It mean that I as an individual can have 3 profiles on my computer at home, and justify to say that I have 2 brothers who also play WG? That would enable  to me playing over the allowable limit (as a standard member) to 30 games. Would not necessarily have to play with myself and always let one profile to win, but you could use "smart cheating" to disturb the opponent with one profile while the other develops. I can attack myself sometimes to avoid suspicion, but mostly attack "stranger" player. Such little things can be easily reversed in my favor. not addressing anyone personally, but I think that multiple profiles with the same IP address should be keep on eye.

1) Yes it's possible that cheaters will be surriptitious enough that they won't get caught.  And there's nothing that anyone is going to be able to do to fix it.  It's a necessary evil.  So trying to police for these situations seems counter productive.

2) I think that if you look over people who have been found to be cheaters, their play was obviously malicious.  As long as we're still requiring that obvious of cheating as the reason to be banned (which everyone seems to agree requires human intervention) then showing or not of matched IPs won't create "false positives" because play over time will be the ultimate indicator

3) Independent of the above comment, I strongly believe that the numbers shouldn't be shown.  False positives or not, I think that the more pertinent argument is that by making that information available to even the lowest of players, the admins now giving all players justification to complain about even obviously fair play.  In the end, it'll just create bad blood in games, and that will deter from the game play.
If the information isn't available on every game screen, then it no longer becomes fodder for an argument.  I'm a firm believer that finding cheaters should stay in the camp of the admins.  Let the admins, whose job it is to police, do the policing.  I want the average players time time playing to be focused on the fun of the game, not worrying about cheaters.

Now, if you want to say that information is available only to a (now somewhat larger) select group of the top ranked (and hopefully unbiased) players so that they can help the admins do the policing, I'm okay with that.  But I don't want Joe "This is my first game and I just got my butt handed to me fair and square" Player to have it available to him.

i think as the only non cheater on the site is that awesome guy rasputin, he should be given all points and all trophies.  Done deal.  Thank you for your time.

Sincerely,

Not Rasputin

gosh im good at grammar

BorisTheFrugal wrote:

3) Independent of the above comment, I strongly believe that the numbers shouldn't be shown.  False positives or not, I think that the more pertinent argument is that by making that information available to even the lowest of players, the admins now giving all players justification to complain about even obviously fair play.  In the end, it'll just create bad blood in games, and that will deter from the game play.

I want the average players time time playing to be focused on the fun of the game, not worrying about cheaters.

I think this is a great comment - that's precisely my concern. You only need to look a little further up this thread for evidence of this.

One other option I have considered is using our resident math / coding geniuses to put together a system for analyzing the weekly game log file and flagging up potential cheaters for review.

The log file would be modified so that there would be no personally identifiable information. For example IP addresses would be hidden via hashing function so they are still unique but not identifiable. Similarly player names would be referenced by a unique id number instead of their player name.

Based on the following information:

GameID
Seat
Timestamp
Action
IP

It should be possible to flag up likely cheaters - e.g. players who play on the same IP who take turns immediately after each other who rarely attack each other and where one player always wins if it gets down to 2 players at the end.

M57 wrote:

I would think that most if not all cheaters can be identified by their play, not their IP addresses.  I've said this before, but I believe cheating is pretty easy to detect because unusual patterns emerge, and once players start to post good numbers they put themselves in the spotlight where their play is observed even more closely.  WG has an active community of players in the top 200, and I'd say at least 10% of them are regular posters on these forums. (some more paranoid than others - and that's OK). 

BYW, I play against co-workers (same IP address) all the time and though I can't speak for their agendas, I'm out to kick their assess. 

Whenever possible M57, whenever possible!

tom wrote:

One other option I have considered is using our resident math / coding geniuses to put together a system for analyzing the weekly game log file and flagging up potential cheaters for review.

The log file would be modified so that there would be no personally identifiable information. For example IP addresses would be hidden via hashing function so they are still unique but not identifiable. Similarly player names would be referenced by a unique id number instead of their player name.

Based on the following information:

GameID
Seat
Timestamp
Action
IP

It should be possible to flag up likely cheaters - e.g. players who play on the same IP who take turns immediately after each other who rarely attack each other and where one player always wins if it gets down to 2 players at the end.

Referring to my earlier comments, I agree with this one as well.  The process for comparing IP addresses should be done behind-the-scenes and should flag the discrepancy for an admin to look at.  The admin then can look at the individual game behavior and determine if the multiple accounts are truly cheating.

Rasputin wrote:

i think as the only non cheater on the site is that awesome guy rasputin, he should be given all points and all trophies.  Done deal.  Thank you for your time.

Sincerely,

Not Rasputin

Too bad he isn't the REALRasputin...that guy was awesome.

AttilaTheHun wrote:

tom wrote:

One other option I have considered is using our resident math / coding geniuses to put together a system for analyzing the weekly game log file and flagging up potential cheaters for review.

The log file would be modified so that there would be no personally identifiable information. For example IP addresses would be hidden via hashing function so they are still unique but not identifiable. Similarly player names would be referenced by a unique id number instead of their player name.

Based on the following information:

GameID
Seat
Timestamp
Action
IP

It should be possible to flag up likely cheaters - e.g. players who play on the same IP who take turns immediately after each other who rarely attack each other and where one player always wins if it gets down to 2 players at the end.

Referring to my earlier comments, I agree with this one as well.  The process for comparing IP addresses should be done behind-the-scenes and should flag the discrepancy for an admin to look at.  The admin then can look at the individual game behavior and determine if the multiple accounts are truly cheating.

I agree.... I don't think we should be privy to anymore information than we have now and I would hate to see the mayhem making the information public would cause. 

Yes - I sometimes play with people that at are at the same IP as me, and I'm not looking forward to the inevitable claims of collusion.

tom wrote:

One other option I have considered is using our resident math / coding geniuses to put together a system for analyzing the weekly game log file and flagging up potential cheaters for review.

The log file would be modified so that there would be no personally identifiable information. For example IP addresses would be hidden via hashing function so they are still unique but not identifiable. Similarly player names would be referenced by a unique id number instead of their player name.

Based on the following information:

GameID
Seat
Timestamp
Action
IP

It should be possible to flag up likely cheaters - e.g. players who play on the same IP who take turns immediately after each other who rarely attack each other and where one player always wins if it gets down to 2 players at the end.

I don't understand.  What's the point of anonymously flagging accounts when you're going to have to look at the individuals' games at some point anyways to verify that there's actual cheating?  Or are you talking about making this information public ..or semi-public for members to evaluate?

tom wrote:

One other option I have considered is using our resident math / coding geniuses to put together a system for analyzing the weekly game log file and flagging up potential cheaters for review.

The log file would be modified so that there would be no personally identifiable information. For example IP addresses would be hidden via hashing function so they are still unique but not identifiable. Similarly player names would be referenced by a unique id number instead of their player name.

Based on the following information:

GameID
Seat
Timestamp
Action
IP

It should be possible to flag up likely cheaters - e.g. players who play on the same IP who take turns immediately after each other who rarely attack each other and where one player always wins if it gets down to 2 players at the end.

Not really wanting to discuss any other part of this discussion, I would love to look at such data (especially with some historical examples where cheaters were present and identified for baseline).  This would be a fun task and the only small suggestion I would make is for the history to be a month or two (not many games end in a week between the same pair of players).

One strong indicator of cheating would be attacking another player with only 1 or 2 units.  In normal game play it is very rare that this is a good idea, but would be common if you are trying to weaken yourself so that another player can take that territory. 

This happens to me about once a game since I use shortcut keys.  I also like to add noise to the data.

Alpha wrote:

 I also like to add noise to the data.

lol  ..Really.

Another easy-code change that could help lay people make an easy assessment of suspect play, and pass the info onto the admins:

I think we all agree that the most flagrant of cheaters will play two accounts in the same game, and only one of them will reap the benefits.
So within the player information screen, maybe the code can be corrected such that sorting of game by "Status" will sort the status alphabetically.
Since all finished games say "XXXX Won", then all games lost by the secondary account to the beneficiary account in the above scenario will be listed together.
This would give someone an easy way to count that 78% of the secondary account's losses were to the beneficiary account.

Just a thought.
...and maybe this should be a post on its own??

BorisTheFrugal wrote:

Another easy-code change that could help lay people make an easy assessment of suspect play, and pass the info onto the admins:

I think we all agree that the most flagrant of cheaters will play two accounts in the same game, and only one of them will reap the benefits.
So within the player information screen, maybe the code can be corrected such that sorting of game by "Status" will sort the status alphabetically.
Since all finished games say "XXXX Won", then all games lost by the secondary account to the beneficiary account in the above scenario will be listed together.
This would give someone an easy way to count that 78% of the secondary account's losses were to the beneficiary account.

Just a thought.
...and maybe this should be a post on its own??

I think all this can be easily seen/sorted via the Profile >> Opponents page, then the pop-up games list window.

Sweet mamma jamma, you're right.  Probably would pay me some to actually open and look at each page on this website, as opposed to just using the ones I know how to use.

Thanks for the tip Yertle!!

M57 wrote:

tom wrote:

One other option I have considered is using our resident math / coding geniuses to put together a system for analyzing the weekly game log file and flagging up potential cheaters for review.

The log file would be modified so that there would be no personally identifiable information. For example IP addresses would be hidden via hashing function so they are still unique but not identifiable. Similarly player names would be referenced by a unique id number instead of their player name.

Based on the following information:

GameID
Seat
Timestamp
Action
IP

It should be possible to flag up likely cheaters - e.g. players who play on the same IP who take turns immediately after each other who rarely attack each other and where one player always wins if it gets down to 2 players at the end.

I don't understand.  What's the point of anonymously flagging accounts when you're going to have to look at the individuals' games at some point anyways to verify that there's actual cheating?  Or are you talking about making this information public ..or semi-public for members to evaluate?

Sorry I should have been clearer.

I am looking for help from someone with statistical / analytic / mathematical experience to assist with putting together a system / algorithm / procedure for flagging up likely cheaters from a sample week or months log data.

As I can't pass on any personally identifiable information like IP addresses to a third party (as this would be a breach of privacy) personal information in this log data would need to be hashed to make it anonymous.

Once a system has been put in place this would be a regular process which would run in the background each week and pass names for checking to me or another admin. There would be no need for hashing / obscuring information as this would be a purely internal function.