:17

Re: Spec Cheat Detector via inverse mouse movement relationship

Post by raaar » 19 Jan 2019, 16:54

this seems exploitable:
- false negative : a cheater with a nearby laptop can move both mouses simultaneously every minute or so to fool the algorithm
- false positive : spectator was afk or just paying attention to another window and didn't move the mouse
0 x

NeOmega
Posts: 62
Joined: 15 Dec 2010, 11:06

Re: Spec Cheat Detector via inverse mouse movement relationship

Post by NeOmega » 20 Jan 2019, 04:22

Jools wrote:
18 Jan 2019, 16:54
NeOmega wrote:
14 Jan 2019, 04:44
If the spectator only moves their mouse when the player does not, and visa-versa, then there is a very high likely hood of twin-box spec cheating.
---
Even after 1 game, the likely hood that two people were moving and not moving (inversely) simultaneously, is near impossible.
After 5 games, it is impossible. The player and spec are spec cheating.
How likely? Would you mind developing your theory a bit...

What is the likelyhood of type I/II errors? That means false positives and false negatives and what is the strategy to minimise both these?
the first test would be a sample test, every 20 frames, that's about 3 test samples a second.
It cross checks each player with each spectator.
If there is a possible inverse match between a player and spectator, it could do a deeper check.

the false positives would be specs that start afk, vs players that play the whole game.
But, with a pair of human readable "timelines" that show black when the mouse is not moving, white when it is, and then gray during intersections of movement, one could quickly read the situation. It's not like it would take months of training. A couple minutes of reading the instructions, and what patterns constitute a dual-box spec cheater.

the chances of false positives would literally be zero.

False negatives could still happen, because dual box spec cheating is not the only way to spec cheat.


- false negative : a cheater with a nearby laptop can move both mouses simultaneously every minute or so to fool the algorithm
This is kind of like bayesian filtering for spam; yes, spam gets through still, but the words say "p3n1s enlarg3ment" diluting the spammers message. Yes, a dual-box spec cheater could periodically move both mouses at once..., but it would start to hamper their gameplay. They would have to be awfully persistent to get it done right. And they would have to do it every game. Is the cheater having fun yet?
0 x

User avatar
PicassoCT
Journeywar Developer & Mapper
Posts: 10229
Joined: 24 Jan 2006, 21:12

Re: Spec Cheat Detector via inverse mouse movement relationship

Post by PicassoCT » 21 Jan 2019, 07:38

Penis enlargments

Reported and deported at once. Spring is big enough, the community does not need further aid.
0 x

Post Reply

Return to “Feature Requests”