What is Chess? What to think and how long for

Your knight for my pawn, and you have to take it. Things are about to get worse, sunshine. This is my game now. You have been out-calculated. I don't want the piece, I want your soul. My attack hurts like a school disco and I'm going to sac sac mate until... wait. What? 10 seconds left. But, I'm winning, right? I must be like +5 right now. Just play any move. But how do you have 10 minutes on your clock? Anyway, just play any move. That one will do, fine. For. Fuck's. Sake. Any move but that move! How could I instantly find the one move that loses? Literally, I could have played anything el...

At first, chess seems to be about the moves you play. But, after a while, you'll notice something weird. You can know more, think deeper, and play better moves, and still lose. That's because better chess is not about playing better moves. It's about playing good enough moves, quickly enough.

Playing moves faster means you play worse moves. If you have to play a game with only 3 minutes on your clock, you'll play about as well as someone 700 points lower rated than you would in a 90 minute game. Put another way, the 3 minute version of you beats the 90 minute version of you roughly 2 in 300 games (that’s what 700 points mean). Ben Johnson and Ken Regan go in to detail here.

It's hard to think of anything that can affect your chess strength this much. How many openings have you learned? How many strategies have you formulated? How many tactics have you solved? How many endgames have you studied? And yet, for better chess, the most valuable skill might be how you use your time.

This feels cruel. You can improve your tactics, openings, and everything else, but if you don’t combine them in the right way, you’ll still lose. You’re like a mathematician who never learned BODMAS. You might perfectly add, subtract, multiply, and divide, but if you do it in the wrong order you get the wrong answer.

Unfortunately, after two years of maniacal chess playing, I do not have this skill. How could I? When I play chess, I just play. I have no idea what I should be doing, let alone how long I should be doing it for. And, somehow, among the endless drift of chess content, it seems few have much to say about to actually practice using the right amount of time on the right things. Only that you should.

It’s not clear what playing chess even means. Only I’m pretty sure it’s not what I’m doing: think an arbitrary amount of time until there's a move I reckon is best, or least bad, then play it. This means I run out of time, all the time. I checked Aimchess and I'm behind on the clock 90% of the time. I get sucked in to all the possibilities, especially when any move will do. The more I've learned about chess the more there is to think about and the slower I’ve got. I win the game the hard way, through calculation and planning. I then lose the game the easy way, because I'm watching the clock more than the board. Clearly, this is wrong.

Other people play chess differently, but equally badly. I run an informal chess club, the Tuesday Chess Club (which ironically runs on Monday or Wednesday but never Tuesday). And Shanjit, a quiet guy who speaks only to tell you you're mated in 5, is top dog. When Shanjit beats you, he smashes you. We play with a 10 second increment, and Shanjit finishes every game with more time left than he started with. But he loses that way, too. Win or lose, it takes Shanjit no time at all. Maybe Shanjit doesn't actually like chess much, so when he's playing chess he tries to avoid it. I dread the day he decides to think for 11 seconds a move.

So how long should you spend per move? And what should you be thinking about?

How long to spend per move

The two triggers

A lot of what I write here is a summary or extrapolation of Dan Heisman's A Guide To Chess Improvement. It's really phenomenal, so go buy it and read the whole thing.

You might say it's not possible to say precisely how long you should spend per chess move. Some moves are harder than others. You'd of course be right. But I suspect you're being pedantic, and we can get a practical answer and improve our game.

If you are optimising for time alone, you should move as quickly as possible. But it is possible to spend too little time on a move. If you miss a win or blunder because you moved faster than necessary, that's a mistake. On the other hand, if you aren't optimising for time at all, you should think almost infinitely. But it is possible to spend too long on a move. If you miss a win or blunder later because you now have insufficient time, that's a mistake.

So we need one principle to stop us moving too fast, and one principle to stop us moving too slow. Dan calls these 'move triggers'. You should move when:

Trigger 1 (if you move too fast): You have proven which move is best, so no further time is necessary. You would be willing to stand up in front of a room full of grandmasters and argue that it would not be possible to find a better move; or

Trigger 2 (if you move too slow): You have not proven which move is best, but time constraints (good time management) make it efficient for you to play the best move you have found so far. There is a time 't' which is the reasonable time to think for that move.

The first trigger is straightforward. You move as fast as possible when you're sure it's opening prep, or when you're sure of the best move. If, like Shanjit, you're used to playing too fast, you probably move before Trigger 1. And, you might wonder 'what do I do with the rest of my time?'. We'll get to that.

The second trigger is more complicated. If, like me, you're used to playing too slow, you might wonder about the mistakes you’ll make by playing a worse move because you've hit Trigger 2. That's the trade off you make to avoid losing in time trouble. Smaller mistakes now instead of bigger mistakes later. But if reasonable time t is our trigger, how long is t?

How long is t?

To answer this, we first need to understand three principles: average moves per game; Botvinnik's Rule; and critical moves. At times some of the numbers below might sound fiddly, but don't worry - you don't need to remember them. The guidelines they generate will be straightforward.

Average moves per game

There are roughly 40 moves in the average chess game (source here). If you use all your time in 40 moves, that's an average of ~2.5% of your time every move. If your average move time is much faster than that, you're often going to leave time - and chess skill - on the table. If your average move time is much slower than that, you're often going to run out of time. It's crude, but our first estimate for t is just 2.5% of your total time. Assuming you play some moves faster because of Trigger 1, that should give you a little buffer. But we can do better.

Botvinnik's Rule

Botvinnik was a chess world champion, and his time management rule was this: In slow games, use about 20% of your time for the first 15 moves. In fast games, use less than 20% of your time for the first 15 moves (Dan has more here).

This rule is fairly intuitive. Games are not usually decided in the first 15 moves, 'the opening', and it's the most predictable and familiar part of the game so preparation and principles can speed up your play. I personally find this one tricky to stick to, though, and there are some situations where this can be ignored. For example, where you are in book for a long time or out of it very early, or where the position is highly critical move after move.

But Botvinnik knows what he’s talking about, so let’s take it as a good aim. 20% of your time divided by 15 moves is ~1.3% of your time per move. So now we have a better estimate, t is ~1.3% of your total time for the first 15 moves, and ~2.5% of your time after that. Our buffer is quite large now: by moving faster in the opening we saved time compared with if we'd just moved at our original 2.5% speed the whole time. We save ~1.1% of our time per move for 15 moves, which is ~17% of our total time. Add that to the time saved by moving faster than Trigger 1, and we've got a lot of buffer.

Critical moves

Some moves are more important than others. Some positions have many easy to find moves which are playable. These are not critical. Some positions have one complex move which can win or lose the game. These are critical positions. Obviously we want to spend more time on moves in critical positions than non critical positions.

Identifying critical positions is an important skill. We have some tactical heuristics:

  • When the best move may be clearly better than the second best move.

  • When you have a tactical (short-lived) advantage which will disappear if not exploited now.

  • When you see a move which seems to win.

And some strategic heuristics:

  • When the game changes from known theory into unknown territory, from opening to middlegame, or from middlegame to endgame.

  • When any pieces are exchanged, especially queens.

  • When there is any change, or possible change in the pawn structure – especially in the center.

In some cases, it may be warranted to use almost all your remaining time on a critical move - for example when deciding the almost certainly final critical move in a concrete endgame position. But, most games will have several critical moves. One easy to remember suggestion here is just to double t on critical moves compared with non critical moves. If t for non critical moves is ~2.5%, on critical moves t is 5% of your total time. Given the buffer created by our faster opening play, and Trigger 1, we have time for roughly 7 or 8 critical moves per game. This seems about right based on our heuristics.

Just give me the answer

With these three principles, we have a practical guide to time management. And, surprisingly, the numbers work out quite neatly. Roughly speaking, if the time control is 10 minutes, we use ~10 seconds per move for the first 15 moves, double it for ~20 seconds per move for the remaining non critical moves, and double that for ~40 seconds per move for the critical moves. If the time control is 5 minutes, it’s 5, 10, and 20 seconds per move. And so on.

Obviously these numbers are just guidelines, and should be flexed somewhat either way based on your intuition in the moment. But if you're going far longer or far shorter, you should be questioning if that's the right decision.

Here's a table for popular time controls:

If you want to change the parameters, for example to change the time control, length of or time spent in the opening phase, average move per game, etc you can copy and edit the sheet here.


What to think about and how to practice

So, if t is the reasonable time to spend on a move... what should we actually be thinking about for t? And how should you practice thinking about whatever that is, but for no shorter or longer than t?

I know I play too slow, but I didn’t know how to fix it. I just play, slow. Like Shanjit plays, fast. You’d think me playing more faster games and Shanjit playing more slower games would be the solution. It isn’t - I’m living proof that bullet chess and time trouble are simultaneous addictions. But, I’ve found a way to practice and improve. And it starts with knowing what to think about.

Poor time management is a symptom of not knowing what to think about. If you move too fast, you aren't thinking all the thoughts you need to play your best. If you move too slow, you're thinking thoughts you don't need to play your best. Practicing time management means practicing what to think about.

When I'm making a move, my thoughts often wander. I go down a chess rabbit hole and emerge in time trouble. Sometimes it's worth it, usually it's not. On the other hand, Shanjit doesn't wander anywhere - he plays the move he thinks is best before he's sure that it is. The answer to playing faster, for me, and the answer to playing slower, for Shanjit, is the same: be clear about what exactly is worth thinking about. Whether you're too fast or too slow, you need to practice your thought patterns.

So what should you think about? We don't want to think any unnecessary thoughts, but we also don't want to miss anything. Our thoughts should be both necessary and sufficient.

The necessary and sufficient thought pattern and acronym I use is TASOTISM, which knits together thoughts from Dan’s book. I often count TASO TISM on my fingers when I play (though sadly I am not good enough at chess for people to think I am cheating with mysterious hand signals). Every move, TASOTISM is necessary and sufficient to play your best. Here’s what it means:

Time

When is my Trigger 2? Is the move critical?

You can check this while waiting for your opponent to move. You know how long t is for your time control, so you can deduct that from your current time left on the clock to know your Trigger 2 time. If you think the move is critical at any point during your thought process, you deduct t again to double your time that move.

All

What are all the things my opponent’s move does?

It's easy to miss a hanging pawn, or that a bishop has become active where it wasn't before. Often beginners miss their opponent's plans, and this step forces us to consider them.

Safe

Is my opponent’s move safe? Is there a tactic?

A move is safe it doesn’t lose material (eg a pawn), concede something horrible positionally (a powerful outpost) or lead to being mated. There are a variety of ways to check this: checks, captures, threats; the burger method; safety tables; etc. This is just one step of the process of making a good move. That's why complicated tactics aren't that practically important, and why your intuitive sense of whether a tactic is present is. If, in a 30 + 20 game, t for non critical moves is ¬60 seconds, the kind of tactics you could sense within a useful amount of time might be tactics you could sense within 20 seconds and solve within 60 extra seconds. In 5 + 3 blitz, where t is ~10 seconds, if you can't sense a tactic in 5 seconds and solve within 10 extra seconds, it's not worth the time it takes to look for.

Objective

What is my objective this move?

Maybe you need to complete development, apply pressure to some weak pawns, to start an attack or to play prophylactically. This step forces us to consider what we know strategically, so that we don't prioritise the wrong things, like grabbing a pawn while our king is weak or starting an attack while we’re underdeveloped.

Time

How much time do I have left before Trigger 2? Is it critical?

Halfway done, let's check how much time we have left this move.

Initial

What are all the initial candidate moves that might accomplish one or all of my wants?

This stops me wandering around random moves, that don't align with our wants. And it forces Shanjit to consider more moves than one.

Safe

Which of the initial candidate moves can I reject because they are not safe?

We apply our tactical radar now to our own moves. Importantly, this doesn’t mean trying to guess how the game might go several moves into the future. It means checking our candidate moves are safe.

Move!

Of the remaining candidate moves, which do I think is best?

You have thought all the thoughts you need to think to make a good move in a reasonable time, so now it's time to pick one.

That’s TASOTISM.

If you’re used to letting your mind wander, it is brutally utilitarian. But that’s a good thing - thinking is your enemy, and now you may only think to serve the greater good. If you dont usually think much, the extra thoughts can seem superfluous, especially if you're used to playing fast time controls. If that's the case, ask yourself, which step should you skip? The answer, I think, is none.

This thought pattern is not a prescription - it is a description of the minimum and maximum thoughts you must have to play your best chess. I suspect that's why lots of strong players recommend beginners play slow time controls: strong players can complete the necessary and sufficient thoughts very fast and mostly subconsciously. For weaker players, you can't play fast without skipping a step. Hoping you’ll develop better thought patterns by playing faster time controls is is like hoping you’ll learn Flight of the Bumblebee by playing along.


Lion's bones

It can seem strange to put such a rigorous structure on our thinking in a game like chess. Are we not caging our inner chess lion? No masters are sitting there counting TASOTISM on their claws! I think this is misguided. A rigorous thought pattern is not a cage, it is an x ray. It shows our lion's bones. Strong players don't notice their thought patterns - you only notice bones when they're broken. The goal is not to think TASOTISM in t, the goal is for good thought patterns to become so internalised that they form the skeleton our muscles hang on. Bones force our chess lion's muscles to work harmoniously - bones make it impossible for muscles to work any other way.

Good thought patterns allow you to figure out which muscles are weak. Do you lack quick tactical sense or calculation? Are you unable to formulate strategies or come up with candidate moves? Are you spending too long in the opening? Are you indecisive? It is impossible to answer these questions without a good thought pattern. If you calculate too long, you might mistakenly feel better in calculation than your peers. If you calculate too little, you might mistakenly feel worse. If you are playing endgames with no time on your clock, you endgames might appear bad because your openings are too slow. When your time and thoughts aren’t assembled properly, it’s impossible to build muscle in the right places.

If you complete a good thought pattern in a reasonable time and still lose, it's time to build some muscle. Until then, it’s not tactics or openings or endgames you need most. It’s time with your thoughts.


2023 update

I moved the A before the S, so TASOTISM instead of TSAWTISP.

I recommend playing long 0 increment games to practice time management. Online this would be 30 + 0. There’s no way to come back after mismanaging your time with no increment. This makes them a good forcing mechanism for practicing time management.

It’s worse to run out of time than it is to play slightly ahead of time, and some games run long, so I recommend extending the move count we plan for to 45, so that we have some buffer if it looks like we’ll go to a long endgame.

Nate Solon recommends thinking in terms of time per periods or time checkpoints, rather than moves. For example, where should I be on the clock at move 15? This is useful. If we’re planning for 45 moves, the most logical periods are 15 (opening), 16-25 (middle game), 25-35 transition or mate), 35+ (endgame). So 25, 25, 35 make good time checkpoints. Including these checkpoints and planning for 45 moves makes the time table look like this:


If you liked this you might like my first article about how to improve at chess here.

I thank Dan Heisman, Ben Johnson, Ken Regan, and Nate Solon for providing the ideas and inspiration for this post.

Why Your Chess Openings Suck

Why Your Chess Openings Suck

The truth is, my openings still suck. As, I’m sure, do yours. But it’s not our fault. It is, I concluded conveniently, our repertoires’ fault. They are not built for humans, because they are not practical. Here’s how to fix things.

Read More

Calculating About Magic: My Startup Failure

Once I’d officially failed, my friend, colleague, and investor Alice asked me, ‘are you happy you did it?’.

A hard question.

I failed last year. Closed down the startup I started. An adulting, unarguable, and public failure. I let people down - my investors and team members were mostly friends. Thinking about it now I feel ashamed and a bit sick. This isn’t a complaint. I’ve had plenty of good luck in life, some earned though lots not, and I deserved what I got this time.

So, am I happy I did it? That seems easy. No, I’m not happy I did it. Starting a startup was a painful way to lose my time, and (mainly) other people’s money.

But it’s not that simple, even if it wasn’t worth it.

Calculating About Magic

Magical products are curious experiences. When you try to explain why they’re magical, it’s often impossible. Yet once you experience their magic, it’s self evident.

Why is TikTok spellbinding? You can (in retrospect) tell a story about short, vertical, music accompanied videos, displayed in an order you’ll like. But no analysis beats the experience. If it did, TikTok would have been obvious without hindsight.

People have written books on trying to reason your way into magical products. There are Customer Developers with Minimal Viable Products, Hackers Doing Things That Don’t Scale, Zero to Oners Keeping Secrets. But there is something difficult and pointless about this reasoning. If it worked so well, magic wouldn’t be so rare. Following a method is a safety blanket, but the method is not the point, magic is.

And magic is easy to see but hard to prove. You may know 1+1=2, but you almost certainly can’t prove it. Probably, without help, if you spent your life on it you couldn't. This is startup complexity theory - there are answers that can be computed easily (P) and answers that can be checked easily (NP). And, you don’t need to. It’s up to writers and artists and academics to prove things we already know are true. Not the founders of TikTok.

Concentrating on the startup method can, sometimes, mean failure. You can’t prove future magic with tests and interviews. And if you try to prove things that don’t need proving, you’ll go slower than competitors that build the right thing the wrong way. The riskiest strategies win. It’s no use trying to explain TikTok when you can build it and show it. Reasoning ends when magic begins.

For smart people, this is hard to accept. In school, getting the right answer isn’t the point - you need to show your calculations. But in startups, it’s enough to make something magical, and not enough to do what you’re supposed to. No one really cares how you win, as long as you do. And if you fail, no one cares how you did that either. There is no right way, beyond winning and losing. No marks for calculations.

If writing about music is like dancing about architecture, eventually, talking about startups is like calculating about magic.

A Gambler’s Logic

When what matters is the magic, not the method, you must seize your chance.

So, 2020. My friend and I built a site that within a month had hundreds of thousands of people watching social esports content millions of times, with zero marketing. Magic. We'd made the site in a weekend and the esports competition was over soon - there were no calculations to be found, and there was no time to do them. We had to decide, do we go all in? Our answer, of course.

This opportunism is a common startup story. Take Entrepreneur First company Souschef, which made automation robotics for F&B retail, and started about as well as you could hope for. Within the first few weeks of EF, they had a sales deal worth $1million, ready to sign. Magic. Did they sign it? Of course.

Today, both my startup and Souschef are dead. We didn't capitalise on our magic, for reasons good and bad. Without magic you can’t do anything, but magic doesn’t tell you what to do next.

Judging by our results, following the magic and starting a startup was the wrong decision. Should we be happy we did it? No. But results based analysis doesn't quite make sense for startups.

In a competitive world, you win by making correct decisions before it's reasonable to. To stand a chance of winning, you have to take risk. By the time you can accurately assess the risk, it’s too late. Someone else skipped that step. No marks for calculations.

Taking risk means sometimes things don't go your way. That doesn't mean, given the situation, you shouldn't have placed the bet. In poker, the starting hand with the highest win probability is two aces. But, on a six person table, even with two aces, your odds of having the winning hand in the end is still less than 50%. Were you wrong to bet on aces because you lost on the final card? Of course not. It's what you're there to do.

Starting with magic is like starting with two aces. Most startups die young, because they can’t find any magic at all. If you find some magic, you’ll never have a better shot. If you might want to start something someday, you absolutely must take your chance. Even if you don’t succeed, it's not wrong to gamble.

Prison Breaks

So once you’ve failed, should you regret it? I think not. Starting a startup is like attempting a prison break with a life sentence: it probably won't work out, but you gotta take your chances.*

I’m still locked up. And while I’m here, I often think about all the people at EF who didn’t succeed either - by definition almost everyone. I can’t help but feel a little guilty. Have I encouraged thousands of people to do something they regret? Do they feel a little sick about it too? I'm sure some do.

But I hope, for most, they see EF for what it’s meant to be. A chance I’m grateful I had. An opportunity, that might never come again, to start with aces, to prison break. And that’s magic.


*Incidentally, I think this is the best argument for people who have no opportunities to make millions to still play the lottery, even though the expected value is negative.