Ask HN: Why are chess problems attack oriented?

It's due to how the puzzles are generated. Positions that have a jump in centipawn evaluation are used as puzzles. If you're under 1500 then stick to the basics. Analyze all your games. The only way to improve is to practice and study.

This is a really fascinating psychology question. because I have been trying to present the same question in cybersecurity for years now.

The closest I have gotten is that a defensive posture motions toward a stalemate, where as an attack mentality mobilises toward a win.

For peace, a (static/deadlock) stalemate is a good outcome. So at heart, the question is about the difference between a war footing or a peaceable one.

Because control of the center and the edges are in fact attack oriented.

Even the top defenses have an attack basis of whether the prevent the control of the center and the edges.

But the real devious defense to force your opponent into a winning a piece that destroys their position attack strategy.

1. Attack puzzle has more clearly defined solution.

2. There is no difference between attack puzzle and defence puzzle, in the sense that mentally you must play both sides - choose optimal moves for the defender.

This is definitely a problem. Defensive problems do appear, but much more rarely than attacking ones. I guess their rating gets pumped up pretty fast, because they're much harder to solve - or people are just used to looking for winning variation - so they disappear from lower ratings altogether. Aimchess[1] worked surprisingly well for me to get at least a bit more well rounded training going on.

A lot of defence can also fall under the umbrella of strategy rather than tactics, as in the dream scenario you'd basically only be trying to keep the position in a state that has all the needed defences built into it. Doesn't really work like that in reality. But studying and practicing strategic principles of the positions you're playing could be another thing that could help, especially the pitfalls of those positions. Most positions have a bunch of well known pitfalls that will improve your defensive game quick if you keep those in mind while playing. A lot of these can be found on YouTube.

One other thing to try out is to also play normal tactics, but with the board flipped.

[1]: https://aimchess.com

One of lichess's criteria for a puzzle is that there must be only one winning move (with an exception for checkmates in 1).

This makes it much easier to select for not terrible puzzles with an algorithm, and to have a good UI for puzzle solving, but it filters out many defensive puzzles both because it excludes positions where there are multiple defenses, and it excludes positions where you just manage to not lose without turning the position into a clearly winning one.

Usually puzzles start from a position where one side has a clear advantage and a single good move. It depends on your puzzle rating, but many puzzles include an important defensive component (e.g. preventing the opponent from mating you while you keep the advantage). However in low puzzle ratings I believe it's more mate-in-X, material grabs, and other simple attacks.

You can also use the defensive theme on Lichess: https://lichess.org/training/defensiveMove

Go to chesstempo and filter by tag "defensive move" (or something like that).

On lichess at least there are definitely some 'defense oriented' as in finding the move that won't lose you material. At any rate, even the attacking puzzles often have a 'defensive' part as you have to exclude the moves that lead to a tactic for your oponent.

Further, spotting successful tactics for yourself is a very related skill to spotting them for the oponent, you can even just press f (I think) to look at the board from their PoV and look for their tactics that way.

Disclaimer: I have little experience with chess but i see good reason for that from a theory of game trees kinda angle.

> Why are chess puzzles not 50% about spotting and preventing attacks?

Chess is a problem where you need to reason about all the actions your opponents can do. Since a lot of chess teaching tradition comes from a linear medium (books) they have a trouble representing how an attacker would attack differently against if different defensive move are played. This problem does not exist as attacker as one can force certain moves on the defender. Thus the branching factor of the puzzle can often be contained. If books would represent the branching reactions, the book will also spoil which moves to pick, finding a solution to a chess problem would be reduced to reading all the replies and pick the most favourable. Another problem is that "defend for 6 rounds" or "defend all material" might allow and train solutions which survive satisfy that criterion but cause the player to loose (a sacrifice would have been the only way to prevent losing) while solutions exists where this isn't the case. In a linear medium it is not possible to handle that well without just printing a solution and say that is the right one.

> Any advice to train that specifically?

You need to leave the linear medium. This is possible very easily by playing certain board configurations against an AI. However generating such a board positions for which there is a few or even one unambiguous chain of moves that constitutes a valid defense is difficult. My wild guess is that they are near the border of endgame libraries for both players. I think a chain of moves would constitute a valid defense if it leads to a state which is in the players winning end game library for all actions the opponent takes however to make this puzzle difficult this chain of move needs to be hard to find and other moves must be actually bad. My proposal would be to search for board positions where the vast majority of moves including some promising moves lead to a position in the attackers end game library while there exists one chain of moves which independent of actions of the attacker lead to a position in the endgame library of the defender. Caveats: finding moves at the border of the two endgame libraries with these properties might be hard. Many positions of these are likely synthetic and would never arise in natural play.

Is it not just perception? And an arbitrary choice (in the POV) in presenting a "chess problem"? Dependent on ranking?

I play defensively so with that underlying mindset I solve chess problems to improve my tactical acuity/prowess ("Beware of that blunder!"). For that I naturally flip the board in my mind back and forth ("What blunder happened?") and at times go through the back-history of moves when I wonder: how the hell did that happen?

I played with some higher ranked player whom I would consider "highly aggressive" but who turned out to be quite defensive players when competing with similar/higher ranked ones, so there is also that ;) Why play defensive if you are way better at anticipating and spotting blunders. So, this comes back full circle at the importance of building a solid base.

Yours is an interesting thought. Do you have experience with games like go? My personal experience has been that the attack-oriented style of most chess problems does help you avoid those positions in normal play. That is to say you learn defense at the same time because you have to think of all the lines in your head.

There are some good midgame books, if you can find a copy of Lisitsky that's one

Problems are attack oriented so that there is a forced solution

Just play more

[Not a chess god but reached ~2170 USCF back in the day]

Basically all these puzzles are trying to train your combinational vision and calculation. An important component of calculation is getting good at executing 'forcing' moves (moves where your opponent has only one or two plausible choices) which are more clearly illustrated in attacking positions or 'Mate in X' positions. Even though the most readily available material is from 'attacking' positions, this skill is transferrable to many other kinds of positions(e.g pawn endings). Defense on the other hand is often more vague with more than one optimal strategy and rarely forcing, so there's just not a lot of good material to draw from.

There is a chess training program called "Dr Wolf" which has a learning mode where you play the computer. At certain points it asks you to switch positions to see the weakness in your previous move - it is striking how effective that is to seeing the weaknesses in one's defense.

I am not particularly skilled in chess but the LiChess puzzles are a daily practice on the subway... I think it is my favorite. I really like the puzzle scenarios where you are down material but the correct series of moves leads ot a win. This insight is a confluence with BJJ where sometimes you can be in a bad position and reverse out of it...

Because chess is 99% tactics and those involve getting ahead in material or mating in max 4 moves or so.

My 6 year old has gotten into chess in a very unexpected way. He was spending a week with his grandparents and I guess grandad got bored and decided to teach him chess and suddenly my kid wants to play chess all the time.

So I play him one day. Now, I'm not a good chess player, but I know all the rules, even the obscure ones, and I my attention span is a lot better than a 6 year old's. And he almost beat me.

We signed him up for chess classes with an instructor over Zoom. He's beating kids with a lot more experience than him. And I find myself trying to train to stay ahead of him. I sit in the room with him during his classes and pay attention, did all the lessons on chess.com and lichess.org, and am constantly doing puzzles on lichess (even using the training dashboard to focus on areas in having problems). I train probably 5x more than my son.

And he's still giving me a run for my money. I let him do chess puzzles with me and he frequently gets frustrated with how slow I am on some of the puzzles. He "sees" the 1-move and 2-move solutions almost immediately. The only place I have advantage is anything 3+ moves. If he doesn't see the move immediately, he doesn't know how to develop to a position that he can solve. I'm not much better, but I can occasionally get it, and that's the only way I'm staying ahead of him right now.

I'm similarly frustrated with the training on lichess. When I play games, I find I'm completely lost on the middle game. I play the openings that my kid's teacher has been drilling, then try to sit in a holding pattern of not making egregious mistakes until the other player makes a mistake I can recognize. I'm basically incapable of moving the game in a direction I want, and my simplistic play is easy pickings for a lot of people as they bait me into--in retrospect--rather basic traps.

I've just about resigned myself to living vicariously through my child on this issue. His classes being online, the instructor is often distracted with the other children being noisy to give real attention to training. My son recently moved up to a new class level (and he's now the youngest in the room!), so hopefully that will get better. I try to fill in that in-person coaching by just parroting what his teacher says and pointing out where my son isn't following along quite right.

The "mixed" set of problems on ChessTempo is an excellent problem set with a large fraction being defensive. ChessTempo has some of the most interesting/instructive problems, and the community commentary on each problem (viewable after you fail or complete a problem) is priceless.

I'm not really sure what the difference is. The best defense is a good offense, no? I'm only about 600 puzzles in, on lichess. Sometimes you must out-attack (race to checkmate), other times you just want to get ahead before you disable their attack.

Maybe you like the Quiet Move type puzzles? Those are a weak spot for me.

There is definitely an iOS chess app that deal with defensive problems. Usually an obvious move where you had to do to avoid checkmate or loss of material

To get better you definitely need to prevent giving your opponent winning tactics. It’s important to study being the active side for two reasons:

1) Not always but very often spotting a winning tactic wins the game immediately 2) You develop an intuition about various piece position combinations and geometries and features that usually produce these tactics in a game.

Now consider these a problem, say white to move, and go back one move. Likely, black just made a move which enabled this tactic. It’s hard to craft a puzzle where you are asking what Black should do: the main thing is to not make the losing move, but are all other moves that avoid it just as good? Not at all!

There are lots of books that try to teach and train this, but it doesn’t fit into an efficient package like the “find the tactic” one does.

One exception here is when one side has sacrificed material for a speculative attack. In those cases just avoiding mate is a kind of puzzle that fits. I have occasionally seen some puzzles like this (eg. Attack and Defence by Aagaard) but it’s a lot more work to put together, and occurs much less often in games which is probably why it’s much less common.

> All problems seem to be attack oriented, not defence.

Probably because it's easier to design puzzles for mating/attacking and people want to attack/mate.

> As a result in a chess match I constantly look for ways to attack my opponent, not to prevent an attack from my opponent.

You will eventually learn to defend too as you play more games. It's pattern recognition.

For example, there are varieties of ways to attack a fianchetto position. Once you feel comfortable exploiting fianchetto positions, you'll notice your opponent attacking your fianchetto positions and act to prevent it. Just like constantly trying to find back rank mates against your opponent will eventually teach you to watch your own back rank. Learning to attack teaches you to defend and vice versa.

When I was a boy I regularly played against myself. You should try it, it really helps for learning how to defend.

You shouldn’t ever play “defense”, its just not how you play. You need to actively attack your opponent.

If you are trying to come up with some sort of passive turtle strategy you are going to get crushed.