Poker Mentality vs. Chess Mentality
How do you semi-bluff in chess?
Richard Rapport caused a stir in the Norway Chess tournament of last year by playing the strange knight retreat 4…Ne7 in a standard position of the Berlin defense.
This move, known as the Mortimer Variation, seems to violate every opening principle: moving a piece that’s already developed, going backwards, blocking in the bishop. What could possibly be the point?
Well there’s a trap if White takes the pawn on e5. That would run into c6 and after the bishop moves Qa5+ picking up the knight on e5.
But Rapport can hardly have expected the trap to have come off, since he was facing the world champion Magnus Carlsen. Oddly enough, Carlsen had even anticipated this very rare line:
“I actually thought about it before the game that he's insane enough to maybe play this move.”
Traps aside, the Nc6-e7-g6 is a common maneuver in the Ruy Lopez, swinging the knight over to the kingside and freeing the c-pawn to advance. It’s just that Black usually prioritizes development and castling before reshuffling the knight.
Here’s where the poker mentality comes in though:
When evaluating the overall value of 4…Ne7, how much should the possibility of White falling into the trap with 5. Nxe5 factor in?
According to traditional chess wisdom you should always assume your opponent will play the best move, so the possibility of Nxe5 would be irrelevant. But in poker, you have to consider the possibility of your opponent making a mistake. In fact, inducing mistakes is a big part of the game. In general, chess players would analyze the line arising from the best moves. Poker players would consider the average of possible outcomes, weighted by their probabilities.
It turns out that a lot of people fall into the trap!
I broke out the stats by rating tier. Below the 2000, 5. Nxe5 is the most common move. It stays pretty common even through the 2200s. You have to go up to the 2500+ range before most players see through the ruse and Nxe5 becomes a rarity.
The thing that makes me really interested in 4…Ne7 though is that even if they don’t fall for the trap, it’s not that bad. The knight retreat is actually one of the computer’s top choices. Objectively, it’s not a bad move. If White doesn’t fall into the trap, the game continues more or less as normal.
In this sense this variation is like a semi-bluff in poker. A semi-bluff is when you bet to try to make your opponent fold, but you still have a chance to win even if you get called. For example, you go all-in with a flush draw: you’re hoping they fold and let you scoop the whole pot, but even if they call, you could still make a flush and win.
When you do the math on this kind of situation, it tends to work out so that your opponent doesn’t have to fold all that often for the whole operation to be profitable. The key insight is that if you can put yourself in a situation with two possibilities, a really good one and a not so bad one, that’s usually a great spot overall. The value of the really good scenario, as long as it’s reasonably likely, carries the day.
This is what makes 4…Ne7 in the Berlin more attractive than a lot of other opening traps. For example, consider this trap in the Englund Gambit, popularized by Eric Rosen.
The problem with this one is that if White plays almost any move besides 5. dxe7, they’re not only better, but winning outright. This completely changes the calculus on the trap, since now your options aren’t really good vs. not so bad, they’re really good vs. disaster. Traps with this kind of downside are good for a laugh but risky for regular play.
This is a great post using data to rank opening traps, combining probability (how likely they are to fall into the trap) with potency (how devastating it is if they do). But it leaves out the downside: how bad is it if they don’t fall into the trap?
Going back to the Mortimer Variation with Ne7, I decided to create a whole repertoire around this line, continuing in the spirit of tricky sidelines that score well in practice.
Against the mainline with 4. 0-0 and 5. d4, I went with a line I knew because chess coach Dan Heisman mentioned that he taught it to radio personality Howard Stern. This line features wild sacrifices and attacks, but it doesn’t have as attractive a risk profile as the Mortimer trap. If White knows exactly how to react, they end up with a much better endgame. However, to get there, they need to play a long string of precise and counterintuitive moves. According to Heisman, in many online games, Stern never ran into the refutation.
One reason that the tradition in chess is to assume that your opponent will play the best move could be that for most of chess’s history, you would have had a relatively small number of playing partners, each of whom you’d play many games with. If a trap works against someone, presumably it works only once. Only recently has the standard become to play anonymous games online with a different opponent every game. Nonetheless, that’s the world we’re living in, and more and more players are interested in what works, best moves be damned.
Of coure bullet/blitz is fun and has its place, but advice geared to that time control could and should be distinguished from long term improvement in classical chess.
In classical the same probabilistic attidude could be usefull and Smerdon also uses it to defend his choice for the Portuguese Scandinavian. Using lichess blitz/bullet games to make a repertoire would entail some risk that you hate your(self) choice for hours while having the worse position.
Lastly, I don't particularly like using dubious ploys trying to exploit holes in ones opening knowledge. It is part of the game, may be a good winning strategy and tricking someone tactically is fine, but to consciously search for that from the get go is not for me. In aggregate it incentivices more choirs for eveybody: learning replys to dubious lines that are seldom played
The major distinction is blitz chess. Implicit in most advice is that you want to improve on the long term for classical time control. You don't want to travel to a game, hope for some trick, let your opponent think for 10 minutes and have a slightly worse positon than you could have had for hours. Even if you win immediately you're not that happy: You produced or learned nothing for that game (maybe even had the same position in some blitz games) and wait till your friends are done with theirs.
I think 3 + 0, which you usually play, is more like bullet than blitz and teaches you significantly less about chess in general and thus classical chess. It is more prone for the tricks you describe. I've written a short essay on the matter, if you're interested I'll sent it to you for consideration in your next post.