You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
One of our admins is on shitty cellular internet (lives out in the countryside, all he can get unfortunately) and when his latency significantly jumps Oryx starts returning false positives for the perfect strafe detection. I know for a fact he's not cheating (as in I've literally stood in his house and watched over his shoulder as he plays, he's a friend and co-owner of the server) so I suspect it's something to do with the connection variability causing Oryx to return strafe offsets as zero during these spikes. I think just increasing the sample size might mitigate this at least partially, but if ping is accessible via Sourcemod one possible solution could be to label or drop detections that occur during significant deviations from a rolling average of ping. I think (maybe) that'd be more robust than an average taken only once as a single derived sample could be exploited by joining, letting it set and then deliberately inducing network congestion with P2P software or some other lag-causing activity.
I'm going to try doubling the sample size to see if that has any effect on things for now.
E: Doubled the sample and detection thresholds for perfect strafes, he still kept showing up and I had a lightbulb moment. Checked status from rcon and his rate was only 20840, which is iirc the lowest possible on 128tick or maybe at all. Had him bump it to 64000 and checked strafe_stats, still some zeroes but below the low detection threshold. Going to install a rate enforcing plugin and set that as the lower bound, should cut the number of bad detections from people with old/weird configurations.
E2: Ok, I'm dumb as heck. sv_minrate was 16000, no wonder things were being weird. Whoops. I guess maybe note in the documentation that it has the potential to generate false positives if left at default.
The text was updated successfully, but these errors were encountered:
One of our admins is on shitty cellular internet (lives out in the countryside, all he can get unfortunately) and when his latency significantly jumps Oryx starts returning false positives for the perfect strafe detection. I know for a fact he's not cheating (as in I've literally stood in his house and watched over his shoulder as he plays, he's a friend and co-owner of the server) so I suspect it's something to do with the connection variability causing Oryx to return strafe offsets as zero during these spikes. I think just increasing the sample size might mitigate this at least partially, but if ping is accessible via Sourcemod one possible solution could be to label or drop detections that occur during significant deviations from a rolling average of ping. I think (maybe) that'd be more robust than an average taken only once as a single derived sample could be exploited by joining, letting it set and then deliberately inducing network congestion with P2P software or some other lag-causing activity.
I'm going to try doubling the sample size to see if that has any effect on things for now.
E: Doubled the sample and detection thresholds for perfect strafes, he still kept showing up and I had a lightbulb moment. Checked status from rcon and his rate was only 20840, which is iirc the lowest possible on 128tick or maybe at all. Had him bump it to 64000 and checked strafe_stats, still some zeroes but below the low detection threshold. Going to install a rate enforcing plugin and set that as the lower bound, should cut the number of bad detections from people with old/weird configurations.
E2: Ok, I'm dumb as heck. sv_minrate was 16000, no wonder things were being weird. Whoops. I guess maybe note in the documentation that it has the potential to generate false positives if left at default.
The text was updated successfully, but these errors were encountered: