Posts: 799
Threads: 52
Joined: Jan 2011
30 Dec 11, 05:25AM
(This post was last modified: 30 Dec 11, 03:42PM by Roflcopter.)
For quite some time I've claimed that unhits affected some players more than others and further that I am affected more than most others. An unhit, as tested below, is a bullet that was discarded due to overzealous cheat detection code.
I made a private server that logs all hits and unhits. With these logs I made a Python script that totals the unhits and number of bullets that hit for each player. Players only played AR during the games. The results below show the proportion of bullets each player unfairly loses (rounded to 2sf):
| Roflcopter | 23% | | MeLonE | 21% | | Vault | 20% | | Bigbaddy | 14% | | yopa | 13% | | Jason | 12% | | Benson | 11% | | Vermi | 5.9% | | Dam | 2.6% |
First of all, you can see that some players are hardly affected while others lose more than 1 in 5 bullets. Some players played in multiple games, in those cases the highest result is shown above (but in all those cases the difference between results for each player on different games were very small).
I'd also like to point out that this rejects the claim that the hit-fix is more unfair to snipers since some players are already getting virtually no hit-loss (eg Dam with only 2.6%).
It also clearly shows that many are affected significant amounts more than others (compare Dam to MeLonE: at least 17% more hits).
Posts: 441
Threads: 13
Joined: Jul 2010
Being part of this test, all I can say: Ha. Ha. I told you f***ing so.
Balance isn't an issue, so get on and fix this properly.
Posts: 2,841
Threads: 44
Joined: Jun 2010
30 Dec 11, 05:30AM
(This post was last modified: 30 Dec 11, 05:31AM by #M|A#Wolf.)
The devs would probably want the actual code where this information was gotten instead of a tables with names and percentages, but thank you for pushing this issue further, it's something that indeed needs to be fixed.
Posts: 799
Threads: 52
Joined: Jan 2011
30 Dec 11, 05:36AM
(This post was last modified: 30 Dec 11, 06:20AM by Roflcopter.)
(30 Dec 11, 05:30AM)|BC|Wolf Wrote: The devs would probably want the actual code where this information was gotten instead of a tables with names and percentages, but thank you for pushing this issue further, it's something that indeed needs to be fixed.
I'd be happy to provide code if anyone is interested.
I think the take-away is there's no reason not to use fixed servers.
(30 Dec 11, 05:28AM)Jason Wrote: Being part of this test, all I can say: Ha. Ha. I told you f***ing so.
Balance isn't an issue, so get on and fix this properly.
Worth noting that SVN does contain a purported fix for this that feels like it does indeed fix my problem. Considering how severe and widespread this problem is I think the developers should aim to release it sooner rather than later.
Also worth noting is that this doesn't account for unhits caused by the anticheat. I've claimed before that the anticheat affected hit-reg significantly but this was only based on anecdotal evidence.
Posts: 239
Threads: 15
Joined: Aug 2010
I'm just a little curious, what exactly accounts for the large difference in values?
Posts: 1,981
Threads: 63
Joined: Jun 2010
(30 Dec 11, 05:28AM)Jason Wrote: Being part of this test, all I can say: Ha. Ha. I told you f***ing so.
Balance isn't an issue, so get on and fix this properly.
it doesn't prove that the fix will not advantage automatic guns.
Also it proves you were wrong when you blamed the anticheat.
you're so arrogant...
Posts: 799
Threads: 52
Joined: Jan 2011
(30 Dec 11, 11:03AM)Luc@s Wrote: it doesn't prove that the fix will not advantage automatic guns.
Also it proves you were wrong when you blamed the anticheat.
It does prove that some people are already advantaged versus non-automatic weapon players though.
I'll be the first to admit that I probably overestimated the role of the anticheat. I can't perform checks on it myself though, so maybe a dev could?
(30 Dec 11, 10:16AM)Habluka Wrote: I'm just a little curious, what exactly accounts for the large difference in values?
I'm not sure actually. It may be to do with FPS somewhat and it's certainly to do with connection too.
Posts: 2,841
Threads: 44
Joined: Jun 2010
30 Dec 11, 01:47PM
(This post was last modified: 30 Dec 11, 01:49PM by #M|A#Wolf.)
Lucas, you can't say that you won't even attempt fixing it just because then automatic weapons will be too OP... If that's really a problem, then reducing the automatic weapons' damage or whatever is needed to make it fair shouldn't bother many. I'd prefer to know that I hit all my shots but still wasn't able to get the kill, than know I couldn't get the kill because some hits didn't register.
Every bullet SHOULD register, and it is highly unwise to keep it that way just because then automatic weapons will be too overpowered. By the table itself, you can notice some players are less affected than others, therefore those players already have an instant advantage over everyone else.
Posts: 2,331
Threads: 45
Joined: Feb 2011
Good job <3. Hopefully the SVN hitreg fix does what it says it'll do.
![[Image: 12700949.jpg]](http://memegenerator.net/cache/instances/400x/12/12403/12700949.jpg)
Posts: 2,136
Threads: 50
Joined: Jun 2010
I guess the (weak) defence at the moment is that the disadvantage is apparently random (or only punishes the more accurate players), rather than specific to a weapon type. Roflcopter, can you get the average FPS on the table, rather than speculate? I can see it's not directly ping related, but it wouldn't hurt to put this data on there too, if it was all on one server.
Posts: 799
Threads: 52
Joined: Jan 2011
(30 Dec 11, 02:17PM)jamz Wrote: I guess the (weak) defence at the moment is that the disadvantage is apparently random (or only punishes the more accurate players), rather than specific to a weapon type. Roflcopter, can you get the average FPS on the table, rather than speculate? I can see it's not directly ping related, but it wouldn't hurt to put this data on there too, if it was all on one server.
As far as I know, FPS values aren't reported to the server so I would have to take everyone's estimates of their own FPS. I can start collecting ping and PJ though.
Posts: 881
Threads: 74
Joined: Mar 2011
Wouldn´t it need to be "Quantitative Values on Units"?
Posts: 441
Threads: 13
Joined: Jul 2010
(30 Dec 11, 11:03AM)Luc@s Wrote: it doesn't prove that the fix will not advantage automatic guns.
Also it proves you were wrong when you blamed the anticheat.
you're so arrogant...
Read the post again, you're wrong.
Posts: 799
Threads: 52
Joined: Jan 2011
(30 Dec 11, 02:31PM)paulmuaddibKA Wrote: Wouldn´t it need to be "Quantitative Values on Units"?
Thanks for the correction.
Posts: 256
Threads: 15
Joined: Jun 2010
if you can, correlate the connection type and if they are using wifi. In addition to that, you can ask what specific OS they are using. (I wonder if this has any influence)
Also, if you're having another session, can i join?
Posts: 799
Threads: 52
Joined: Jan 2011
30 Dec 11, 02:49PM
(This post was last modified: 30 Dec 11, 02:51PM by Roflcopter.)
Additional people:
| Shorty | 18% | | Undead | 16% | | Xenon | 5.3% | | Jorik | 3.4% |
(30 Dec 11, 02:45PM)SKB Wrote: if you can, correlate the connection type and if they are using wifi. In addition to that, you can ask what specific OS they are using. (I wonder if this has any influence)
Also, if you're having another session, can i join?
I'll aim to gather what I can. And you're quite welcome to test. I'll be idling on IRC.
Posts: 1,981
Threads: 63
Joined: Jun 2010
30 Dec 11, 02:57PM
(This post was last modified: 30 Dec 11, 02:58PM by Luc@s.)
(30 Dec 11, 01:25PM)Roflcopter Wrote: It does prove that some people are already advantaged versus non-automatic weapon players though.
I'll be the first to admit that I probably overestimated the role of the anticheat. Exactly.
But just calculate the average loss. And assume that we don't get this loss anymore w/o re-balancing the weapons. Automatic guns will become stronger. ;)
(30 Dec 11, 01:25PM)Roflcopter Wrote: I'm not sure actually. It may be to do with FPS somewhat and it's certainly to do with connection too.
Yes
(30 Dec 11, 01:47PM)|BC|Wolf Wrote: Lucas, you can't say that you won't even attempt fixing it just because then automatic weapons will be too OP... If that's really a problem, then reducing the automatic weapons' damage or whatever is needed to make it fair shouldn't bother many. I'd prefer to know that I hit all my shots but still wasn't able to get the kill, than know I couldn't get the kill because some hits didn't register.
Every bullet SHOULD register, and it is highly unwise to keep it that way just because then automatic weapons will be too overpowered. By the table itself, you can notice some players are less affected than others, therefore those players already have an instant advantage over everyone else. of course! i've never said we should NOT fix this.
But it requires to balance all weapons again imo
Posts: 441
Threads: 13
Joined: Jul 2010
30 Dec 11, 03:38PM
(This post was last modified: 30 Dec 11, 03:39PM by Jason.)
(30 Dec 11, 02:57PM)Luc@s Wrote: of course! i've never said we should NOT fix this.
But it requires to balance all weapons again imo
Then balance all the weapons again, no time to waste.
Posts: 799
Threads: 52
Joined: Jan 2011
30 Dec 11, 04:48PM
(This post was last modified: 30 Dec 11, 05:21PM by Roflcopter.)
Yet more players:
You can see that the two Belgium players (Dam and Jorik) are lowest.
Posts: 205
Threads: 3
Joined: Jun 2010
Good job Roflcopter, i had noticed this difference before, i just couldnt explain why or prove.
How about releasing a V1105 with this simple update while the weapons get balanced in the svn? Just an idea.
Posts: 256
Threads: 15
Joined: Jun 2010
It would be 1106 then, macm, and it's unlikely to happen.
Posts: 799
Threads: 52
Joined: Jan 2011
(30 Dec 11, 04:55PM)macm Wrote: Good job Roflcopter, i had noticed this difference before, i just couldnt explain why or prove.
How about releasing a V1105 with this simple update while the weapons get balanced in the svn? Just an idea.
I support this. I could patch my own client but it would be considered unfair. However if it's an official release then it's available to all equally.
Posts: 890
Threads: 16
Joined: Jun 2010
(30 Dec 11, 02:29PM)Roflcopter Wrote: As far as I know, FPS values aren't reported to the server so I would have to take everyone's estimates of their own FPS. I can start collecting ping and PJ though.
You're correct, fps is not reported to the server but with a stable connection and enough data you can likely figure it out.
Posts: 799
Threads: 52
Joined: Jan 2011
Posts: 1,436
Threads: 7
Joined: Jun 2010
30 Dec 11, 09:46PM
(This post was last modified: 30 Dec 11, 09:48PM by tempest.)
Here's a graph I quickly made to test the shot-dropping misbehavior for the AR: http://img840.imageshack.us/img840/6862/dropsk.png. I tested with /maxfps values of 40, 60, 90, 120 and 150.
Apparently, the only thing that matters (at least with a decent connection) is FPS. But before you all run for /maxfps 120 or something, remember that this is for the AR only, and I suspect it may look quite different for other weapons.
@Roflcopter: would be nice if you could ask some of those people for their average FPS.
Posts: 3,462
Threads: 72
Joined: Jun 2010
(30 Dec 11, 09:46PM)tempest Wrote: Here's a graph I quickly made to test the shot-dropping misbehavior for the AR: http://img840.imageshack.us/img840/6862/dropsk.png
Apparently, what matters (at least with a decent connection) is FPS. But before you all run for /maxfps 120 or something, remember that this is for the AR only, and I suspect it may look quite different for other weapons.
Why does it peak on multiples of a typical screen refresh rate? 60 and 120, that just seems odd.
Posts: 1,436
Threads: 7
Joined: Jun 2010
I don't think that has anything to do with the refresh rate. It's related to the firing speed. Just look at this graph for the SMG: http://img853.imageshack.us/img853/6081/dropssmg.png
I think this whole thing is due to the supposedly FPS-independent shooting, which was implemented in the client, but not in the server, so...
Posts: 617
Threads: 20
Joined: Sep 2010
larry i wanna try out your server. I want to see how bad this affects someone with dial-up.
Posts: 152
Threads: 20
Joined: Jul 2010
30 Dec 11, 11:16PM
(This post was last modified: 30 Dec 11, 11:22PM by Milandrag.)
is it just me or is it really that almost only w00p,MyS,legacy,epic w/e.... guys have the most low numbers? :?
Posts: 441
Threads: 13
Joined: Jul 2010
30 Dec 11, 11:22PM
(This post was last modified: 30 Dec 11, 11:30PM by Jason.)
(30 Dec 11, 10:59PM)lucky Wrote: larry i wanna try out your server. I want to see how bad this affects someone with dial-up.
Get on w00p| teamspeak / #mys IRC, and I'll hook you up with him. olol.
(30 Dec 11, 11:16PM)Milandrag Wrote: is it just me or is it really that almost only w00p,MyS,legacy,epic w/e.... guys have the most low numbers? :?
I agree with you on the w00p| thing, but you'll find Vault and MeLonE have it incredibly bad, who are =MyS= players - and quite frankly Benson and I aren't that great off. Larry aka Roflcopter from legacy is also one of the worst affected. That's a silly argument.
|
