Republic Veteran
Join Date: Jun 2012
Posts: 613
# 71
03-04-2013, 10:49 AM
Vote to kick works fine in every other game I play. You guys are freaking out over nothing. As normal. Bring it on.
____
* * *
STO community Doom-O-Meter: |||||||||
Career Officer
Join Date: Jun 2012
Posts: 867
# 72
03-04-2013, 10:51 AM
Quote:
Originally Posted by therealted View Post
Not sure if you're referring to my post. If not, eh, I got time to kill anyway...

If I'm reading logicalspock right, what he's talking about is an alternative to the votekick system, where players who "vote down" a target just add a demerit of some sort to the target's record. Once the target accumulates enough demerits, he/she/it is banned from the queue for a certain amount of time. So it's not a 4-votes-you're-out system.

Given that the current chat ban system works on a similar basic model, and given the nature and number of complaints concerning abuse of that system here and in CO, the griefing potential should be apparent.

Again, if I'm reading things right, logicalspock is claiming that a sufficiently well-designed system will ignore "background noise" from casual griefing (e.g., people voting someone down for character or personal names, or even just nuisance effects). My argument is that the system, if automated, can be easily broken to the point of uselessness, and that there are people who will happily band together to do so.

Of course, if I'm not reading him right, I'd welcome clarification.


IMO a vote-kick system simply meets abuse with more abuse, but maybe that's another thread...
It is not that you are "not reading this right". It is that your belief is disproved by the central limit theorem. If it really can be, "broken to the point of uselessness," please point out how. Similar systems have proven incredibly reliable in other applications.
Career Officer
Join Date: Aug 2012
Posts: 953
# 73
03-04-2013, 10:52 AM
Quote:
Originally Posted by commanderkassy View Post
Vote to kick works fine in every other game I play. You guys are freaking out over nothing. As normal. Bring it on.
I've had exactly the opposite experience with such systems, so I guess both our opinions are equally valid or invalid and therefore do not really further the discussion.

Considering the difficulty Cryptic has with introducing most new things into the game can you really guarantee it will "work as intended" first off and if it will be a perfect systems even after patching?

as it stands I couldn't make that claim on behalf of Cryptic.
Career Officer
Join Date: Aug 2012
Posts: 953
# 74
03-04-2013, 10:55 AM
Quote:
Originally Posted by logicalspock View Post
It is not that you are "not reading this right". It is that your belief is disproved by the central limit theorem. If it really can be, "broken to the point of uselessness," please point out how. Similar systems have proven incredibly reliable in other applications.
You know CLT is not perfect right and does not always return accurate normal distributions.
Career Officer
Join Date: Jun 2012
Posts: 867
# 75
03-04-2013, 11:08 AM
Quote:
Originally Posted by thisisoverlord View Post
You know CLT is not perfect right and does not always return accurate normal distributions.
The central limit theorem is proved true. I think you mean to say that real distributions involving a finite number of data points will not always conform to normal distributions.
Career Officer
Join Date: Jun 2012
Posts: 867
# 76
03-04-2013, 11:15 AM
Quote:
Originally Posted by thisisoverlord View Post
It proves nothing as it does not work accurately with human beings as variables. Especially when they can have multiple accounts and alts, this is not the simple solution you think it is, applying statistical systems to human beings has always been flawed in practice when it often seems perfect in theory.

I would be interested in you actually pointing out specific weaknesses in the idea. However, making vague and unsubstantiated claims is not useful.

For example, you claim people have multiple accounts and alts, but you fail to point out how that is relevant. Of the millions of STFs that have been played, do you really think that such a large number of the players are part of some grand conspiracy to alter game mechanics? It would be a rather easy theory to test.

And it would also be a rather easy problem to solve. Just like the developers would have to pick and monitor a reasonable critical point to ban players from STFs for receiving too many reports, they would have to pick and monitor a critical point to ignore a players who report others too often.

It would be trivial to build in safeguards, such as using a hypothesis test to determine if a certain day's data is too skewed to be a coincidence.

The key would be to spend months gathering the data before the ban system was turned on, so there is a sufficient amount of test data from which to draw conclusions.
Lieutenant
Join Date: Jul 2012
Posts: 53
# 77
03-04-2013, 11:19 AM
Quote:
Originally Posted by thisisoverlord View Post
You know CLT is not perfect right and does not always return accurate normal distributions.
Even if it did (and you're right, it doesn't), it completely falls apart outside the domain of independent random variables - which is exactly where you'd be when a group of players take it upon themselves to mess with the system.

If the system is designed to figure out a "reasonable" amount of demerits (or whatever) to trigger a queue ban via statistical analysis, then even a small group of people consistently and deliberately slapping demerits on everything that moves will push that threshold above the point of usefulness, and broaden the standard deviation so much, that even chronic AFKers would look normal.

In real life, statistical analysis works when the behaviors it covers are things that people can't muck about with in a way that isn't detrimental to themselves. In a video game, there is no such limit - and as I said, there is no shortage of players who will push the envelope just because they can.

logicalspock, I really feel like I am missing something critical in your proposal. Some pseudo-code or other description of the algorithm you'd use could really be handy right about now...
____________________

The gorilla formerly known as Kolikos
Career Officer
Join Date: Aug 2012
Posts: 953
# 78
03-04-2013, 11:23 AM
Quote:
Originally Posted by logicalspock View Post

The key would be to spend months gathering the data before the ban system was turned on, so there is a sufficient amount of test data from which to draw conclusions.
Once again, the point as has already been raised in the post above this one is your water-tight statistical theory is a little leaky when it comes to application and there is indeed a question over whether it is entirely useful as an application in this instance.

But ignoring all that this final line of yours is what is really important, Cryptic simply wont bother to waste that amount of time and effort to get the system working right. so your solution as it is would never be considered.

Furthermore the problem is without the system actually running you will never be able to collect data on the frequency with which players were abusing the system, nor would testing it on Tribble help considering the playerbase sample size there and the demographic make-up which leans towards a lack of trolls/griefers/abusers

Last edited by thisisoverlord; 03-04-2013 at 11:29 AM.
Captain
Join Date: Jul 2012
Posts: 560
# 79
03-04-2013, 11:34 AM
Why not just track the contributions of each team member, then reward the ones who contribute over a certain threshold?

Track how much each player kills/damages/heals in the team. I don't think AFKers who are using keyboard macros would reach that threshold by fire autofiring a few potshots at enemies and AFKers who sit there doing nothing get no reward at all, but a reprimand from the issuing NPC for doing nothing and a cooldown on joining any PUG mission of that type.

But maybe that would be a headache for programmers to implement...
Career Officer
Join Date: Aug 2012
Posts: 953
# 80
03-04-2013, 11:39 AM
Quote:
Originally Posted by lonnehart View Post
Why not just track the contributions of each team member, then reward the ones who contribute over a certain threshold?

Track how much each player kills/damages/heals in the team. I don't think AFKers who are using keyboard macros would reach that threshold by fire autofiring a few potshots at enemies and AFKers who sit there doing nothing get no reward at all, but a reprimand from the issuing NPC for doing nothing and a cooldown on joining any PUG mission of that type.

But maybe that would be a headache for programmers to implement...
This would be a better solution but you'd have to be careful how you implemented it.
Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is Off
HTML code is Off


All times are GMT -7. The time now is 05:12 AM.