How to get better at chess tactics?

Depends on how you define "successful". It's probably the most efficient way to get from 1200 to 1800 USCF rating. But if your goal is 1900 or higher, you'll need to study other areas of the game besides tactics, or else you'll hit a hard wall around 1800.

But it does show the importance of tactics training, and the idea of repeating the same sets over and over to learn them well is a good one. I mix that in with other types of chess study, doing sets of tactics puzzles over and over, but also studying other areas of the game (openings, endgames, positional play). I already hit the 1700s USCF, and now I'm shooting for 1900 or more. Besides, varying my chess study keeps it more entertaining, as opposed to just drill tactics over and over.

I am undecided on MDLM's program and the 1800-2100 range.  I think you are right about studying other things, for the most part, but I still think that ( a few?, some?, many?, most?...I don't know) players can benefit from the program in the 1800-2100 range.  I've snooped as best as I can to find players in the USCF A-class and Expert rating, and I have found: 1) a number of them have low tactics ratings on some servers, and 2) there are published games in which they make extremely simple tactical oversights.  For players 1800-2100, I'm not going to present an opinion, I'll just do what I think is giving me success, and I measure success in a number of ways, the most relevant being that I am cashing in on players USCF rated 1800+ in rapid games and longer time controls.  Being that the vast majority of my time has been spent on MDLM's program, I have to think there is a causal relation.  Maybe I just take to the program better than most, and it won't be as effective for others.  I don't know.

For 1800+ USCF players, probably as important or more important than solving basic tactics problems is finding complex combinations, such as in the Encyclopedia of Combination or in Polgar's big "Chess" book.  I have begun spending 1/3 of my tactics time on slowly studying complex combinations, the remainder on MDLM's Seven Circles.

That's an interesting post, worth quoting:

"... [students] can find a 3-4 move problem in 10 seconds if they know the pattern, and that they can fail to find a mate in 2 for 10 minutes if they don't know the pattern). look at the answer, and now go over the answer 3 more times in your head to help the pattern take hold. your brain can probably take on 2-3 new patterns between sleeping, so you should stop once you've been stumped by 2 or 3 problems (usually will take about 10-15 min). there is no point in doing more than that in one day. and any day you miss, you can't make up for. a semi-random estimate on my part is that you need about 2000 of these patterns to become a master. so you need to do this for 2 years or more."

I think I already explained this though. Am I just bad at chess? Do most people see things instantly that I have to think about? Maybe it isn't true for you, but there was a point in time where I had to think about it a while just to see if there was a hanging piece on the board. But now, after much practice, I see hanging pieces instantly. I don't see why I shouldn't apply that same patience to learning these tactical patterns.

If I were to go to fast tactics, all that would happen is that they would get easier and easier until eventually I could see them instantly, and I would just get a bunch of instant mate in 1s, easy skewers, immediate forks etc.) How does doing a bunch of mate in 1s and basic one move tactics help me? It wouldn't challenge me at all, or force me to think.

It seems to me the question boils down to whether the ultimate value is in developing simple pattern recognition or in learning to calculate.  The MDLM/Heisman approach seems to emphasize pattern recognition, whereas doing unfamiliar sets of more complex problems probably develops one's ability to calculate.

My own view is that both skills are probably important and complimentary, though like xman, I'm currently focusing on the latter approach.

Once in Tactics Trainer, click the "Customize" button and then uncheck "Show Timer". The timer gives us an indication of the difficulty of the puzzle. We don't get any hints when playing an actual game, so we should train without those hints as well.

Think of it like a little kid learning to read. First, they learn the alphabet. They know what sounds each letter makes, so the first time they see the word "cat" written down, they can sound it out. The more often they see the word "cat" in writing, such as over and over in Dr. Seuss books, the easier it is for them, until the word "cat" just jumps off the page, and they recognize it instantly.

Now think of two kids: One who has learned the alphabet, sounded out the word "cat" when they saw it in writing once, but has never seen it again. The other has seen the word "cat" in writing dozens of times and recognizes it instantly. Which do you think will have an easier time sounding out the words "catastrophe" or "cataclysm" the first time they see it on the page?

They need to work on those short, building block words until they become instinctive, slowly mixing in more and more longer words as they go. Just reading every word from a dictionary exactly once, even if you go in order from easiest to hardest, will make it harder to learn the long words than if you've seen the easy words hundreds of times already.

Chess is the same way. Think of knowing how the pieces move as the alphabet. Think of one move tactics as short, easy words. Think of longer combinations as longer, more difficult words. The better you know the easy stuff, the easier it is to figure out the hard stuff the first time you see it.

Your first sentence nails it - are you shooting for calculation ability or pattern recognition? The answer is that you should be shooting for both!

No, the de la Maza/Heisman approach isn't just about pattern recognition. They both say to use puzzle sets that start easy and get harder as you go. You should be doing some calculation the first couple of times through the set, especially on those tougher puzzles. But by the time you've been through them 7 times, you'll recognize most of the patterns, and only have to do a little calculation to confirm the solutions to the tougher ones work. Then it's time to move on to the next set.

So repetitive, untimed puzzle solving helps you develop both calculation and pattern recognition ability. Playing slow games helps with calculation. Playing blitz helps with pattern recognition. This is why all of these are essential elements for training if you really want to improve.

How to get better at tactics?

TT trainer, period.

That's the point.  You don't want to have to think about simple tactics.  

They've done studies that show IMs and GMs don't do more calculation than experts.  In fact, they often do less.  The strongest players have such strong pattern recognition skills and, as a result, chess "intuition" that they don't have to look at nearly as many variations as a lesser player.  They are not analyzing and calculating, they are recognizing.

But they only way to learn to recognize the patterns behind comlicated tactics is to do them.

If I do simple tactics, I am not challenging myself and I will not learn to do more difficult tactics. I can see every skewer on the board instantly, but that doesn't help me see a skewer after a check and a coersian sacrifice. I need to know that particular pattern, knowing the building blocks doesn't help me. In that way, phonix doesn't make a good analogy to chess tactics.

I think Fromper, there is something else.

The person who went through the dictionary and saw all the words once had the advantage of see catastrophe, and cataclysm and every other "cat word" already. He learned the pattern of "cat" just as much as the other kid, they just learned it differently. He has read "cat" hundreds of times too, just as part of a word. It still improves his ability to spot and sound out cat.

In the same way, I've found that doing slow tactics improves my fast tactics more than doing fast tactics does.

I think that I shouldn't try to say this advice works for everyone though. I know how my brain masters things because of my experience playing the piano. Some people don't, and perhaps this method wouldn't work for them. I've gone through all of these same dilemas with my piano practicing, and chose the slow practice route for sight reading, which is very, very similar to tactics training. I can tell you that I started out by "calculating" (reading each individual note in each chord as slowly as I needed to to get it) but now I sight read by "intuition" because I recognize the patterns in the music. 

So while I understand the allure of having the building blocks, I believe that I know enough about how my own mind works ot have a good reason to design my own curriculum.

Obviously, at some point, I will see my tactics and/or my gameplay stagnate, and then I will have to shift my strategy again. It's not about one particular training method when you are learning a skill as deep as chess, it's about avoid complacency in your training. Obviously, right now I don't feel much of a need to change my strategy when my tactics rating on chesstempo is improving by about 60 points a day and climbing hundreds of points past 1200, but that will stop eventually, and I'll apply some of what I've learned here to my thinking. I'm especially intrigued by picking 1000 problems and doing them over and over again as a baseline. I'm not very intrigued by "learn 3 new patterns a day" because I don't know how to learn or define a "pattern" and so I wouldn't be able to apply that method very well. That might work very well for other people, because that's the way they think.

Maybe a year from now I'll look back and say "that method for learning tactics was awful, why did I ever do that?" But there's nothing I can do but try my best now and ultimately, the only thing that I've come to accept is constant about chess improvement is that it will be hard and slow, regardless of the method I use.

Thanks for answering the questions everyone.

Again, nobody's saying you should just do one move puzzles over and over. And I don't think anyone here is arguing in favor of timed tactics. I think we all agree with turning off the timer and doing them untimed, at least on the first pass.

When de la Maza said to use 1000 puzzles over and over, he suggested the set that comes with the CT-ART computer program. It starts easy, but they get harder as you go, with probably some 7 or 8  move combinations by the end. But by doing them over and over, he did the couple of hundred easy ones several times, just like he did the really hard ones several times.

Personally, when I first started applying this method, I started with John Bain's book "Chess Tactics for Students", as recommended by professional chess coach NM Dan Heisman. These are all easy beginner puzzles, 1-3 moves deep. I did the puzzles from the book once, which made some of them easier, since they're grouped by themes and stuff, so that provides hints when you do several in a row. Then, I cut up the book and put all the puzzles on flash cards, and did them in random order over and over until I could solve all 400+ of them in roughly an hour.

Right around when I reached the point of having those puzzles down cold, I went into a USCF tournament with my 1250 rating and scored two wins and a draw against opponents with rated 1800, 1620, and 1850, so my rating shot up to 1420 in one tournament.

Next, I moved on to Murray Chandler's book "How to Beat Your Dad at Chess". Don't let the silly title fool you - this is a great book on common attacking patterns. Again, I did them in the book once, which provides huge hints because they're grouped by theme, with an explanation at the beginning of each theme. But you still have to calculate as much as 8 moves deep in some of them. Then, I photocopied the book, cut it up, and put all the puzzles on flash cards again, to repeat in random order until I could spot the solutions instantly.

For the longer sequences, I would frequently be able to tell the theme of the puzzle by looking at it, but always need to calculate and visualize the longer sequences every single time, even after I knew the puzzle. And there are plenty of examples where there are similar puzzles, so you need to look for specific factors to tell them apart to know which variation this is. There are even some in that book where the tactic fails, so it's good to recognize that, too.

I hate using random puzzles from the internet. Books of puzzles chosen by a professional coach for their instructional value will be far more useful. And you can choose a book based on the proper difficulty level for you, and work them again and again. I'd say do books with 250-500 puzzles, not the 1000 puzzle plan that de la Maza recommends.

I like this Fromper guy.

Except for all of those people who have not developed the ability to recognize patterns, but have memorized the particular problems by doing each one a dozen time, and who still suck at tactics yet are pretty good at Tactics Trainer.

Becoming better at tactics - use a tactic trainer and use your brains. Do puzzles on chess.com and read the comments. Read about them, do them, remember them. Exercise and eat healthy. Take care of yourself. Chess is a mind game. You can't be good at it without a sharp mind. When learning anything, be patient, perseverant and practice a lot. If that doesn't make you a better player, then I'm Daffy Duck.

That's a cool way to do it Fromper. I've never considered actually spending money on chess before. I started in this place a few months ago because I figured chess is fun, but before I knew it I decided that I could become an IM before I died. I appreciate all the ideas, and will apply them to my thinking.

One thing I've found is that I am extremely bad at avoid perpetual check in puzzles. I have to calculate every queen check because I don't know any rules of thumb. Would it help me to do lots of avoid (or cause) perpetual check puzzles, or would it not help me if I know what the idea is about? They are the last type of puzzle that really stumps me. I'm started to get every puzzle on chesstempo very quickly, even as the difficulty rating goes up I am improving faster. It's only the avoid perpetual puzzles that I still get wrong, and I don't know how to solve them or best how to practice them.

The best way to get better at any particular puzzle type is to just do lots of them. The more you see them, the more you'll remember the ideas behind them. So even if the positions in your games are slightly different, you'll have a better idea of what to shoot for, or avoid.

You are a smart guy, i believe you will improve lots. For tactics i can advice you to get good books for both theory and practice. I mean two different things by theory: a book demonstrates themes, a book offers advice on thinking methods, calculation etc. I had real difficulties at tactics starting studying some books helped me a lot only in 4 months (i didnt read those books by half yet) Jacob Aagard's "how excell in ...." books great but Valeri Beim's books are even better but i can say both are above my level and probably much above yours as well. I had to work hours on a single exercise in Beim's book sometimes but i can say i enjoy the book and learn a lot.

I think the best way is practice. Tons of puzzles does some of the work. Another way is to play sharper games, not necessarily just openings. In any given somewhat open position, there is usually a way to inject a little dynamism into the game, so look for those! Thirdly, studying games that feature extremely complex tactical ideas is probably a must. I can't speak from experience, but doing all of this would probably help me improve! :)