I decided to review some problems I had solved over many months on chesstempo.com, so I bought a one-month silver membership and started working through problems that were red (not correct), yellow (caused tactics rating to drop), or took 60+ seconds to solve. There were so many that I only reviewed each problem about once. When the month was over, I went back to solve new problems, my ability seemed to have tanked (dropped 80 whole points!).

What went wrong? I think the problem was that I was not reviewing problems enough. To really recognize a pattern, you need to review it often--and I was not doing so during the month. Hence the problem-solving atrophy. But there is a solution: see my next post.

By the way, I do not want to discourage anyone from paying for a subscription to chesstempo.com. My problem was that I was not using my subscription wisely, I think.

Cognitive psychology suggests that, to memorize something, you need to review it frequently at first, then less frequently later. So if I want to really ingrain a puzzle solution in my brain, a reasonable approach might be to solve it as follows:

0 day - first attempt
1 day - first review
3 days - second review
7 days - third review
15 days - 4th review
31 days - 5th review
63 days - 6th review

At this point, I will have looked at the puzzle 7 times, and should have the pattern strongly etched into my brain.

Yesterday I played with a spreadsheet to work out a puzzle-solving regimen to use with chesstempo.com that would allow me to solve puzzles according to the above schedule. Based on these assumptions...

• I have time for 15 - 20 puzzles/day
• a history of the 160 most recently solved puzzles is available to me

... I can follow this schedule:

Day 1: History problems 150 - 160; 5 new problems
Day 2: History problems 1 - 10; 15 - 20; 35 - 40
Day 3: History problems 70 - 80; 5 new problems
Day 4: History problems 1 - 10; 15 - 20; 35 - 40

After completing the 4 days, you repeat the cycle (on day 5 follow day 1's schedule, on day 2 follow day 6's, etc.). Then repeat it again, and again, slowly gaining tactical strength. (That's the theory, anyway!) The reason you can just keep repeating the 4-day cycle is that solving 5 new problems shifts the history by 5, so for example problems 1 - 5 become 6 - 10 and problems 156 - 160 disappear from your view.

The only possible problem with this schedule is the assumption that you have a history of 160 problems available. In fact, the free membership on chesstempo.com provides a history of only 100. So I will be modifying the schedule as follows:

Day 1: History problems 90 - 100; 5 new problems
Day 2: History problems 1 - 10; 15 - 20; 35 - 40
Day 3: History problems 60 - 70; 5 new problems
Day 4: History problems 1 - 10; 15 - 20; 35 - 40

This will squeeze the review schedule such that the 5th review is on day 26 (instead of 31) and the 5th review is on day 40 (instead of 63). Less than ideal, but hopefully not a big problem.

Expect a report on this experiment in about 3 months!

My last post discussed a program of puzzle solving designed to help the brain acquire patterns that can be used in game situations. It should be possible to do something similar with regard to my own games and subsequent analysis:

Monday - play a slow game
Tuesday - analyze the game, publish analysis here
Wednesday - review games from 2 weeks ago, maybe play some blitz
Thursday - play a slow game
Friday - analyze the game, publish analysis here
Saturday - review games from 6 weeks ago, maybe play some blitz
Sunday - day of rest

I suspect I need a little less frequent review, both because I invest a lot of time in the initial game and initial review, and also because if I tried to review more I might not have enough time left to play! If I had more time, I would prefer a more aggessive schedule, and would heartily recommend more games/analysis to those who can afford the time investment. This is simply the pace that I can sustain.

Memorizing a specific puzzle solution is not the same as internalizing the pattern.
It would seem that mastery of a pattern requires a review of many many related examples that are not exactly the same.

In any case, if I'm not mistaken, the repetition of certain puzzles is built into many tactics servers and programs. For example, i know the same puzzles come up on chess.emrald at times, and I know that Chessimo software repeats each puzzle 6-10 times over the course of the program. If that's the case, one could rely on the program to supply the repetition automatically without having to manually keep track of such things.

I have an adult student who started at 1400 (now 1600 7 months later), and my advice for him on tactics training is:
(1) keep 1 tactic book in your bathroom, 1 near the TV couch, and 1 at your bedside. pick it up and read it when/if you feel like it - no pressure. the important thing is select an EASY book and keep it in plain sight and easy reach. if u love chess you'll find yourself reaching for it provided it's easy and not too hard! (How to Beat Ur Dad and Ch Tactics for Kids and Alburt's PocketBook are great for this)
(2) 10 minutes of tactics puzzles per day. this one is not optional. he uses chess.emrald but ChessTempo or a book are obviously fine too.
(3) repeated viewings of Zibbit Pattern Recognition videos never hurt anyone.

GL with your diligent mission of chess mastery!

I am truly enjoying the positions I obtain as black in the Tarrasch Defense. No doubt a FM or even a NM could target my isolani, gradually restrict my piece activity, and make me struggle with the long-term pawn structure weakness. However, I am not playing at that level, and the Tarrasch gives me a lot of fun. So why not?



1. B 2. B 3. B

4. W

My analysis:

1. White has a long-term advantage in pawn structure...but "before the endgame the gods have placed the middlegame." Black's Re8 attacks the e2 pawn down the half-open file, the Bc8 can get to g4 adding to woes on e2, and the bishop on b6 can come to life if white's tenuous blockade breaks down. I played 13....Ne4, placing the knight on a strong central outpost. However, it makes Bg4 harder to play, since the response of f3 can fork black's minor pieces on e4 and g4. Thus 13....Nxd4 14. Nxd4 Bg4 might be a little better. Either choice is good, though, IMO.

2. Here I played 16...Bd7 to connect the rooks rather than 16...Bg4 because I feared the simple fork 17. f3. However, targeting e2 with Bg4 is stronger because 17. f3 doesn't work: 17....Nxg3 18. hxg3 Rxe3 19. fxg3 Qxd4 and white's pawn deficit and weak dark squares leave him in big trouble. White might have hopes of a draw due to the opposite-colored bishops, but black is going to have all the fun along with good winning chances.

3. Here I elected to strike hard with the forcing 18...Nxg3!? The idea is that white has a lot of vulnerabilities on the dark squares: Be3, Nd4, Rc1, and of course Kg1. Thus a continuation like 19. fxg3 Rxe3 20. Qxe3 Bxd4 21. Rxf6 Bxe3+ 22. Kh1 Bxc1 and white is hopelessly lost. However, I did not calculate all the lines accurately, as we are about to see. Thus a move like 18....Bg4 piles the pressure on white's position with no risk, and is preferable. Note that 19. f3 would fail due to 19....Nxg3 followed by Rxe3, etc.

4. After a long series of exchanges white can emerge with equality: 19. fxg3 Rxe3 20. Rxf6 Rxd3 21. Rxb6 (the move I missed) Rxd4 22. Rxb7 Be6 23. Rcc7 and white has excellent compensation for the pawn. Instead white played 19. Bg5, which looks like it saves the bishop with a tempo-gaining attack on white's queen. However, after 19...Qxd4 white was simply down a pawn with no compensation. He played one more move, 20. Qg3, but resigned on the spot. The resignation was too quick for an amateur game, but not by much; white's position is an awful mess after 20....Rxe2--down 2 pawns with many weaknesses.

Thanks for the excellent suggestions, Katar! I do keep tactics books around the house, and solve them regularly. I'm not so sure about chessimo and CTS, though:

• chessimo has only 400 tactics problems for $50, so it doesn't seem like such a good value. Of course it also provides strategy and endgame training, which are probably the best value it does provide. Since I'm strongest in strategy/endgame and weakest in tactics, though, it probably doesn't make sense for me. For others, though, I certainly wouldn't discourage it.
• CTS offers very little history (only 20 problems). In fact, I have used CTS in the past. However, I've found that using chesstempo.com with a program of review works better for me, as per the brief but illuminating discussion we recently had. At the same time, I've read favorable comments about CTS by many chess players, so I would certainly not discourage someone from checking it out.

I think that the reason that chesstempo.com with review works for me is that I don't just look for the first move, then click to the next puzzle when I spot it. Instead, I review the analysis as well--in particular, why the responses are inadequate, and why alternative first moves are not as strong. Evidently reviewing the thought process (and not just identifying the first move) also helps to acquire the pattern.

Note to readers: Zibbit's videos are available here on chessvideos.tv, and at least some of them are free. Just search on the keywords zibbit and pattern to find them.

Again, Katar, thanks for inserting some good ideas in this thread.

Hey, thanks. A couple quick notes: (1)Chessimo has 4,000 not 400 tactics puzzles-- value is open to debate.
(2) Zibbit's 5 pattern recognition videos are all 100% free and essential viewing.
they are linked at the top of the page here: http://www.chessvideos.tv/wiki/index.ph ... _by_Zibbit

Chris, what a nice idea (...Nxg3!?) in the Tarrasch game! I had to pause and analyze all the complications. Part brilliant, part insane, but definitely in the spirit of the Tarrasch which i've always felt should be played with a freeflowing gambiteering mentality. (Notably the defense was a favorite of Kasparov and John Nunn in the early 1980s...)

Any chance you will publish a video? Would love to see it. Your attention to detail is legendary.
EDIT: Also i am a fan of your no-nonsense classical style and the way you think about chess.

I wanted to call your attention to this very good video about critical moments: viewtopic.php?t=3654

It is a free video, worth watching and thinking about.

Thanks, looks interesting and helpful!

Thanks for the clarification. How does the problem repetition work in chessimo? Does a problem get repeated frequently at first and less frequently later (which I outlined as the cognitive psychology-inspired pattern above), at even intervals, or less frequently at first and more frequently later (like de Maza's 7 circles)?

These do look excellent; thanks for the link!

Since I started playing openings like the von Hennig-Schara Gambit and the Tarrasch, I'm looking for opportunities to play effective sacrifices. It's a lot of fun! And the shock value often dazes my opponent, so a sac that is not objectively winning (as in the Tarrasch game) or even unsound can still score the point. (But not always.)

I've actually been quite surprised at the interest in my journal; I started the training journal basically for my own learning and thought that maybe the occasional reader might wander by for a gander. But I'm happy that a good many fellow chess players are finding some value in my posts.

I'd like to learn more about the time investment required to make a video before I commit to it. I'm pretty much maxed out right now....

A week ago I scored my first win in the von Hennig-Schara; white moved his queen too frequently, I used a tactic I had blogged about recently, and I KO'ed white in just 14 moves. White resigned with mate in 2 looming.

A couple of days ago in a von Hennig-Schara I sacked a 2d pawn, and then a 3d, but I absolutely would have preferred black's dynamic chances to white's 3 pawn material advantage. (See if you agree with my assessment of the second position below.) Even after missing a couple of the strongest moves, I eventually recovered all the material with good winning chances. Then I overlooked elementary tactics that would have garnered the point--agony, agony! Oh well, I keep learning and having fun anyway; hope you learn and have fun as you analyze alongside me.



1. B 2. B 3. B

4. W 5. B 6. B

7. B 8. B 9. B

10. B 11. B

My analysis:

1. I played 16...Bxf5, thinking that I would be able to play the crushing 17....Nc2+ next. Unfortunately, I overlooked the simple zwischenzug continuation 17. Qc4+ Kh1 18. axb4. Better for black is 16....Rxf5 17. Qc4+ Rf7 18. axb4 Nxf2 19. Rg1 Qb6 and white is in a heap of trouble. Note well: if you think you have a forcing move against your opponent's queen, make sure it cannot escape with a check or counter-threat.

2. White is up 3 pawns, but black has a huge lead in development and piece coordination. It's the 18th move, and white has only 2 pieces developed! The ficsgames.com silicon evaluates this as -0.66 in spite of white's material, and I have to concur. I considered playing 18....Nxf2 (19. Kxf2? Bd3+ -+), but ultimately decided to connect the rooks and pile on more pressure with 18...Qb6, which is a bit more accurate.

3. I wanted to play 19....Nxf2!?, but I was afraid of the fork 20. Nd5. However, it would fail to 20...Bxb4+ 21. Nxb4 Nxh8 and the knight should be able to re-emerge on f2, leaving white with an undeveloped mess and no material compensation. To prevent the fork I played 19....Bxb4 which is somewhat inferior.

4. White can improve his development and coordination with 20. Ra6 Qb7 21. f3 Ne5 22. Qb5 Qe7. However, he played the "natural" 20. Bd2?, leaving himself open to a devastating continuation....

5. ....which I did not find. I couldn't find a winning continuation with my pieces in my play, so I played the inferior 21...Bxc3, which nevertheless leaves black with a very strong position.

The keys here to the winning move are:

• Black has no winning continuation with the developed pieces, so look around to see if anything is undeveloped. And sure enough, there it is on a8: a rook that's pouting because no one invited it to the party!
• Nxf2 did not just threaten the Rh1 and block queenside castling; it also undermined the defense of the e3 pawn.

Combine the 2 insights, and instantly you should be slobbering over 21....Rae8! Silicon suggests the subsequent 22. Rd1 Bc2 23. Be2 Nd3+ 24. Bxd3 Rxe3+ 25. Bxe3 Qxe3+ 26. Be2 Bxc3+ 27. bxc3 Qxg1+ 28. Kd2 Bxd1 -+

6. 22.... Qh6 23. Qe5 (23. Kxf2 Bd3+ 24. Ke1 Qh4+ 25. Kd1 Rxf1+ 26. Rxf1 Bxf1) Bd3 24. Be2 Bxe2 25. Kxe2 Rae8 26. Qb5 Ne4 27. Raf1 Qxh2 28. Kd1 -+. I didn't see that the knight was not hanging because of the white king's peril, so I played 22....Ne4 instead.

7. White is hoping that the threat of doubled rooks on the 7th rank will be so terrifying that black will abandon the defense of his f pawn. However, black has time to protect it with 37....Rd4 because after 38. Rdd7 Rb4+ 39. Ka2 Ra8+ white must exchange a rook to avoid mate.

8. The simplistic 43...f2 is drawish after 44. Rf7. However, black can keep the white rook from getting behind the f pawn with 43....Kg8!, and suddenly black is winning.

9. The simplest win is the deflection 48...Rd3+! 49. Rxd3 f1=Q -+. Even after 48...Kg6?! black still has an easily winning position, though.

Why did I miss this easy tactic? I was so focused on trying to gain time on the clock that I was blitzing my moves almost instantaneously. Lesson: Sit on your hands a little bit! Don't just play the first move that pops into your mind, even in zeitnot (if you have an incremental time control).

10. Again I blitzed out a plausible move (50....Kxg5?), overlooking the winning skewer 50...Ra3+!. I spent less than a second on the move, and I saw the skewer the moment I released my mouse. Again...sit on your hands a little bit!

11. I calculated that white would easily draw the king and pawn ending after trading rooks, so my mind was telling me that I should check the white king with 52...Ra3+ and force white to fight for the draw. In fact, 53. Ke4 h5 54. Rg2+ Rg3 55. Rf2 Kh3 and the Nalimov tables pronounce a draw. Korchnoi would of course be able to draw this easily as white--but my opponent was not Korchnoi! So I should have played 52....Ra3+ and made him keep working for the draw.

Lesson: Don't stop fighting after a discouraging turn! Will yourself to keep playing the best moves; take encouragement from the fact that if you keep putting pressure on your opponent, he might crack.

First, about the intervals of repetition... that applies to rote memorization. Like nailing a long complicated musical passage, which would be performed exactly the same way every time. Obviously such a musical passage must be memorized cold for performances. There is no way this concept should apply to repetition of chess puzzles which are merely examples -- of which there are literally many thousands -- and are not meant for rote memorization nor are they meant to be repeated/performed exactly the same way in a performance situation. As for Maza, he was an unemployed, single bachelor who decided to spend 2 years on nothing but chess. Of course he would reach 2000 in those circumstances regardless of his method (i would hope so!), yet somehow inexplicably he became the demigod of an army of optimistic American class players who blogged their attempts to replicate his success, which obviously none of them did. Anyway Maza clearly destroyed his love of chess (if he ever loved chess) b/c he never played ever again.

Your diligence and organizational acumen is staggering. So you seem to be the type of chap who could pull off the rather complicated tactics regimen you've outlined. I mean you are a software architect after all. I just think the important thing is regularity/frequency of practice.

I believe it would take no more time than the detailed writeups you do now.

Interesting that you bring up the music analogy, because the way that skilled musicians keep improving is by working to master passages that are a little beyond their current skill level. Then having mastered that formerly forbidding passage or piece, they move on to another one. (Actually, those who want to make serious progress work on more than one passage/piece at a time, of course; I'm just outlining the principle.) On the other hand, musicians who just play for enjoyment don't improve very much. (Of course, they enjoy music, and why criticize them for doing what they enjoy?)

The analogy to chess is obvious, I think. If you want to improve, rather than merely enjoy,

• reach a little beyond your current level in some skills/patterns,
• practice them until you are comfortable with them,
• then move on to higher ground (while occasionally reviewing what you've acquired).

How does this affect the choice between mostly repeating puzzles vs. continually solving new ones? I think that the approach you take to solving the puzzles should influence your strategy. If, when you repeat a puzzle, you simply find the correct first move and then go to the next puzzle, you are only improving your ability to solve that particular puzzle. I agree that such an approach is not a productive use of time, and you should prefer a continual stream of new puzzles if that's how you're going to solve them. On the other hand, if you are using the repeated puzzle to review the whole process of how to get at the correct solution...for example,

• the tempting sac doesn't work bc/black has a zwischenzug,
• another tempting line doesn't work because the position after 5 ply is not quiescent,
• the obvious recapture doesn't gain any advantage, so you should look more broadly for alternatives, and finally
• the winning move looks quiet, but threatens a double attack that black can't cope with

then you can solidify some habits and patterns, making them "second nature."

So there is not one right approach for everyone. As you can tell by these blog posts, I'm the sort that carefully reviews the thinking process that leads to the right solution, rather than just looking for the solution and moving on, so I seem to benefit from repetition. At least that's what my experience with different approaches to solving puzzles would indicate.



Although some of them have shown good results, if not quite as striking as his. I think that Maza's chief contribution has been shifting the focus away from gaining more knowledge by reading yet another chess book plus memorizing openings (a frequent club player approach) to gaining skill by deliberate practice. Others have subsequently elaborated on this with a lot more nuance and insight (e.g. Soltis' recent work on chess study), but Maza should still get some credit for sounding the alarm, IMHO.



Solving tactics puzzles is a lot like a golfer's going to the driving range, or a swimmer's practicing the flutter kick every day. These are good activities, but you do need a lot more variety to stay motivated. The golfer also has to practice chipping and putting in various situations, figure out better club selection, lift weights, etc.; the swimmer also has to practice many strokes, do intervals, do body position drills, lift weights, etc; and they all have to compete on a regular basis. Likewise, someone learning to play a game that requires a network of skills--like chess--needs many different types of practice to stay motivated and keep improving. So Maza's fanatical dedication to (mostly) one form of practice probably contributed to his loss of enjoyment and motivation.



I agree wholeheartedly. Just as half of winning is just showing up, half of chess improvement is just practicing frequently--regardless of the method.

Diagrams 3 - 6 come from this Kalashnikov Sicilian



1. B 2. B 3. W

4. W 5. B 6. W

My analysis:

1. White's strategy is to undermine black's e pawn, so black must play 5....f6 on the spot, followed by Ne7 at a later date. Instead, I decided to indulge the luxury of playing for the 2 bishops with 5...a6, only to discover after 6. Bxc6 bxc6 7. c5! Bxc5 8. Bxe5 that I was definitely inferior. Fortunately, black can barely defend with 8....Qg5 9. Nf3 Qg6, and both the g7 and the c6 pawn are defended. Of course, white has a big lead in development....

2. Black enthusiastically played 10...Bg4, but the pawn on c7 is going to drop, and (far worse) white's dark-squared bishop is going to become very strong. It was time to give up the bishop pair with 10....Bd6 in order to save the pawn and keep the positional equilibrium.

3. O-O keeps black under pressure while maintaining white's initiative. Black cannot play 1...Qxc4 because of 2.Rac1+-, so 1....f5, but after 2. Rfe1 the pressure on black is only increasing. Instead I played 1. Qxb3?! Qxc4 2. Qxc4 Rxc4 and not only had white's initiative evaporated, but black's rook had entered the game with considerable force.

4. 1. b4! and white will create a queenside passed pawn that will become a force of nature. Instead I played 1. Rac1?! Rxc1 2. Rxc1 which gave black the opportunity to achieve near equality (see the next diagram).

5. 1....d5 unleashes the bishop, after which a5 and Bb4 and white will have difficulty making any progress. Instead black played 1...a5, which has a hard-to-see flaw.

6. 1. b4! axb4 2. a5 and the white a pawn will tie down black's army, while the black b pawn is harmless. The pawn sac is definitely worthwhile, because if black is given another opportunity to play ...d5, white can make no progress. Best move of the game for white.

A recent brain imaging study of jazz musicians shows that when they are doing their most creative improvisations, the prefrontal cortex (the manager/regulator of logical thinking) essentially goes dark. In other words, the improvising musician stops trying to filter his musical thoughts; instead, s/he just lets the ideas flow. Of course, what emerges is not random (unless it's Sun Ra on the sax)--it's strongly based on patterns that the musician has mastered.

Once you're past the opening tabiyas, playing chess can be a lot like jazz improvisation: you need to let the ideas bubble up. In fact, almost all my best chess ideas just kind of hit me square in the eyes; the horse rears up, chomping at the bit, and requests permission to wipe out the opposing king's pawn shield--so it seems.

Too often, though, my thinking during the game is goal-oriented analysis. Starting with a candidate move, I insist on trying to figure out the analysis tree. This is prefrontal cortex thinking, rather than jazz artistry thinking, and as a result I miss a lot of possibilities. Then I come back afterwards to analyze, relaxed and thinking more playfully, and discover all kinds of wonders that I publish on this blog.

Honestly, I'd rather play those wondrous moves before I blog them! So my goal is to try to think more improvisationally as I play chess, and let the interesting ideas flow. There is a time for tree analysis, of course, but I need to learn to switch back and forth between the 2 types of thinking (improvisational and analytical). I don't know if I'll gain 200 rating points, but I hope to get out of the rut I seem to have been stuck in recently.