Chess, Starcraft II, Aikido, and the Teaching Process

This is part 2 of a series ruminating on the process of learning to play Starcraft II at age 58, and what it might tell us about learning and teaching chess (with a little aikido thrown in for good measure).

"But teacher, you said--"

When I'm teaching chess to relative beginners, the following position causes trouble:

There are general principles one would like to teach.  Develop your pieces.  Meet enemy threats and make threats of your own.  Knights before bishops.  Control the center.  All of these principles converge on the conclusion that Black's best move is 3...Nf6.  But when your student tries that out, this is likely to happen:

As a young local player said when I asked him how his game had gone, "I fried his liver."  Among beginners this attack will win much more often than not.  In the 1980's, as a young player myself, I played two identical tournament games that ended in mate on move 13. 

When this happens to your student, they are likely to ask--plaintively or belligerently according to temperament--what went wrong.  You can recommend the master line to them, which goes like this:

to which they will quite naturally respond that 5...Na5 loses a pawn, decentralizes the knight, and generally looks terrible.  It isn't terrible at all--in fact it's why I stopped playing the Fried Liver, after a nasty trouncing from an NM--but the reasons why are hard for beginners.  Or, you could say that while 3...Nf6 is not technically a mistake, they'd better not play it just yet, despite all the reasons you just taught them why it should be a good move.  At which point you realize that, while chess is hard, teaching chess is also hard....

I learned my first StarCraft opening from a brief YouTube video in which the pro player extolled its simplicity, reliability and power.  Then he gave a slight, nervous giggle and said, "Uh, though we'll hope they don't attack too early.  Just do your best if that happens."  Of course it did happen, and I lost a lot of games to those early attacks because I lacked the technical skills to fend them off.  The whole experience seemed awfully familiar, and eventually I thought of the Fried Liver Attack.  Early attacks in StarCraft have much the same quality.  They are very concrete:  general principles won't save you when your opponent is at your throat in the opening.  And while they can be countered, it is hard to teach the needed technique to a beginner. 

In StarCraft even the top players fairly often lose to instant attacks.  One top Taiwanese pro, Has, has made a career out of them, and bags World Champions from time to time.  This doesn't really happen in chess.  A key difference between the games is that SC2 has hidden movement.  If Has' opponents knew what he was going to do they could surely deal with it; but he's spent years coming up with lots of different instant attacks, so his opponents are left guessing.  The apex of this strategy was a game in which Has did nothing untoward whatsoever, just quietly developed, and his opponent wasted enormous effort and expense frantically trying to find the non-existent threat.

You could compare these to tournament novelties in chess, but my impression is that with very rare exceptions, a good TN will give you an edge in chess, but won't win the game outright.  In SC2 the game can literally be over, even at the top level.

"I won't let that happen again!"

As a titled player I get a lot of requests to play lower-rated opponents, and sometimes do, especially in Daily Chess.  It means I get to see the following opening fairly often:

Chess.com calls this the Anti-Fried Liver Defense.  You can guess right away that this player has had their liver fried a time or two too many, and they are not going to let it happen again.  Unfortunately the move is bad:  it violates all those opening principles we were talking about.  --The student retorts:  "But so does 5...Na5."  So maybe they have to try it, and lose some games, and convince themselves it actually is bad.

The StarCraft equivalent is building a whole lot of static defenses at your base right away, cannons or bunkers or spines.  That does a decent job of stopping the early attacks, but it's economically terrible--such defenses cost money, do not contribute to your economy, and can't be used to attack, so your opponent can outproduce you.  A chessplayer will outdevelop you but the result is much the same:  they show up on your side of the map/board with more units than you can handle.

Another SC2 version of this arises when you have learned that a Protoss player can really mess you up by building cannons in your base.  So whenever you see a Protoss worker, you panic, pull all your own workers off their jobs, and pursue him.  You probably won't even catch him, and in the meantime your workers aren't working, which will cost you dearly.  I've done this an embarrassing number of times, despite the fact that it never turns out well.  (As in chess, such weaknesses are bad if you're playing anonymous games on the Internet, but even worse if your opponent actually knows you.  My son is a Protoss player....)

In the case of the Fried Liver the teacher can at least recommend 3...Bc5.  It's not too hard to convince most people that this is better than 3...h6.  StarCraft has a rock/paper/scissors dynamic which means that no plan is consistently successful.  You can play a "greedy" strategy, optimizing your economy, and lose to a sharp early attack.  You can make the sharp early attack, and if you run into strong defenses, your attack will be broken and you are unlikely to recover.  Or you can defend strongly and hope your opponent is not greedy, because they will own the whole map before you are ready to peek out from your defenses.  Some days it feels like anything you do is wrong. 

"Learn to scout," say the masters in response.  But scouting is hard for beginners:  it takes a lot of attention and you don't have any to spare.  You end up staring at your scout while your economy languishes.  Somewhere around this point, in chess or in SC2, you start to realize that these games are really hard.  --I sometimes wonder why I don't have any easy hobbies, something to do after a hard day's work when my brain is worn out. 

All this leaves the teacher with some difficulties.  You'd rather that your students didn't spend too much time playing the Anti-Fried Liver or its StarCraft equivalents, because it's not building good habits.  But a player can't succeed at either game on pure memorization, so at some point your students will have to find out why some approaches aren't as good as they look, take their lumps, and learn to think for themselves. 

I think it's easy for a teacher to imagine that if they just explain very clearly why a particular way of playing is bad, their students won't do it.  But for at least some students this is a vain hope.  (I'm not slagging such people:  I'm one myself.  I have trouble grasping why something is bad until I actually lose because of it, and I'm not inclined to take anyone else's word on the matter.)  I see myself falling into this with my own chess students:  "Bring all your pieces to the party," I repeat over and over, but in the heat of battle they still leave half of them at home.  And my older student said to me a while back, "I've learned from you that my biggest weakness is greedy captures"--and two moves later took a piece the younger student had apparently left en prise and immediately lost his queen to a discovery.  (Good on you, younger student--that was no accident!)

The StarCraft teacher ViBE spends a lot of time ranting about this in his videos:  one has the sense he's never reconciled himself to the fact that he says "Don't do this" and his students turn right around and do it.  But they always will:  for some, at least, that's how they learn.

There's also the fact that players have emotions, and those influence what we play.  A predominant emotion in both chess and StarCraft is panic.  I suspect just about every player who learns SC2 goes through a phase where if they're attacked, they put their finger on the "make more army" button of choice and hold it down until all their resources are gone.  You can tell them not to do this but it does no good:  that's what panic is all about.  All you can do is teach strategies and wait for them to get over the panic on their own.  There are similar deep-rooted forces which lead us to either lash out or cower helplessly when pressed hard.  I've lost a lot of StarCraft games by thinking "Oh gosh, my position is bad--maybe I'd better take all my remaining units and attack him right away."  Occasionally it works (just about anything works occasionally) but usually it doesn't.  I've lost a lot of chess games exactly the same....

"But it worked!"

Starcraft has a lot of established openings.  None of them involve Zerg building both ravagers (souped-up roaches) and mutalisks (flying harassment units) early on.  But gosh, that looked good to me, and I played it incessantly for a while.  After a week or two I could tell you why that isn't an established opening:  both ravagers and mutalisks require the same limiting resource, and lots of it.  Your "advisor" will be singing an incessant chorus of "We need more vespene gas" all game long, and you will tend to lose.

--You won't lose all your games, though.  Another thing that makes teaching hard is that wrong strategies sometimes work.  In fact at the lower levels they work pretty often, because your opponent is just as random as you are.  My first "promotion" in StarCraft was on the strength of my ravager/muta opening, misguided though it was.  This makes stubborn students even more stubborn.  I have a young student who will never resign:  maybe I could have convinced him otherwise, except he managed to draw my slightly older student when down a whole queen, due to stalemate.  And he has never forgotten this.

What's a teacher to do?  In the end I think you are there more to nudge and perhaps guide than to lead.  This is particularly true in a highly individual sport like chess or 1v1 StarCraft, where during the actual game you have to make the decisions on your own. 

It's easy to feel you aren't accomplishing anything as a teacher when your student persistently ignores your advice.  I recall playing a young C player who didn't castle.  In the post-mortems his coach chewed him out:  "How many times have I told you to castle?!  Why didn't you castle?"  I felt very prickly and defensive on my opponent's behalf, because I wouldn't have liked to be talked to like that myself. 

I played the same person, now a B player, a year or so later.  He didn't castle, and I won again.  I don't know what his coach said but I imagine it was pointy.  Another year went by, and I ran into him in the Class A Championships.  He didn't castle, provoked me into an attack which turned out to weaken my own king's position--and beat me handily.  Evidently he had better things to do than listen to his coach!

It can feel like you are talking to a brick wall.  But to counterbalance that I'll bring in a story from a third difficult skill, aikido.  I am not a physically graceful or athletic person, but I fell in love with aikido and after thirteen years of diligent and frustrating practice managed to earn my black belt.  (And then, alas, fell down a flight of stairs and sprained both ankles badly, which was the end of my aikido practice.  By the time they recovered, the pandemic struck and the dojo closed down.)

One day, prepping for a belt test, I was working on kokyunage, the "breath throw", which is sometimes called "twenty year technique" because it is hard to get right.  My dojo taught it right from the start (if it's going to take 20 years you better start early) and I knew how to do it, in some sense, but it looked and felt terrible and my partner could generally stop me.

And then, to my vast surprise, I did a good kokyunage, and my opponent swirled around and fell on the floor, looking completely surprised.  I did a couple more and they were all fairly good.  My dojomates said, "Wow!  What are you doing differently?"

And I said, very shamefaced, "You know that thing that sensei has been telling us to do for the last five years?  I just did that."

I don't know why I couldn't just do it right away, except that it never made any sense.  The impulse in kokyunage is always to grab the opponent, because intuitively, if you don't grab them you won't have any leverage.  Sensei patiently explained that this is wrong.  I parroted that explanation to my own juniors (our dojo required everyone to practice teaching).  Having been pointedly told not to grab, we would refrain from grabbing with our hands, but we'd try to "hook" the opponent with our wrist or elbow instead. 

It doesn't work like that.  In a correct kokyunage you genuinely don't grab them at all.  When you try this as a beginner, it feels like you have left a key part out of the mechanism and it can't possibly work.  In fact, when I first did a correct kokyunage I wondered if my partners were just being nice to me. It took being thrown that way myself to really grasp that no, it actually does work if you do it right.  Sensei was awfully patient about this, probably because she knew from experience that there was nothing she could do to hurry the process.  But when she saw me on that belt test, she said "Well!  That's significantly improved."  She was sparing with her praise, so that was a big deal.

So this is something you can at least hope for as a teacher:  that sooner or later, one of the seeds you planted will sprout something useful. 

Thinking like a beginner is hard!

ViBE's SC2 teaching series involves him trying to play at a speed and complexity appropriate to the student's level, starting at a very beginner level, and showing techniques that such a player can readily grasp.  He does this in the context of playing actual games on the competition ladder.  When he is playing Bronze opponents he aims to have a full army in 12 minutes.  (For comparison, my first attempts hit at around 15, but it was pretty easy to get to 12.)

One of my favorite bits of the video series is an episode in which he is doing this, and someone on Twitch asks him an intriguing question about the past history of SC2.  "Distracting," he mutters, and then ventures an answer.  His mouse flickers across the screen, he tells his story about the past, his army springs up, and then he looks at the clock.  8 minutes.  "Oops.  Sorry." 

I have never been able to break 9 minutes with full concentration, so I'm both impressed and infuriated by this game.  He wasn't even playing at full speed the entire time!  I think it may be the only time I've seen extreme distraction make someone play much better.

But what this says is that playing below one's strength is actually very difficult.  I didn't really understand this until my son asked me to emulate the White side of a Sicilian at a level appropriate to his (around 900).  I recommend this exercise--trying to play like someone much weaker than yourself--to any medium to strong chessplayer, because it's unexpectedly educational, as well as difficult.

The first thing it drove home is that chess is tactics all the time.  You can be quietly playing a normal opening, nothing untoward, and think it's a strategic game.  But by my level (class A) you are constantly testing your moves to make sure you aren't giving your opponent a tactic, and constantly probing your opponent's moves in case they have given you one. Very frequently, you can only play the strategically desirable move because it is backed up by tactics--for example, opening praxis is full of lines which apparently hang a pawn or even a piece, but for tactical reasons capturing doesn't work.  If the tactic weren't there, or you didn't know about it, you couldn't play that line.  (Try to avoid learning a memorized line but not its tactical underpinnings!  This never works out well.)

You can improve at SC2 by just developing your economy and, when the time comes, blindly throwing your army at the opponent.  It's not quite as easy as ViBE makes it look, but up to a point, it works.  While SC2 looks blindingly tactical when you watch masters play (partly because their economic moves are so fast you can hardly see them, partly because commentators tend to focus on the tactics), chess is more so, and it is hard to imagine teaching someone to play even remotely capably without first teaching them a whole lot of tactics.

I was White in this line (or something transpositionally similar) against a newcomer in a recent event.  What was I thinking in this position?  I played a4 to gain a little space on the queenside, which my coach wants me to experiment with more.  And it's not a loss of tempo, because there's a tactical threat which Black must respect.  Unfortunately my opponent did not:

I had no particular intention of springing this trap--I've played this opening a lot lately and I'm used to Black playing ...a6 or ...a5.  But the possibility of the trap underpins my moves, because otherwise Black could response to a4 with a developing move and I would get behind in an opening which, despite its name (Giuoco Pianissimo -- "quietest game") can easily yield sudden attacks by either player.

Of course it's a truism that chess is, up to a certain level (often given as class A) 90% tactics.  But trying to refrain from calculating anything beyond "he takes me, I take him", as I imagine a very low-rated player would do, drove home that this is literal.  I have sometimes wondered whether we should, as teachers, emphasize strategy more early on.  (I coach my students through survival-mode Puzzle Rush almost every week--but is that because it's a good teaching tool, or because it's easy and requires little preparation?)  But trying to imagine chess without tactics drove home that it would be difficult and frustrating to do so.  A quiet positional opening is a nice, level trail--through a murky swamp of tactical diversions and disasters.

When I tried to play like a 900 player, my son nearly beat me.  I calculated that a certain natural defensive move would lose my queen, counted the moves in that combination, gulped, and played it anyway.  He pounced at once--okay, it's easier to see tactics for yourself than for your opponent, but did I underestimate how good 900 is?  But then I got a passed pawn with a rook behind it.  There's a simple technique to stop this, explained by Nimzovitch:  blockade the pawn with a piece.  "The passed pawn is a criminal who must be kept under lock and key; milder measure like police surveillance are not sufficient."  I said this to my son afterwards, when he was tearing his hair over having lost the game:  probably it will stick.

Here's the game:

Did I cheat? I thought queening the pawn was an almost mindlessly simple strategy, easily accessible to a 900 player.  But I'm not sure.  (Reader comments are welcome here!)

The other thing this exercise drove home is that I actually don't have a fully functional mental model of the way weaker players think, despite having taught my kid students weekly for the entire length of the pandemic. (I certainly don't remember being a 900 player myself; it was over 45 years ago!)  I don't need such a model to beat a weaker player.  But do I need one to teach them?

GM Julio Sadorra offered a different approach.  At summer camp he would set up a position and ask two of us, in consultation, to play it against him.  He chose positions that were objectively far superior for us, but then he brought his full skills to bear.  Owen Xuan and I managed to beat him...once, in a week of trying.  It was infuriating but definitely educational.  I have tried this once with my two students.  They correctly recognized a potential Greek Gift, chased my king all the way to the fourth rank, and then lost the thread, several pieces, and the game.  (While they consulted, I turned off the sound on my Zoom.  It's definitely unfair if the stronger player can hear the weaker ones' discussion.)  I'll have to try this again.

I am not a very experienced chess teacher, and I'm really not sure if trying to play to my students' level is pedagogically useful or not.  But I will recommend the experience, just for what can be learned from it by the teacher.  I am amazed by the gulf of ignorance I feel when I try to imagine how my son, or my students, are looking at a position. 

This also shows up when my son and I look at Tactics Trainer together.  There are some puzzles where I can lay out my thought processes, and they're similar to his; we understand the position the same way, and try stuff until it clicks.  There are others where I can explain nothing, because I calculated nothing:  just looked at the puzzle and immediately knew the solution, sometimes quite a lengthy solution, without calculation.  I have to hope he finds this inspirational, because otherwise it doesn't seem useful at all!

I saw ViBE do something similar in SC2.  He was looking at a replay of one of his games, and paused the replay at the point where his scout had entered the enemy base.  It hadn't yet seen the key buildings which would reveal the enemy strategy:  just a couple defenders, some workers, and a base structure.  But, he said, now you know exactly what strategy your opponent is using, assuming they are playing efficiently.  This is an early mutalisk attack.

Then he had to explain.  It was around 10 minutes of explanation for a moment in the game that took two or three seconds (that's as long as you can spare to look at your scout, which is why scouting is so very hard for low-ranked players).  I'm sure his explanation was correct, but I don't think he did the complicated calculations and comparisons implied by it.  Not consciously, anyway.  I think he looked at that base and said "Mutalisks incoming."  Just that.  (He was quite correct, and when the mutalisks showed up 30 seconds later, he was ready and waiting for them.)

My experience is that people who want to learn this skill in chess try timed puzzles--I tried that too--and I think this is often counterproductive.  Until you know the pattern, trying to execute it fast makes no sense.  (I could look at that base for ten minutes and probably not work out that it's mutalisks, but I'm sure if I looked at it for 3 seconds I would have no clue--I don't even register the timer bars when I'm rushed like that, let alone interpret them.)

My best success came from doing non-timed Tactics Trainer and having a penalty for missing a puzzle.  A small house chore works well.  Then you grind on each puzzle until you get it right, because doing dishes is less fun than doing chess puzzles.  This seems like the opposite of training for intuitive speed, but you have to have ground on enough of them to form patterns before you can get to speed.  Another, similar approach is Survival Mode in Puzzle Rush.  Getting reset back to the beginning is annoying, so you try to get the puzzles right rather than accepting a failure and moving on, and I believe it's more useful.

Summing up

That's a welter of thoughts, and I'm left with as many questions and answers--particularly, should I try to play at my students' level, or do puzzles at my son's level, in order to help them?  Or is that exercise purely for my own benefit?  And, when is it worthwhile to insist that your students should castle, and when do you have to give up and get out of their way? 

The one thing I feel pretty clear on is that learning a new skill is good for you:  it may have no practical use in itself--I can win money at chess but it is vanishingly unlikely I will ever be good at SC2--but it stretches your brain and makes things click together in interesting new ways.  If you're at the point in chess that I am in SC2, I salute you and encourage you to continue.  It's fun; it can be even more fun as you improve and can see more complex and lovely ideas; and it keeps the mind lively and agile.