ESPN published a fascinating feature story on disgraced former NBA referee Tim Donaghy last week, to mark the 10th anniversary of his sports betting scandal first coming to light. It’s a very long read, so if, like me, you missed the story when it came out a decade ago, let me give you the nutshell version before I launch into my own take on the story.

A classic downward spiral

Donaghy was a veteran NBA referee who, at the time, was regarded by many of his peers as being among the best in the business. He also happened to acquainted since adolescence with an extended group of friends involved in gambling generally and sports betting specifically. Some had gotten into that world deeply enough to be acting as brokers for other high rolling bettors, or even running their own offshore sportsbook, PlayASAP.

NBA refs are contractually forbidden to gamble (except on horse racing, for some reason, an interesting aside from the ESPN article) but the circles he ran with were heavily into that world, and he himself was inexorably drawn in. It began innocently enough, betting on rounds of golf with his friends, then making trips to the local casinos. Eventually, though, his friend Jack Concannon convinced him to pool their money together to bet on NFL and college football picks supplied by another friend and bet broker, Peter Ruggieri. From there, they moved on to NBA games, and finally to games Donaghy himself was refereeing.

Unsurprisingly, bets on games officiated by Donaghy had a tendency to be winners. At some point, Concannon got greedy and started raising additional money to bet on the games without Donaghy’s knowledge. But he was placing those bets with PlayASAP via Ruggieri, and his acquaintances at the company didn’t take very long to notice that some bets were much larger than others, and that those bets were winning too often. Because everyone in the circle knew each other at least a little, it wasn’t hard to make the more important realization that these bets were all on games for which Donaghy held the whistle.

One partner at PlayASAP, James Battista, approached Donaghy and informed him that Concannon was getting rich off his picks without him, and convinced him to give him the picks instead. In return, he’d give Donaghy a freeroll: any pick that proved correct would earn Donaghy a fee, initially $2000 per game, later more.

Inevitably, Battista got greedy as well, and began wagering so heavily on games that sharps around the world were taking notice of how the lines were swinging on certain games. Some put the pieces together themselves and began profiting off the scheme as well, despite not having direct access to Donaghy’s picks. Some of these had connections to the Gambino crime family, which eventually brought matters to the attention of the FBI. From there the whole thing came crashing down on Donaghy, Battista and others who’d become involved over the years.

Universal lessons

Donaghy’s slide from recreational gambling to unwittingly helping to fund the operations of the Gambinos is an interesting true crime story in its own right. What’s most fascinating to me, though, is how his activities managed to go undetected for years when they seem like they should have been obvious in retrospect. There are several factors at play, and it’s interesting and illuminating to look at how they relate to other forms of cheating, including in poker.

In large part, Donaghy wasn’t caught because no one wanted to catch him. Other gamblers were more interested in getting in on the scam than bringing it to light, and the NBA’s primary concern was avoiding a scandal. In fact, he might have gone on fixing games for years longer if not for the fact that a third party, the FBI, came into the picture with a separate motivation entirely – removing sources of income for the Gambino family – which happened to require taking Donaghy down in the process.

The many-observer fallacy

The NBA and other major sports leagues generally claim that match-fixing is impossible because there are simply too many parties with a vested interest watching at once for any funny business to escape detection. A similar argument is also put forward quite commonly in the “is online poker rigged?” debate, namely that the volume of online play, number of people involved and ability to automate data collection (either by the site or by players using third-party software) taken together render the odds of a cheater escaping detection vanishingly small.

There are two basic problems with this, one more relevant to poker than the other. The first is the noise generated by false positives, or the “Boy Who Cried Wolf Effect.” When it comes to interpreting a chaotic set of data (like board runouts in poker or the distribution of referee calls in a sports match), humans fall victim to a whole host of different psychological biases, especially when – as in poker or sports betting – there is both a hope for one outcome and a worry that things might be slanted towards the other.

As a result, virtually every sports fan, including those who don’t bet but are merely emotionally invested in one team, feels at some point as if there’s some sort of officiating conspiracy against them. Likewise, even the most rational poker player can’t help but entertain paranoid fantasies about cheating opponents, a rigged deal or some sort of supernatural curse when experiencing the downside of variance. Some people become so convinced based on their own anecdotal evidence that they end up going public with unsubstantiated claims, which poses a credibility problem for other such claims, even those that are backed by evidence. Getting others to even look at the evidence requires first convincing them that you’re not just another fool in the throes of confirmation bias.

The second problem can be more or less of an issue depending on the nature of the wrongdoing. This is that not all observers are as scrupulous as others. Upon taking notice of what’s going on, some might prefer to look for a way to use that information to their advantage rather than bring the matter to light. This was certainly the case in the Donaghy story, though ironically, what ultimately brought it to an end was that there were simply too many people simultaneously trying to keep it under wraps so they could cash in themselves.

Fortunately for poker, this isn’t typically the case, though it’s possible to construct hypotheticals in which it would be. If opponents are colluding, or running bots online, or dealing seconds in a live game, there’s probably nothing an astute observer could do with that information except go public with it. But imagine, for instance, a situation in which a player in a live game notices that certain cards are marked, but doesn’t know who did it. Some players would call the table’s attention to that fact and get the deck replaced, but others might elect to keep it on the down-low; after all, having noticed this, they’re no longer at a disadvantage against whoever did it, but rather now share in an advantage against the rest of the table.

Selective myopia

One striking aspect of the Donaghy story is that even after he was exposed and an internal investigation was carried out, the NBA’s official story was that despite betting on his own games, Donaghy had done his job properly on the court and had not attempted to influence outcomes. Not only that, but the FBI didn’t even attempt to make that case in court.

It shouldn’t even be a question that Donaghy influenced results. Even if he were not consciously attempting to, refereeing involves a certain margin of error, and common sense dictates that a referee who stands to benefit from one outcome will tend to err on the side of that outcome. Donaghy himself publicly denied having deliberately manipulated scores, but effectively admitted in private that this was a lie (by saying he’d fail a polygraph test). Interviews and testimony from others involved in the scheme indicated that the understanding was that he was doing so.

Once again, though, the fact that the NBA failed to confirm the obvious is merely depressing and not shocking when you consider the incentives involved and the human capacity for self-deception. The NBA’s primary interest was in maintaining the fiction that it’s “impossible” to rig a major league match and get away with it, and the FBI was only interested in getting to the Gambinos; what happened to Donaghy was of no concern to them either way.

Lies, damn lies, and statistics

Meanwhile, getting from “intuitively obvious” to “confirmed by evidence” requires a statistical analysis of Donaghy’s calls. The trouble with statistics is that the conclusion you reach tends to depend on the approach you take.

Randall Munroe’s web comic XKCD addressed this issue in amusing fashion. It’s too big to include directly in this article, but you can click HERE to read it yourself. If you don’t feel like doing so, the synopsis is: Scientists are urged to test whether jelly beans cause acne. Using a method with a 5% chance of producing a false positive, they conclude that they do not. However, it’s then suggested that only one color of jelly bean does. After repeating the experiment with 20 different colors, they do find one that appears to cause acne. The joke, of course, is that one in twenty is precisely the rate of false positives they expected. In other words, if you run the same experiment twenty times on different colors, you’d expect to find one that seems to cause acne even if none of them actually do.

There’s a well-known quote from political writer Upton Sinclair: “It is difficult to get a man to understand something when his salary depends on his not understanding it.” He was talking about the difficulty in advancing political discourse, but the principle is as true for the world’s best scientists as it is for arguing about tax policy with people on Twitter.

In the XKCD comic, the experiments are motivated by a pair of journalists looking for a sensational story. Because they hold that motivation, they’re going to push the scientists to keep modifying the experiment until it produces the result they’re hoping for. But it works the other way too; sometimes you’ve got a correct hypothesis, but the first pass at an experiment turns up a false negative or seems inconclusive. If you were in fact hoping for the experiment to turn up negative, there’s a high risk of deciding to stop there. Or, if you do come up with a positive result, continuing to plug away at the problem with different methods might eventually get you what seems to be a refutation.

In other words, if you keep reformulating a problem and testing your hypotheses indefinitely, you’ll probably find support for any number of conclusions along the way. The trouble with that is that humans usually think about something exactly as long and as hard as is necessary to find support for the conclusion they were hoping to reach, and stop precisely there.

In the Donaghy case, reporters for ESPN came up with a very different result from the NBA, largely because their interest – like the journalists in the XKCD comic – was in getting a good story, and it would make a better story if they could find evidence that he had in fact rigged matches.

The NBA relied on reviewing video footage and looking for evidence that Donaghy had overlooked obvious fouls or called bad ones at a higher rate than other refs, but of course he wasn’t that overt. The ESPN team made a different assumption, that he was exploiting technicalities and taking a more literal interpretation of the rules in order to call more fouls on one team than the other in spots that could go either way. Considering only games that were close enough that a few fouls could make or break the spread, they found an imbalance in the calls between one team and the other that would occur by chance less than 5% of the time – 5% being the threshold for “statistical significance” in many branches of science.

Similarly, in poker – especially online poker – most people attempting to collude or runs bots are operating on the assumption that the site is making at least a token attempt to detect them, and are going to avoid “smoking gun” behaviours that are going to show up on cursory analysis. Finding the signal in the noise requires making certain assumptions about the type of funny business going on and picking the right things to look for. The odds of picking the right method depend heavily on who’s looking and, more importantly, on what they’re hoping to find.

Protecting yourself

If this seems to you to paint a pretty bleak picture of the world, you’re not wrong. Whether we’re talking about sports, poker, politics or the stock market, there is always going to be a certain amount of shady business going on that should be easily spotted in principle, but which evades detection by the simple fact that those in a position to spot it don’t have the right incentives to do so.

You won’t always be in a position to spot cheating yourself, but if you assume that there will always be people willing to cheat in any way they can get away with, you can infer a lot about the safety or lack of safety in a situation by thinking about who is providing oversight and what their incentives are. If it’s directly important to them that things actually be conducted fairly, then there’s a good chance that things are on the up-and-up. If it’s only the perception of fairness that matters, it’s a good bet that things are being swept under the rug in order to maintain that illusion.

The good news for poker, as I said before, is that players themselves generally have little or no ability to capitalize on others’ dishonesty, and a clear incentive to stamp out cheating in the games they play. Despite the omnipresent casino cameras and sites’ online security teams, the players are therefore the first and sometimes the last line of defense. If motivations were equal, it would never be the case that internet sleuths operating solely on their own hand histories are the ones to expose collusion and bot rings that security teams with full access to the data have missed, but that has often happened in practice.

Even so, not all operators share the same motivations, and so not every site is as safe as the next. In a vacuum, a player who wins by cheating is no better or worse from a site’s perspective as one who wins by honest means. A site which has identified a cheater may even have an incentive to allow them to go on winning until they attempt to withdraw, so as to seize as large a balance as possible.

The role of government

Regulatory oversight provides additional safety in that sites need to provide the appearance of safety not only to their players but to a regulatory body that has greater access to the data. Once again, though, the motivations of the regulatory body need to be examined, as it’s possible that some are more interested in collecting fees than they are in genuine enforcement. A regulatory body with no higher accountability may simply be in the business of charging for rubber stamps.

Ultimately, and despite its unpopularity, government is the only body higher than players themselves that can be trusted to have the proper motivations. A government might not be interested in the wellbeing of individual players, but it does want to avoid money flowing out of the country and into the pockets of foreign actors. The FBI may not have cared about the results of NBA matches, but it did care about Donaghy’s misdeeds once they resulted in money flowing to the Gambinos.

Likewise, a government which has liberalized online gaming has a vested interest in preventing player funds from getting swept up by criminal rings, especially foreign-operated ones. Its powers may be limited and indirect, by holding accountable its own regulatory bodies which in turn hold the sites accountable. It is, however, better than player oversight alone, and so a site regulated by a local authority accountable to its government is safer than a grey or black market one, and players should therefore have an interest in encouraging their governments to take a positive but active role in online gaming, rather than hoping for a hands-off approach.

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