I tend to believe that the first approach helps you more and is also something you need to do from time to time in order to stay in shape.

However, this presupposes that you also play normal games, study books on other subjects, etc. i.e. doing only tactics is not enough, but when discussing tactics, working in isolation seems to me by far the best approach.. unless of course you hate doing it!

@Fromper: Where do you take your clear cut wisdom from? Is there any empirical research about this?

It's very difficult to find a experimental design for answering this question. Chess players tend to do all mixed up.

Jupp53 wrote on 07/23/10 at 09:08:27:

Why do you specifically ask Fromper this question? I am not aware of any empirical research at all on subjects like these.

My experience is like that of BPaulsen.  I never did much studying of tactics per se, and I have been struck by how often people here refer to doing so, relative to mentioning e.g. studying annotated games.

So I seem to be in the minority here in liking tactics puzzles. They aren't my only study, but they made up probably half of my study time in going from 1250 to 1750 USCF.

To me, the difference between doing tactics puzzles and reading master game collections is the difference between doing something and being shown something. I have to do it myself to really learn it. I may understand something while it's being explained to me, but remembering the details comes from doing it myself, preferably repetitively.

When I just read a book of annotated master games, I come away feeling like I didn't learn anything from each game. I realize that the cumulative affect is that I'll get used to seeing common patterns and strategies that come up a lot in high level games, and I'll get good instructional advice from the annotator, so I know it's still worth doing.

And as mentioned by someone else, the tactical variations are in the annotations. Given the sheer quantity of variations in the annotations of any game, I don't play through most of them. When I've tried to do so, I've gotten lost going back and forth between the variations and resetting the board in front of me to the main game position to continue with it. To me, the value in studying annotated master games is to see the explanations in words. The variations are just there to answer questions if I have them (ie Why didn't he just take the pawn?), not for tactics study.

By doing lots of tactics puzzles, I've come to recognize common tactical patterns, so they jump out at my much more quickly during my games. And it's improved my visualization ability, so I'm less likely to overlook moves when calculating ahead during games. That's not everything necessary to be a good chess player, but it's a start, and it's what I most needed when I first started playing USCF tournaments and was rated 1250.

I've seen recommendations to go through master games in a "guess the move" format (preferably from the winner's perspective only), but I haven't tried it yet. It would probably be good study, but it would take me as long as playing a slow game myself in a tournament - an hour or two just for my moves.

Of course grandmaster games are a great way to learn and become better, but for purely tactical improvement studying tactical examples is probably the best. For your friends who might have languished behind in chess ratings, who professed to studying lots of tactics, I wonder if there problem is that they don't realize that it's good positional play that leads to tactical opportunity (atleast that's what's seemed to be true for me).

Thank you all for these thoughtful responses to my original post.  I suppose that without a controlled long term experiment we may never know which path is more effective, but in the meantime we should all probably spend some time doing both.

I'd like to add an additional question for which there may be an answer.  If you look back at the games of players like Pillsbury, Marshall, Alekhine etc, we see a lot of the patterns we now study in our tactics books.  But how did these guys learn the patterns?  They did not have "How to Beat your Dad at Chess" or CT-Art 3.0.  Did they have tactics puzzles to look at or was this all a product of their own creativity and playing lots of games?