[0 carbon]

Some people cheat. Either they play both sides in 2 windows/tabs, or they share screens with P, or they talk with P..
But how can BBO identify them? Some players complain & someone may look at them, but that is labour intensive.

Some signs of cheaters:
- monthly or longer average >60%
- always play with same few Ps
- exceptional defensive play

Could BBO discover them via a computer program? AND distinguish cheaters from good players?

Well, I suggest using a sieve to identify cheating pairs.

• consistent average >60% resulting from successful bidding and play
• play on defense is much better than when declaring, since on defense they know the entire distribution. "Better" means more tricks, either directly compared to other players playing the same hand/trump/declarer, and/or using GIB play as a yardstick. (even better if BBO can compare same hands, same suit trump, same declarer)
• on defense, 1st lead is more successful than other players's 1st leads, or is consistently GIB's choice for best results.

This should identify a few pairs which are very lucky. If they continue to be successful in the next week or month, this is a good indication that they may be cheaters.
Good players will make good plays both declaring and defending, so will escape the sieve.
Lucky players will "revert to the mean" in following results, so they were "just lucky" for that period of time.

Then a manual review can done, especially of hands identified where the defensive play was exceptionally good. Was the play obvious from the particular bidding? Or did it show knowledge of partner's hand?

Then those IDs can be banned

tOM

[Vampyr]

Good players may well consistently achieve scores >60% on BBO, especially if they play only with one or a few partners -- which is again not a sign of cheating, but usually just means that a person is not willing to play with random strangers.

As far as identifying players who do better on defense than on declaring, it sounds like a pretty daunting task, not only in bterms of the size of the databases that would have to be created, but all the variables that would be involved.

I think you just need to accept that people will cheat, and hope that communication with Abuse will help weed out the worst offenders.

[nige1]

You can automate the 1st pass of cheat-detection by using specially selected extreme diagnostic deals, lucky and unlucky e,g, deals where...

• Game or slam makes with theoretically inadequate values because of benign breaks and a lucky lie of honor cards.
• Good-looking games or slams fail due to bad breaks and failing finesses.

Unfortunately, it's against the rules of bridge to deal such trap-hands deliberately. BBO could automatically search for them, however. The software could search for deals where the losing trick-count and law of total tricks fail miserably to predict the double-dummy result, because of unusually favourable or unfavourable distribution.

The software could flag such deals and monitor them for players who consistently achieve good results on them. This could provide an initial sieve to create a short-list of candidate-cheats for human investigation,

[diana_eva]

There already are a number of automated reports tracking unusually good performance for pay tournament players, but any cheating investigation eventually requires humans checking thoroughly why someone made it to the shortlist.

It would be very expensive and time consuming to do this for all the free games. Also pretty useless since offenders can find a way to get back under a new name relatively easy. Such a process can't be fully automated unless you find it acceptable to auto-ban someone for being too good.

[The_Badger]

diana_eva, on 2017-September-14, 01:11, said:

Also pretty useless since offenders can find a way to get back under a new name relatively easy.

Just block their IP address, Diana. At least they will have to purchase a new bit of computer kit to carry on, and that is inconvenient in itself. (Yes, I realise people can access the internet through phones, shared computers, etc. but block those IP addresses too. In the end they will get the message loud and clearly.)

[steve2005]

nige1, on 2017-September-14, 00:52, said:

You can automate the 1st pass of cheat-detection by using specially selected extreme diagnostic deals, lucky and unlucky e,g,

This is a daughting task. It is my understanding that BBO does this already to some extent or can do it for a specific person if suspicion is read. Obviously BBO for security reasons won't say what specifically they look for.

2 examples that show how it could be hard,
1) my partners cat managed to click 6N response to my 1H opening! Of course it makes and nobody would be it. I guess you could bar the cat!
2) This was in face to face but would raise suspicion if online. 2D mini-roman open -> 2H sign-opp 4+h 0-8 pts. (with 3 queens 1444) 4H bid from opener. 20 point game rolls. Opener says I was max(15) with a singleton so had to.

You would have to slog through large numbers of strange hands. Could only be done with computer to find possible cheats.

[barmar]

The_Badger, on 2017-September-14, 02:56, said:

Just block their IP address, Diana. At least they will have to purchase a new bit of computer kit to carry on, and that is inconvenient in itself. (Yes, I realise people can access the internet through phones, shared computers, etc. but block those IP addresses too. In the end they will get the message loud and clearly.)

We do block IPs, but people get around them.

And don't forget that lots of people still access the Internet from ISPs that don't assign permanent IPs to customers.

[j_with_a_B]

I host free tournaments on a daily basis. Cheaters are a perennial problem. When players in the same tourney regularly reach the top 5 in the standings, it makes me wonder. When a pair makes 85%+ as top pair and the next is in the low 70's, my attention is peaked. I used to blacklist some pairs for excessive "luckiness" in an attempt to preserve fairness for the rest of the players. I do this now only in exceptional cases, simply because they too easily return with freshly minted nicknames and carry on as before. Then I need to spot them all over again. It never ends. Use of the TCR and banning new players can help to some extent, but it also catches newcomers to BBO as well as many that have been away for an extended period and have had their TCR lapse.

Banning on the basis of IP address is extremely problematic since most of them are assigned by ISP's on a dynamic basis. All that is needed is to turn off and unplug the modem or router for a short time and a new one gets assigned when they reconnect. Also, there is the problem that it is possible for innocent players to acquire a banned IP address through no fault of their own and be unable to connect.

I have had some entertaining encounters with players that play both hands on a few occasions. I have asked a question privately of one partner only to have the other one reply. I have prompted players to play on a few times and had them say they could not. I have helpfully suggested that they play from the other ID and have been thanked.

I very often have players claim the opps are cheating because of a single lead or bid. When the results come in at the end I try to point out that when they scored 39% and came in in the bottom third of the pack, then, if they are indeed cheating, they are not very good at it and should continue since they obviously need to practice more.

Cheating is very difficult to prove. However, rudeness is not. I have a macro that I run as needed. It runs like this:


I consider unsubstantiated allegations of cheating to be extremely rude!
Some players suggest the opps are cheating when they misclick or play slowly or answer another opp's question or even when they just play better and beat them fairly...
If you suspect an opp of cheating then report it to me or another TD!
If that is not to your liking, then send your proof to BBO at: abuse@bridgebase.com
DO NOT be so RUDE as to accuse the opponent directly!
I may remove you from the tourney instantly for rudeness!
Too often such allegations are unsubstantiated and are, at best, only suspicions or misunderstandings. They amount to nothing more than gossip.
If you have real proof, then it can be verified and the players involved can be suitably dealt with...


I think Tom (0 carbon) is on to something. Surely a simple algorithm could be devised to spot "outstanding" "lucky players". I hope further discussion by people that know more about such things than I might lead to some simple automated tools that could identify a short list of candidates for human attention. Perhaps they could be used on an ad hoc basis if suspicions arise and suspects are presented for examination. I know many regular average players comment to me about how discouraging it is to see the same players win some tourneys all the time. I watch the results each time and know my regular players well. Some do win on occasion but I am confident they do not cheat when I see them near the bottom of the results as often as at the top.

John

[Vampyr]

j_with_a_B, on 2017-September-14, 16:25, said:

I know many regular average players comment to me about how discouraging it is to see the same players win some tourneys all the time.

Well, is a game of skill. Chess, backgammon, poker... the top players win much more than most.

[j_with_a_B]

Vampyr, on 2017-September-14, 20:48, said:

Well, is a game of skill. Chess, backgammon, poker... the top players win much more than most.

Very true. But it becomes obvious that skill at cheating is the only one involved when the margin of the win is so wide and consistent. Even champions do not win all the time. It is hard to believe that something more than skill in the game is not involved when players bid their partners' suits, always make finesses that most would never try and lead to P's voids instead of their bid suits. Skill and luck are no match for foreknowledge. I very often argue on behalf of the accused in their defense when I realize they play well and do so several times every day, day after day with their regular partner of several decades. I am lucky to have had the chance to get to know most of my regulars and their levels of competence. When 2 players are partners and both nicknames are new to BBO and created the same day or only a few days apart or are names I have never seen before they often catch my attention. When they score in the high 80's in Match Points over 9 or more boards even one time I tend to note them mentally. If they return and come first or second every day for a week or more I cannot help but be more suspicious. As I said, in the past I often would blacklist them until a new pair with the same patterns quickly took their place.

I am hoping some sort of software tool can be developed that can examine such cases objectively so the burden of detection and analysis is somewhat lifted from the TD's. My tourneys are all free ones and there are no Master Points at stake and the prize money is not a factor. I appreciate BBO's policy that in these free tourneys the Hosts and TD's are free to exclude anyone for any reason, no questions asked or justification needed. I am sure I have blacklisted a few players in error in the past. I know I allow some to continue to play that other TD's have long prohibited. The respectable players would simply prefer not to swim in a pool where the top is covered in pond scum. Surely a decent tool for analysis of hands for evidence that one player or pair of players may have known more about other hands in the deal than they should have been able to would be worth developing. BBO would be the ideal place for it to be developed. From what I hear, the world of high level international competitive bridge could benefit from something of the sort if only to exonerate some high money prize winners of wrongdoing.

John

[0 carbon]

j_with_a_B, on 2017-September-14, 16:25, said:

...
Banning on the basis of IP address is extremely problematic since most of them are assigned by ISP's on a dynamic basis. All that is needed is to turn off and unplug the modem or router for a short time and a new one gets assigned when they reconnect. Also, there is the problem that it is possible for innocent players to acquire a banned IP address through no fault of their own and be unable to connect.
...
I think Tom (0 carbon) is on to something. Surely a simple algorithm could be devised to spot "outstanding" "lucky players". I hope further discussion by people that know more about such things than I might lead to some simple automated tools that could identify a short list of candidates for human attention. Perhaps they could be used on an ad hoc basis if suspicions arise and suspects are presented for examination. I know many regular average players comment to me about how discouraging it is to see the same players win some tourneys all the time. I watch the results each time and know my regular players well. Some do win on occasion but I am confident they do not cheat when I see them near the bottom of the results as often as at the top.

IP banishment and ID banishment have problems...

Since they play to win, maybe when identified, BBO can just maladjust their hands, preferably with a few tricks to play, so they do poorly and get continually frustrated. Especially when the hand has mystical bidding.

TDs could do this too, but hard to watch 1 table all the time.....

tOM

[zhasbeen]

After signing in I see that the discussion is about live games, rather than the Robot tournaments that I play in. From all I understand there is no way to cheat in those games. What drew my attention were comments saying that anyone who averages 60% is probably cheating.

As a robot fanatic, I record the date, tournament number, number of tables at start of game, and overall placing, for every tournament I play. For match points I record percentage and IMPS margin for IMPS.
I currently have a 61.06% average for a 145 match point games I've played since joining in mid-August. For IMPS I am averaging +17.99 for 92 matches.

Right about now I'd be thinking, man, is this guy ever full of himself.

The truth is is that, while I'm a considerably above average duplicate player, I'm not close to being in the top tier, even in my district, let alone nationally. My point is that you don't have to be great player or cheat to average 60% playing against the BBO robots. I'll take credit for being pretty good, though.

While I'm on this topic, any of you who play only the live games can improve your play a bunch by playing is these robot games. They will drive you nuts sometimes with their crazy bids, but they do lots of good things. While they make plenty of errors, they also lay down some defenses that would make Zia Mahmood proud. They make more good plays than bad. This is where they can help you the most--on card play. I guarantee you will improve your declarer play if you play enough of these. A huge plus is that you can see "movies" of every hand you played. Although it can sometimes be depressing/embarrassing to watch them, they are extremely helpful.

I only keep records of the 12-Board ACBL sanctioned events that cost $1.25, although I occasionally play in the 8-Board 25 centers. Only points earned in ACBL games count on your offical record kept by ACBL, although the point totals displayed on BBO include all points.

If you haven't tried it you ought to. Just make sure you print out a copy of the GIB convention card and familiarize yourself with it. In the clubs you aren't supposed to look at your card during the bidding, so I guess in that respect I could be accused of cheating. However, you can run your mouse cursor over the bids as they are happening and there will be cryptic meaning. I don't have to look at the card much anymore, but I use the mouse cursor thing regularly.

[johnu]

0 carbon, on 2017-September-18, 21:58, said:

IP banishment and ID banishment have problems...

Maybe BBO could send a 220 volt feedback loop to those *ssh*l* cheaters. Not enough to permanently fry their brains, but enough to make them permanently stay away from BBO.

[0 carbon]

zhasbeen, on 2017-September-22, 21:56, said:

After signing in I see that the discussion is about live games, rather than the Robot tournaments that I play in. From all I understand there is no way to cheat in those games. What drew my attention were comments saying that anyone who averages 60% is probably cheating.
As a robot fanatic, I record the date, tournament number, number of tables at start of game, and overall placing, for every tournament I play. For match points I record percentage and IMPS margin for IMPS.
I currently have a 61.06% average for a 145 match point games I've played since joining in mid-August. For IMPS I am averaging +17.99 for 92 matches.

Yes, good players can get caught in the "good results" sieve. That is why more analysis is needed. I suggested psychic defence is the best significator.

For DD analysis of many deals, download http://bridgecaptain...downloadDD.html You can do this for your own or others' deals.

[0 carbon]

Microsoft to the rescue: https://arstechnica....r-windows-devs/

[barmar]

0 carbon, on 2017-October-21, 10:56, said:

Microsoft to the rescue: https://arstechnica....r-windows-devs/

I don't think the checks in that software are relevant to the kind of cheating we're worried about.

[johnjo42]

I can’t see how you wil ever stop the most blatant cheating such as talking to partner on phone but I do wonder what the point of these people is. What exactly do they hope to gain. (Not talking about money bridge or Robot tourneys).

[barmar]

johnjo42, on 2017-November-15, 16:35, said:

I can’t see how you wil ever stop the most blatant cheating such as talking to partner on phone but I do wonder what the point of these people is. What exactly do they hope to gain. (Not talking about money bridge or Robot tourneys).

I have two main theories:

1. Ego boost.
2. Some people just like the idea of pulling a fast one.

When OKbridge first started, and it was only usable from Unix computers, I had a bridge friend who worked in a computer lab, so he was able to login to the computer on multiple terminals. He created multiple OKbridge accounts and played with himself as both North and South. He was a decent player, so he didn't need to cheat like this, he just saw it as an easy opportunity to scam the system. He was a college student, so maybe not the most mature at the time.

[0 carbon]

zhasbeen, on 2017-September-22, 21:56, said:

As a robot fanatic, I record the date, tournament number, number of tables at start of game, and overall placing, for every tournament I play. For match points I record percentage and IMPS margin for IMPS.
I currently have a 61.06% average for a 145 match point games I've played since joining in mid-August. For IMPS I am averaging +17.99 for 92 matches.

You can also see your past hands at http://www.bridgebase.com/myhands/
and download them all & analyze them using http://bridgecaptain...downloadDD.html

Robots can play pretty well, but the bidding tends to get lost....

tom