When Should You Take a Risk?

Because if you want to win, you'll have to eventually.

Great news! Tonight, the San Gabriel Valley Chess Club returns to OTB play, having found a new location in San Marino after our previous venue burned down in the Eaton Fire. I’m very stoked to get back to the board, and quite pleased and encouraged by the fact that our club was able to find a new venue so quickly.

Next week, I’ll report back on how my game goes tonight, hopefully with some Chestnuts of Wisdom™ to share. (And as always, there will be this newsletter’s regular note of suspense: Did Kevin Lose to Another Child?) In the meantime, I want to get warmed up for OTB play by revisiting one of my most interesting positions from last year and trying to draw some lessons from it that I can keep in mind for this week.

A Position.

The date: July 2, 2024. The tournament: Summer Fun 3. The opponent: a teenager rated 1516.

Twenty-six moves into a dynamic, evenly matched game, I found myself in an interesting position. White to move:

What’s happening here? Three things stand out: first, Black weakened his kingside two moves ago with f6. This in and of itself wasn’t a bad move, but as a chess player, this sort of thing should make your Spidey Sense tingle: precise play becomes far more important when you introduce structural vulnerabilities like that, and tactics become possible for the other side of the board.

Second, I have a really nice centralized knight on d5. That’s the reason behind Black’s last move, Nd7: he wants to trade those knights off, eliminating my star player.

And third: Black’s pieces are being held together by duct tape. None of them are defended by a pawn, meaning that they’re vulnerable to being manipulated: I can move his pieces by taking the ones that they are defending. Once again: TACTICS ALERT.

If there’s a single chess concept that stands out most in my brain, it’s probably this idea from Bobby Fischer: “Tactics flow from a superior position.” On the verge of my 27th move, I knew that I had a superior position — and I thought that there must be some sort of tactic there, lying in wait. Moreover, I needed to find that tactic now, because Black has fully signaled his intention to trade off my strong knight if I don’t do something with it soon.

That’s when I noticed Nxf6. If I can force Black to take back with the g-pawn, then Nxf6 blows up his kingside pawn structure. And it turns out that, because of his interwoven pieces, I can force this: by playing Bxd7 first, I eliminate Black’s ability to retake on f6 with a piece. So I’m looking at:

27. Bxd7 Bxd7 (or Rxd7) 28. Nxf6 gxf6.

And then I have an obvious follow-up: Qxh6. I’ve sacrificed a knight but gotten two pawns in exchange, and even more importantly, I’ve exposed Black’s king and shoved my queen in to his position.

But… then what?

For a good ten minutes, I puzzled over this. I stared at the board. I got up and walked around. I thought, and thought, and thought, my brain feeling like it might overheat and explode. The minutes ticked away, and the more I calculated, the more obvious it became: I just couldn’t see what came next. Did I have a knockout punch? Would I be able to find checkmate, or would my attack fizzle out, and I’d merely find myself down a piece for two pawns? I didn’t know.

I could still try to play Nxf6. But I’d be taking a risk.

A Lesson.

Okay, so, first of all: Nxf6 is obviously a good move. Part of why I wanted to write about this position is that I’ve been kicking myself over it for six months: had I followed through on my instincts and played Bxd7, then Nxf6, it would’ve been the best sequence of my OTB career to date.¹

If Black does, in fact, play the continuation that I was considering, then it’s just totally winning for White. The end result of best play for Black is this:

But therein lies the beauty and the pain of chess. A continuation is only as good as your understanding of it. I thought that, by the end of Qxh6, I would be winning, but I wasn’t sure: I couldn’t quite see far enough. I had to trust my intuition and my sense that there would be something on the other side of that move.

Like all sports, chess is objective, but its objectivity is only meaningful in relation to the sophistication of the observer: a position might look clearly winning to Magnus Carlsen and like absolute nonsense to me, just as a passing situation might be a touchdown for Joe Burrow and an interception for Will Levis, or a shot might be smart for Stephen Curry and bad for… I don’t know. Anyone else? That’s what makes a good performer, beyond your sheer ability: how accurately you can read the situation in relation to your own skillset.

At the end of those ten minutes, I just couldn’t pull the trigger, and as so often happens in scenarios like this, I played a “safe,” i.e. bad, move instead: Nxe7, which gives up my best piece for nothing and totally defuses my attack. After that move, the engine has Black slightly better. My mojo gone, I would later make another mistake, and Black would be winning; I managed to complicate the position enough that he offered me a draw on move 44, which I eagerly took.

So, as always, one lesson here is to be better at chess, but there’s another, more subtle lesson that has wider applications: learn when to take risks.

Pursuing the Nxf6 sacrifice was a risk; I knew that. But was it a risk worth taking? And how do you decide such a thing?

When I researched this question, I discovered an article by Brad Stulberg — who is quickly becoming a regular reference-point for this newsletter — that provided a fascinating response to my dilemma. Brad quotes Andreas Wilke, a professor of decision making at Clarkson University, who offers an equation for deciding when to take risks:

“If the chance of success x benefit of success > chance of failure x cost of failure = go for it.”

Interesting. Let’s allow the objectivity of math to make up for the lack of objectivity caused by my badness at chess.

In game, I couldn’t have calculated exactly what the chances of the Nxf6 continuation’s success were, but I probably would’ve pegged them at at least 50 percent. So let’s say, conservatively, 50 percent. The benefits of success, however, would have been huge: I knew that strong chances of checkmate existed, considering that I already had my queen, knight, and dark-squared bishop on the kingside.

Obviously, if my chance of success was 50 percent, that means my chance of failure was also 50 percent — making the last remaining variable the cost of failure, which, as long as it was less than the benefit of success, would mean that I should’ve gone for it.

The cost of failure, in this case — assuming that the calculations I had done were correct, which I felt pretty confident about — was a mere pawn in material, since I would have traded a piece for two pawns. And I would have been able to reasonably assume that it would have been even less than that according to the evaluation, since the damage to Black’s position would have made up for at least some of the material loss.

I’m not a computer, so I can’t precisely calculate the numbers here. But winning positions generally equate to a +3 or above in the eyes of the engine. So even if I had evaluated the situation at, once again, its most conservative values — benefit of success at +3 and cost of failure at -1 — the answer is clear: I should’ve gone for it.

This is actually pretty eye-opening to me. I think you could describe me as a person who is open to risk by temperament: I have quit good jobs to go freelance, changed careers without the guarantee of employment, and, well, I spend a lot of time playing chess.

At the same time, I’m not exactly a daredevil: I don’t partake in any risky behavior health-wise², I have a conventional lifestyle as far as cultural norms go, and I like to be in bed by 9 pm.³

I would say that I’m open to risks in the macro, but in the micro, I tend to play it safe. That’s why I like to play over-the-board chess against people who are better than me, but also why I tend toward safe moves. This is a relatively common problem for chess players, I think; on a recent episode of the Perpetual Chess podcast, Michael Franco was saying basically the same thing, that he struggles to make the move that will blow up the position and cause havoc instead of the solid, strategic choice.

So I would be well-served by keeping this equation in mind: if the benefits of the risk I’m considering seem big enough, and the chances of success are at least reasonable, and it isn’t going to absolutely destroy my own position if it doesn’t work out, then I should take the chance. Particularly if I’m going to beat the good players that I’m, at best, drawing now, this will be essential.

And obviously, it isn’t hard to see how this could apply to any decision. Big upside? Tolerable costs of failure? Reasonable chances of success? Give it a whirl. Sometimes, just having a heuristic you can use helps turn what feels like an unmanageable decision into something you can work with.

I mean, the numbers don’t lie.

READING/WATCHING/LISTENING

READING: I really enjoyed this piece by FM Andy Lee about the tension between the “country club” and the “circus” in chess culture.

WATCHING: Minx. It’s very funny, Jake Johnson is terrific, and we got the whole first season for like, five bucks on Amazon Prime. Looks like it costs the normal amount now, but it’s still worth it.

LISTENING: Combat, Stay Golden. If you like Modern Baseball, you’ll love this album. If you don’t know who Modern Baseball is and you listen to this, you’ll probably think I’m actually 16. Whatever, this album rules.

1. The continuation is, in fact, so brutal that the engine would prefer Black not to take back on d7 — the recommended response is Bxd5, allowing White to go up an exchange after Bxc8 Qxc8. That’s because Nxf6 is so devastating. Even after it’s played, the engine still doesn’t want Black to take back with the g-pawn: it prefers Kh8, which allows Nxd7 Rxd7 Nxe5 Bxe5 Rxd7, putting White up the exchange and two pawns.

2. Aside from maybe playing chess.

3. Before I had a baby, I would often stay up until the wildly late time of 9:45.