If you read much poker news, you’ve probably heard about Patrick Leonard’s controversial river behaviour in the PokerStars Championship Barcelona Main Event. That was three weeks ago now, and three weeks is an eternity in the media world, so I realize that I’m extremely late to this party. I wanted to write something thoughtful on the subject, however, not just my own gut reaction; I try to limit my hot takes to Twitter and collect my thoughts more carefully before writing up an article.

The controversial move

If you haven’t been following the story to this point, you should probably just watch the video of the hand yourself and see what you think.

The short version is that, after checking to his opponent Andrea Shehadeh on the river, Leonard found himself facing a bet and took a long time to think about it. After several minutes of mulling the decision and saying things like “it’s hard to be bluffing, I guess,” Leonard put down his chips and said “alright, you’ve got it, nice hand.”

He didn’t, however, fold his hand. Rather, he waited for a reaction from Shehadeh, then said, “oh, that’s strong,” waited another couple of seconds, and finally folded at that point. The contentious issue is that “you’ve got it, nice hand” is not a binding statement in most cardrooms, but tends to be something that players say in the process of folding, rather than while they’re still making up their mind to do so.

When instinct contradicts consensus

Part of what’s taken me so long is that my own gut reaction was the opposite of most people’s. Virtually everyone seems to be in agreement that Leonard’s actions constituted an angle, even if he didn’t think so at the time, or that they’re sufficiently angle-like that he should have known better. Although he initially defended himself in a video, explaining his actions and the motivation behind them, Leonard himself has since come around and acknowledged that he was over the line and apologized.

My instincts, on the other hand, were that it wasn’t something I’d do myself, but that I’d have no objection to if done to me by an opponent. It seemed to me that Leonard was implying that he was about to take an action in order to get a read. The immediate comparison that popped to mind is the way many beginners will reach for chips out of turn to imply that they’re likely to bet; not only does no one object to this, it’s actually a reliable tell for a lot of people, as if it’s done deliberately, it usually means they’re hoping to induce a check.

When essentially everyone disagrees with you, you have to consider the possibility you might be wrong. At the same time, it’s dangerous to assume you are, unless you can figure out why. Herd mentality is a real problem, and it’s easy to find examples of bad ideas propagating simply because they were held by sufficiently many people that they came to be seen as self-evident.

The trouble is, although everyone but me seemed to feel that “you’ve got it, nice hand” was an angle, they couldn’t quite explain why, at least not to my satisfaction.

No chance of a muck

The most obvious way this could be an angle is if Leonard was hoping to trick Shehadeh into mucking his cards, thinking he’d won the pot, then announce “call” once those cards were irretrievable. The trouble is that, given the players involved and the context, there’s no way this would work and no way Leonard could conceivably think it would work.

It’s the sort of thing that’s done in nightly tournaments by unscrupulous regulars to clueless newbies, but it’s just not the sort of thing that one professional would attempt to pull on another at a crucial stage of a main event, when everyone is being sure to protect their hand. What’s more, since they were playing at a feature table on camera, the downside to Leonard’s public image if such a ploy succeeded would outweigh the immediate financial benefit.

So, if that were all there were to it, I’d say that if there’s no harmful intent, no harmful outcome, no real likelihood that a harmful outcome could have come about, and no explicit rules violation, it’s hard to make the case that the behaviour shouldn’t be acceptable.

Bad definitions

Part of the reason that I didn’t see Leonard’s play as an angle-shoot originally is that I already had my own clear definition in mind, which happens to be wrong.

Most classical angle shoots involve the player knowingly doing something which binds them to an action, then pretending that it was not their intended action. For instance, throwing in a 5000 and 100 chip, then announcing 600 a couple of seconds later; because the chips were thrown in before the bet was vocalized, it should stand at 5100, yet the player can pretend that they simply confused their 5000 for a 500. The opponent may then make a different decision than they would have if the large overbet had been clearly intentional.

Based on that, my feeling about angle-shooting was that it necessarily involved bringing the dealer and/or the floor staff into the game in a way not intended by the spirit of the rules. By that definition, it would be similar to, and unethical for the same reasons as the practice of “diving” in soccer and other sports.

That definition misses out on some other clearly unethical practices, however. For instance, a player could protect their cards with a high-denomination chip, then slide the rest of their stack in; to an inattentive opponent, it might look like an all-in, even if the dealer doesn’t mistakenly announce it as such. Misleadingly stacking one’s chips under guise of sloppiness, or intentionally over- or under-estimating when asked “about how much” one has would fit the category as well.

My incorrect definition isn’t the only one out there by a long shot, either. Another one I’ve heard from several people is “exploiting knowledge of rules technicalities not familiar to the average player.” Although this would arguably cover a scenario like Leonard’s – if one assumes his opponent may not know that “you’ve got it, nice hand” isn’t a binding fold – it misses some of the others I’ve mentioned, like misleading chip stacking. Furthermore, it suggests that careful study of the rules is unscrupulous, when it should be every player’s responsibility to make sure they know them. Finally, it also fails to provide any concrete guidelines to establish which rules are “technicalities” and which ones should be part of a player’s game plan.

Is angle-shooting like pornography?

Pressed hard enough about what angle-shooting really means, most people will fall back on a statement like, “It’s anything you do at the table that’s unethical but not expressly against the rules.” Or, as Steve Ruddock put it to me on Twitter, “anything that would get you thrown out of a home game or beaten up by Mike Dentale.”

These lines of reasoning resemble the words of United States Supreme Court Justice Potter Stewart in the 1964 case of Jacobellis v. Ohio, where the question of censorable obscenity vs. protect speech was at issue. Regarding the phrase “hard-core pornography,” Stewart stated that he would “not today attempt further to define the kinds of material I understand to be embraced within that shorthand description […] but I know it when I see it, and the motion picture involved in this case is not that.”

Stewart was praised for his common-sense approach, so is a similar “I know it when I see it” philosophy about angle-shooting good enough? I would argue it isn’t, because it leads to circularity in a way that Jacobellis v. Ohio doesn’t. To see that, you have to lay out the argument in full.

For Jacobellis v. Ohio, it’s: “If the movie is hard-core pornography, then it qualifies as obscenity and is not protected speech. However, I know hard-core pornography when I see it, and this isn’t that.”

When it comes to angle-shooting, however, the whole point is to discourage unethical behaviour. If the only definition of the term hinges on the ethics of the acts in question, the statement as a whole becomes a tautology: “It is unethical to do things which are unethical.”

So what is it?!

This would be a pointlessly long article with no payoff if I didn’t think I’d managed to pin down a more concrete definition than that. Fear not, dear reader, I won’t let you down.

The conclusion I’ve reached after a lot of thought is that to understand angle-shooting, we have to look at the game’s information structure, and specifically what constitutes a “game state.” This concept is critical to the programming of artificial intelligence, which is useful because it means it’s pinned down very precisely: Computers follow objective rules, and semantic arguments about the nature of neural nets notwithstanding, they don’t just “know things when they see them.”

The game state for poker consists of the player’s own cards, the distribution of chips between all players’ stacks and the pot, and all actions taken throughout the course of the game. Most contemporary poker AI being non-adaptive, that last is often truncated to merely the actions of the current hand, but properly speaking it includes all hands the player has ever played which involved any of the other players currently at the table, plus all cards that have been shown down in those hands.

In more intuitive terms, the game state is the hand history. It’s all the information that comes in the hand history file you would download from an online poker site, and which you would post in a forums thread to do a hand review.

A simple and complete definition of angle-shooting, then, is that it is an attempt to deceive another player about the current game state.

What’s critical to note, here, is that other players’ cards in the current hand are never part of the game state. The player’s (or AI’s) decision-making may include making inferences about what the other might be holding, but it’s not supposed to be available explicitly. Attempts to interfere with opponents’ efforts to infer one’s holdings are therefore “part of the game,” while attempts to deceive when it comes to actions taken, chip stacks, or even cards shown down in earlier hands are not.

A caveat about intent

There’s one little wrinkle in this definition, in that we have to include in the game state whether an action was intentional or not. This isn’t something that computers need to worry about; they simply assume that all actions were intentional. If we assume differently, however, then every action actually has two components, its practical component and its informational component, though usually these are the same.

The only way players have to infer their opponents’ holdings is by way of the actions they’ve taken voluntarily. In terms of their practical components, the small and big blind are the same as any other raise; informationally, they’re very different, as posting blinds is an obligation not a choice, and therefore implies nothing about the player’s holdings.

Looking at the example of betting with a 5000 and 100 chip, to understand it as angle-shooting, we need to accept that doing this as a genuine accident (having meant to bet 500+100) is a different play – from a game state perspective – than doing it intentionally. When it really is an accident, the informational component is identical to the bet of 600, and only the practical component is a bet of 5100. Put another way, it’s as if two bets were made, a voluntary bet of 600, followed by a forced bet of 4500, akin to a posted blind.

The opponent will naturally respond to these two game states differently, so an attempt to make them believe it’s one when it is actually the other constitutes shooting an angle. Understood correctly, it’s not very different from deceiving them as to how many chips you have, or making them think you’re out of the hand when you’re not.

But what about Patrick Leonard?

Unfortunately, this doesn’t tell us whether Leonard was or wasn’t angle-shooting when he said “you’ve got it, nice hand.” There’s still one final hair to split to make that call, which is whether he was attempting to make Shehadeh think he was about to fold, or that he had folded. What a player has done is part of the game state, but what they’re about to do is not, unless they’ve in some way bound themselves to an action, for instance by checking out of turn.

In defending himself, Leonard said that he was not attempting to get Shehadeh to muck, nor was he genuinely unsure about what he would do. Rather, the time he took thinking and everything he did prior to actually folding was intended to gauge Shehadeh’s behaviour in a high-pressure situation, in the hopes that it would be useful in a future hand.

That doesn’t make it not an angle, however, because how a player behaves depends on the game state. People will give away much more when they believe a hand is over than when there’s still action pending. Misleading someone about the game state to get a read isn’t any different than misleading them to change the outcome of the hand.

Having heard Leonard’s explanation, I don’t think that in his mind, at the time, he believed that he would fool Shehadeh into thinking he had folded, or had verbally bound himself to a fold by the rules. Rather, it was just part of his efforts to pretend he was facing a very difficult decision and was reluctant to fold. However, many others have indicated that they might have been misled by his statement, and that it would have been reasonable to expect Shehadeh might be, even if he would never release his cards.

In other words, it may not have been an intentional shooting of an angle in this case, but a different player behaving identically in a different context could very easily be angle-shooting deliberately. That being the case, there’s another rule of thumb to consider, one which applies as much to real life as to poker: If you’ve done something outwardly indistinguishable from an unethical act, then regardless of what you intended, you should apologize and try not to do it again.

Which is, fortunately, what Leonard has since done.

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