Saturday, July 30, 2005

100 CTS Failures

After my last 1830 tries on CTS I have got well-documented 100 failures now, so I did not get exactly the 95% success rate I set as goal. There was one mouse slip, two doubtful cases where my move should be better or at least equal, and 22 blitz blunders. I have collected all these failures in an excel list, and now I begin to go over them again, trying to figure out the best (most logic) way to the solution. I've gone over 9 so far, the rest will follow in the next couple of days...

Wednesday, July 27, 2005

Defense Patterns

I give away a lot of rating points on CTS now because I move only if I have seen all defense moves. It is interesting to see where most tacticians fail: in positions with a lot of targets that are already attacked or could be attacked or double-attacked. In most cases the winning move is a silent defense move. In the Circles I have learned mostly attack patterns, and so I welcome the many defense problems on CTS. I try to list what I have noticed these days:
  • Taking forward as opposed to taking back: The winner has (a) the richest capture path or (b) the least investment or (c) both.
  • Zwischenzug: Check, capture or threat before taking back, in order to save from double attack or to gain material.
  • Aggressive guard, type I: A piece captures and same time guards an own target, and if this one is taken, the aggressive guard survives after taking back.
  • Aggressive guard, type II: Giving check or making a threat by taking back on a safe square, thus gaining a tempo to save a second target.
  • Double defense: To parry a double attack, one of the targets moves to a safe square where it guards the second target.
  • Desperado: A piece cannot be saved so it sacks itself at highest price and with maximum threats.

Tuesday, July 26, 2005

CTS Failure Stats

Today I invested 30 rating points to regain 94.8% success after some blitz blunders. My failure stats (n=77) is roughly 30% blitz errors, 30% missed defense moves, and the rest is missed mates (10%), missed trapped pieces (10%), false counting (10%) and missed geometric patterns (10%). But it seems that I am learning: Yesterday I missed 9 defense moves, today only 3.

Saturday, July 23, 2005

Caissa's Tactics Server

That's how I dub CTS now. The longer I use it for training, the more I am convinced of it. With low-rated problems I took more risk today and so I made a couple of blitz errors. But I also proudly present this trophy that no less than four 1500+ tacticians had failed to solve before! I just decided to stop my session then and enjoy.

Thursday, July 21, 2005

Time management on CTS

I optimized my time management on CTS as «Slowmouse», beginning yesterday evening and continuing today: Play low rated problems as fast as possible (but still safe) to gain points, play higher rated problems slowly (not care losing points), taking minutes for 1700+ rated (losing no points anyway). Got up from 1220 to 1260 (hello PawnSensei), still holding my 95% success rate.

Wednesday, July 20, 2005

Trapping, Defense, Counting

These three weaknesses emerge from my recent CTS training. I try to keep a 95+% success rate, unfortunately I did not manage to reach 96% again. I have begun to keep book about my disasters. Very interesting. So I learned that I frequently (3/17) miss a Queen Trapping when there is a near mate on the board. I also miss many defensive moves (5/17) of my own or of the opponent. Very difficult is escape from near mate and then win. I even made counting errors, one by playing too fast, and one in a complicated captures sequence. But I also got this trophy today. I focus my training now mainly on CTS. Really useful motifs, just as in games. Excellent tool. But you have to take time and analyze your blunders. Then it may help to improve. At least I hope so.

Thursday, July 14, 2005

Busy CTS Day

Slowmouse did 280 today, success rate down to 96%, rating up to 1244. Two trophies of 1700, both as first solver. Excellent training of time management: Play obvious moves fast, take enough time for others, take more than a minute for difficult ones. Nice to watch rating rise steadily. My goal: catch up Mousetrapper (1406) with a success rate of 95+%

Wednesday, July 13, 2005

Happy Slowmouse

Quite happy with CTS as training device now. Got 98% success while rating climbed to 1200 - Slowmouse now 200 behind Mousetrapper. Still, one nightmare every 40 moves is too much. Would mean one blunder every game. Taking time allows that you really learn. And the problems present a lot of themes, sometimes it is just the simple decision «can I take back?» or «how parry this check?». One main advantage over the Circles is that there is much less repetition on CTS. More variations. More patterns. On CTS I can do X-Ray Scans, Target Scans and Filtering, blundercheck, Candidate Move Evaluation, all this with ever changing and always interesting positions.

Tuesday, July 12, 2005

A slow session on CTS

Today I logged in as guest in CTS and decided to play as carefully as I would do in a slow game. Fastest move was 17 seconds, 5 were below 30 seconds, 2 were exactly 30 and only one more than a minute. Ratings of the problems were between 1180 and 1700. My success rate: 100% of 10 problems. My rating dropped from 1500 to 1060. The shortcoming of this strategy is clear: The system rates you as patzer and serves you only trivial problems. So, unless they set up longer time delays, you have to constantly log out and in again to get problems of a 1500+ level.

Update: I logged in with a second personality as Slowmouse, and after some 30 tries I am now the Percentage King with 97%. The strategy of Slowmouse is to get near 100% and not to care about rating. Interestingly Slowmouse is climbing steadily, and the 1200 are already in sight.

Slowmouse does careful target scans, defensive scans and blunderchecks before he moves.

Monday, July 11, 2005

Blitzing patzing on CTS

Today I lost 15 points on CTS, it was not my day. I fell behind Tempo and had to struggle hard to get him again. 2 mouse slips. Many oversights. Not looking at the whole board. Seeing a move but not believing, then playing wrong. And the like and the like again. Just now I see Tempo struggling to get me again. I suppose he will. I see tomorrow.

At the Zurich garden chess scene at Lindenhof I came by and saw a winning move just the sort of simple CTS puzzles in 3 seconds. The guy took about a minute before he moved. Of course he won with «my» move. Two remarks on this: Chess is about winning, not about winning fast, and CTS is a tough school but helps to get a quick eye.

Saturday, July 09, 2005

Targets ready for tactics

When I scan carefully all targets through a whole game I usually find 4 to 8 targets on each side at any given moment. Of course most of them mean nothing, because they cannot be attacked or an attack can easily be parried. The number of targets means absolutely nothing, so the winning side can have more targets than the losing side. In the Korchnoi game I recently analysed the targets I found that Korch, when starting his final attack, had nearly all of his pieces and pawns to be labeled as targets. But of course his opponent had nor time nor means to attack these «targets».

So, today, I updated my list of Target Filters with a final and decisive filter. It allows to sort out the targets that are ready for tactics. In any case it is important to look at the geometry of the targets: Are they on squares of same color, in the same line or diagonal, on a Knight Circle, or susceptible to a Queen Fork? This can be done in advance: As soon as a new target emerges, just check the geometry with the other targets. I hope that by this method I should be able to find tactical shots in a forced way, not just by intuition.

Friday, July 08, 2005

Target Filters

Any piece or square on the board can be a target. After several attempts to make a list of useful targets I try it the other way, to define a sort of formula or algorithm that can be applied on every piece or square and that works as a sort of filter, and the outcome should be either «target» or «no target».

First filter:
What objects can be targets?
  • pieces
  • pawns
  • squares next to the King

Second filter:
An object is a target, if it is either not protected or its defender is
  • a King
  • a Queen
  • attacked
  • pinned
  • overworked
  • compensating an attacker

Third filter: An object is a target, if it
  • is or can in one move be attacked by a piece of less value
  • is under x-ray of a piece of less value

Final filter:
Targets are ready for tactics if
  • two of them can be attacked by one move (fork, discover, ...)
  • they are unable to move (pin, trap, ...)
  • their guards can be removed (deflection, capture, ...)

Yesterday I applied these filters, scanning the whole board after every ply, going through a brillant Korchnoi game, which took me about an hour. The finish was a double piece sacrifice leading to a mate. I had identified all the targets before I saw how Korchnoi made use of them in his combination. Then I applied my target filter to a puzzle I had missed before. After I had identified the targets, the solution was quite obvious. So I feel that I am on the right way.

Board Vision Day 24
X-Ray Jogging: 1.60 sec/scan, 1 hour

Tuesday, July 05, 2005

Target Scan Training revised

Blogs are no manuals. So be prepared that I revise my training methods all the time. Just one major misconception of my yesterday's post: Target scanning is supposed to take place mainly on opponent's clock, so in a slow game there is no hurry. This means, it can and must be done in a very thorough way. Secondly, even more to blame as a blunder, is my idea to leave pawns «behind» for the moment. Rubbish! If you do not appreciate pawns, then you deserve no better than lose them.

So, here again, is my (revised) list of targets:

  • a King that can be checked
  • an X-rayed King or Queen
  • a piece that can be attacked by pawn or minor piece
  • an unprotected piece or pawn
  • a piece attacked by as many (or more) pieces as defend it
  • a pawn attacked by as many (or more) pawns or pieces as defend it
  • a pawn that must protect a piece attacked by at least 2 pieces
  • a piece or pawn protected only by King or Queen
  • a piece or pawn protected by an overworked piece
  • any square around the King protected only by the King

Now, after every ply, I do a total board scan and enter the number of spotted targets in a pocket calculator. Stop watch running. After an hour or so, I stop the watch and count targets. Besides that, I do some X-Ray Jogging described earlier.

Board Vision Day 21
Target Scanning: 4.72 sec/target
X-Ray Jogging: 1.57 sec/ply

Monday, July 04, 2005

Target Scan Training

When I saw the nice rake of Temposchlucker I thought to myself, this is what I need: Pull a rake over the crowded board and collect the targets. Well, there are many targets, but not all can be attacked at once. So my job must be to create a filter. As a first step I decided to concentrate on pieces, leaving pawns behind at the moment. A piece is a target when it is vulnerable. To keep it simple, I define primary targets as follows: A loose piece; a piece that is attacked by as many pieces as defend it; a piece defended only by a King or Queen; a piece defended only by an overworked piece (that has to defend another piece, a pawn or an important square). A piece attacked by 2 pieces and defended by 1 pawn is not a target in this sense, but the defending pawn is hanging; this is a different story.

So, after some experiments, I came to this Target Scan Training, part 1 (pieces): Go through a game with the computer. Count the target pieces. At the beginning, it is the 4 rooks. After every ply check what has changed in the position. If one more target piece emerges, count it. If one disappears, take note of it, but do not count backward. Proceed with this until the endgame begins. Write down the target count. Take the next game. Have a stop watch run. Take total of target counts and time, calculate the time needed for one target.

In my games of today there was roughly one new target with every 2 or 3 plies.

Board Vision Day 20
348 targets scanned, 9.6 seconds/target