There’s a lot of talk about collusion these days, and not just when it comes to Special Counsel Mueller’s investigation. TwoPlusTwo user FarseerFinland feels that team play is a rampant problem in the high stakes multi-table tournaments at PartyPoker, and appears to be correct. A representative for PartyPoker first informed him that some of the accounts in question had been suspended pending investigation and, just yesterday, made a public announcement that the investigation had resulted in the banning of a set of accounts and seizure of associated funds, which will be distributed to victims of the collusion.

PartyPoker has taken a very aggressive strategy with its tournaments over the past year, running a $2,600 Super High Roller every Sunday, plus $530 High Rollers most days. Many of these have generous guarantees and often run with overlays. All of this is part of a far-reaching strategy by PartyPoker to draw serious, profit-minded players away from PokerStars as the latter has turned its focus to a more casual crowd.

And yet, PartyPoker still isn’t nearly as large a site as PokerStars, getting less than 15% of its traffic at most times. As a result, many of these tournaments – weekday tournaments especially – run with small field sizes, and it’s that fact which makes them an appealing target for collusion teams.

Small field tournaments: an appealing target for collusion

There are many different ways for unscrupulous players to benefit from team play, but soft play on a tournament bubble is one of the easiest and safest. Other methods, like hole card sharing, often require making unnatural and inconsistent moves, which provide a site’s security team with concrete evidence of wrongdoing, and may even be picked up by automated systems. On a tournament bubble, all players naturally want to avoid all-in confrontations, so the difference between a colluding play and a tight one can be harder to recognize.

The main requirement for effective collusion – other than not getting caught – is that the team members can reliably find themselves at the same table in a scenario where they can create an advantage by working together. Single table sit-and-goes are the natural choice, but they’re also the most heavily scrutinized by site security for that reason; furthermore, they’re heavily populated by players who grind the same stakes for hours at a time, which makes the collusion much more likely to be spotted and reported by opponents.

Multi-table tournaments with large fields don’t fit the bill, because there are too many tables running at once when the bubble arises for a team to expect to find its players sitting together. And, of course, the value of a min-cash has to be significant before it’s worth it for the cheating parties to organize themselves in this way.

Weekday High Rollers at PartyPoker are therefore an ideal sort of target, with a min-cash being worth four figures and field sizes typically running in the vicinity of 100. That means money bubbles coming when play is down to one or two tables, which in turn means that a modestly-sized team runs a strong chance of finding itself in a position to collude.

That appears to have been the case in the High Roller Turbo on Wednesday, October 18, when FarseerFinland found himself at the final table with six opponents all apparently very reluctant to get all-in against one another. Effective stacks were in the 20 BB range, and yet in 48 hands, he noticed that no all-ins were called, except those by or against him. He nonetheless eliminated three of the six, after which the other three finished off the tournament almost immediately.

The smoking gun

On its own, that’s unusual but by no means conclusive, as weird runs of luck can certainly happen over the course of a few dozen hands. It was enough to trigger FarseerFinland’s intuition that something was amiss, however, and cause him to look up the opponents in question. As it turns out, the six opponents all have new accounts and all play very limited and nearly identical schedules, consisting most of the “Clubber II” tournament, a $100+9 Turbo event with a low guarantee and a field size often smaller than 50 entries.

While it’s entirely possible for individual players to have favorite tournaments and to play only those on a regular basis, it’s extremely unlikely to find six at the same final table together. It’s even more unlikely when the tournament in question is one of the site’s less popular ones, and the final table in question is a different tournament entirely, but one with a similar structure.

Good news or bad?

What’s most interesting to me about this story and others like it, is that neither the poker-playing public nor the poker media ever really seems to know how to respond to it. Just as with the PLO bot ring scandal at PokerStars in 2015, many people’s knee-jerk reaction is to lose confidence in online poker. Lee Davy, writing about the story for CalvinAyre.com, exemplifies this reaction.

Others, like former PokerStars media liason Michael Josem, have pointed out that news of cheaters being caught should be seen as positive. There’s nowhere in this world that significant amounts of money are changing hands without some people attempting to turn a dishonest profit. Counterintuitively, then, it’s possible to argue that it would be far more worrisome if we never heard about cheating in poker.

In between these two views is the uncomfortable truth that although it’s good that cheaters are getting caught, players would really feel a lot safer if it were site security teams catching them, rather than players doing their own sleuthing. The fact that the biggest cheating scandals are always first uncovered by posters at TwoPlusTwo leads to the understandable perception that sites are either incompetent or indifferent when it comes to catching cheaters.

Why players catch cheaters

There are two defenses that can be made of sites. The first is that when a site catches cheaters without player involvement, we don’t generally expect the public to hear about it. Security is the least transparent aspect of an industry which lacks transparency overall. The reason usually given is that any information disclosed by site security may be of use to cheaters in guessing at their methodology and avoiding their efforts. As a result, the amount of policing actually done by a security team may be quite different from the amount it’s perceived to be doing; PokerStars, for instance, has claimed that 80% of accounts banned for illegal software and 90% of those banned for collusion are caught internally, and only 20% and 10% respectively by player reporting.

That’s fair enough, but it’s understandable that players are unhappy not knowing how prevalent cheating is, how quickly cheaters are caught, how much money is being seized from them and so forth. Fortunately, this may change in the near future, as the UK Gambling Commission has implemented new regulations requiring licensed sites to report all details of account closures to the Commission. That doesn’t necessarily mean that the general public will have access to that data, but we can hope that the Commission will issue some kind of report in future which will shed some light on the situation.

The other excuse given by sites – including PartyPoker in this case – is that their methods are rooted in statistical analysis and therefore require a large volume of hands from a given player before a red flag can be raised. Players at the table, on the other hand, have their intuition to guide them; data-based methods require that all hands by all players be scrutinized, which is an enormous amount of work even with computer assistance. An instinct for spotting “suspicious” players and hands means that a human sleuth is implicitly discarding the overwhelming bulk of available data as non-suspicious, and examining only the situations most likely to involve cheating.

Collusion and the handshake problem

In the case of collusion specifically, there’s an additional problem in that it’s very difficult to spot by looking at individual players’ decisions in a vacuum; rather, it’s the interaction between two or more players that one needs to consider. A colluding player might have a perfectly normal 3-betting frequency from the big blind, for instance, but be doing it less frequently than usual against her partner, and more frequently than usual against other opponents.

Here, too, human intuition provides a useful shortcut, in that when we’ve seen slightly suspicious play, we’ll keep it in the back of our minds; if we then see similar situations recurring between the same pair of players, then we know where to look to find collusion. An automated system, on the other hand, has to struggle with what’s known in math as the handshake problem. If you have N people, then the number of possible two-person interactions (“handshakes,” or potential collusion scenarios) is equal to [N x (N-1)] / 2. This of course means that there’s a power law at work, and the number of scenarios that must be considered without the benefit of intuition grows as the square of the number of players in the group being scrutinized.

PartyPoker claims that it is expanding its security measures, both by introducing new, “highly advanced” detection techniques, and by forming a panel of elite online players who will manually review hand histories submitted to the site by players who’ve spotted something suspicious.

Unfortunately, the nature of digital security is that it is always an arms race. Whatever measures one has in place, someone will always eventually find their way around them; the only effective security system in the long run is one which is constantly adapting and improving, which is expensive and time-consuming.

This isn’t the first time a big cheating scandal has broken out, and it certainly won’t be the last. They say vigilance is the price of freedom, and that’s as true at the poker table as anywhere else. Sites will do what they can to protect their players, but it’s hard to imagine a future in which cheating never happens, or is always quickly caught. It’s one of the realities of the game we’ve chosen to play, and ultimately, your own intuition is your last and best line of defense.

Alex Weldon (@benefactumgames) is a freelance writer, game designer and semipro poker player from Dartmouth, Nova Scotia, Canada.