Wednesday, October 31, 2007

CT-Art 767 narrative

Some problems in level 40 of CT-Art are too complex for me to learn through repetition. For these problems, I will do what I learned from Temposchlucker and and BlueDevilKnight (are there others that did this that I am forgetting?), I will write narratives for these problems.

CT-Art 767:

White to move

The weakest points in black's position are g7 and h7. It would take many moves to get to h7 and if black plays h6, there's probably no route in. So g7 is where to focus the attack.

The immediate attempt to crash through, 1. Nxg7 Bxg7 2. Bh6 f6 allows black to defend g7 laterally with the queen. We would like to do 2 things better. The first is do not allow the lateral defense with the queen. Secondly, don't give up the knight for the bishop since the knight is a good attacker of g7.

There is a safe check that will allow Rxf6, so we examine 1. Nh6+. The king must move to f8 or h8. On 1. ... Kf8 we have 2. Bc5+ Re7 and now the pinned rook will lose for black after 3. Rxf6 gxh6 (gxf6 Qg8#). The critical line is 1. ... Kh8 2. Rxf6 gxf6 3. Qh4. The queen attacks the newly weakened f6 square.

Black to move

3. ... Qe7 Black must defend f6 (3. ... Re6 4. Rxd8 Qxd8 5. Nxf7+ loses quickly). White attacks this defender and the weak g7 square 4. Nf5 Qe6 5. Qh6.

Black to move

Black cannot defend the g7 square. 5. ... Rg8 6. Rxd8

Labels: , ,

Monday, October 29, 2007

Taking a break

For a few reasons, I'm going to alter my training regimen. I'm going to stop doing the 3 week repeats on the CT-Art problems.

I finished level 40 on Thursday at 93%. Even with time left to start on level 50 for the first time since I started 3 week repeats, I decided not to. For one, it has been helpful to write narratives on the level 40 problems. There are some on that level that I don't think I could get just with repetition. So that seemed more useful than trucking forward on level 50.

I believe I need to play more games, and in the way that blitz doesn't count. I generally don't play long time control games because I simply don't have time during the week if I'm solving CT-Art problems. Since I recently signed up at, I've fired up some turn-based games there (I don't understand the terminology here, aren't all chess games turn based?? I would call it correspondence).

I've noticed in recent games that I have a tendency to overestimate my tactical chances. The simplest hypothesis is that constantly solving problems where I have a tactical win has made me over agressive. Hopefully if I put CT-Art aside for a spell and play some real games, I'll get a nice balancing dose of reality.

Of course, the blog will change with me. I may post a few more narratives from CT-Art, but annotated games will likely take a more central focus.

Wednesday, October 24, 2007

CT-Art 765 Narrative

I'm once again delving into the narrative writing, and I've once again found an error in CT-Art in the problem set.

CT-Art 765:
White to move

White is attacking the king, but should be wary that his own king position is not perfect, the rook on f3 is pinned and the knight on c3 is attacked. At the moment, g7 and h7 are weak near black's king. The queen attacks at h7 adding a second attacker with g6 or Rh3 is not currently possible because these pieces are pinned. The bishop attacks g7, an attacker can be added with f6 and Qh6.

The immediate f6 suffers from the fact that opening the g-file gives some play to black. 1. f6 gxf6 2. gxf6 Rg8+ 3. Kh1 Qxf3+ 4. Rxf3 Bxf3#.

In order to play g6, white can sacrifice a bishop with 1. Bxg7 Kxg7 2. Qh6 Kh8 3. g6 fxg6 4. fxg6 White is threatening mate and CT-Art gives a poor defense for black, 4. ... Qc5+ 5. R1f2 Qg5+ forcing a trade of the queens, but giving up a defender of the f8 rook, 6. Qxg5 Bxg5 7. Rxf8. Instead, black can give up material temporarily with

White to move

4. ... Rf7 5.gxf7 Rf8 where black is getting his material back due to the pinned rook on f3. The idea is that if black wants to survive, he needs to keep his counterplay active, not trade queens.

If white wants real chances he needs to keep the lines to his king closed. A much simpler practical choice is 1. Qh5 with 2. g6 coming. 1. Qh5 bxc3 2. g6 and now even ... h6 fails to Qxh6, so black doesn't have time to take the knight on c3. 1. Qh5 f6 2. g6 h6 and now white sacs his bishop on h6, 3. Be3 with Bxh6 coming. In this way, white attacks the king position without opening dangerous lines to his own king.

Black to move

White attacks the black king, threatening Bxh6 without opening lines of attack for black.

Labels: , , ,

Monday, October 22, 2007

CT-Art 735

I was going to write a narrative on CT-Art 735 and decided to double check it on Fritz. It turns out the CT-Art solution has mistakes in multiple places.
CT-Art 735:

White to Move

The pawn on f7 is weak (only defended by king) and is the only defender of e6. In this kind of position, 1. Nxf7 Kxf7 2. Qxe6 immediately comes to mind as so many squares around the king can be controlled. 2. ... Kf8 is mated immediately by Bh6 so 2. ... Kg7 is forced. We can stay with forcing moves by 3. Bh6 Kh8 and now simply pry open the h-file with 4. Bxg6 where black loses quickly if 4. ... hxg6 5. Bf8+ Nh7 6. Rxh7+ Kxh7 7. Qf7+ Kh8 8. Qg7# (5. ... Nh5+ 6. Rxh5+ gxh5 7. Qh6+ Kg8 8. Qg7#). Other 4th moves for black give back excess material, e.g. 4. ... Rg8 5. Bf7.

The above is not the CT-Art solution. CT-Art starts off with 1. Nxf7 Kxf7 but then CT-Art gives 2. Rxh7. This is a good invading move, rooks are typically very well placed on the 7th rank:
Black to Move

The rook is immune from the knight, Nxh7 Qxe6 Kg7 Qxg6+ and mate follows with Bh6 or Qxh7. So black continues with 2. ... Kf8. Here CT-Art gives no credit for the very strong 3. Bxg6 instead favoring 3. Rh6 Bf8 (other moves may be better) 4. Rxg6 and then a serious error from the defense 4. ... Bg7? (Kh8 is more complex and doesn't lose nearly so quickly) which allows 5. Bh6 Nf4 6. Rxg7+ Qxg7 7. Bxf4 where white has three pawns for the exchange.

The CT-Art solution, while winning, is very complex compared to the first solution I gave. Even then, the solution given is helped along by inferior moves from the defense. A proper defense would have made this solution extremely difficult in practice.

Labels: , ,

Sunday, October 21, 2007

Writing some narratives

So far I'm doing well on level 40, but there are some problems where I'm just drawn to the wrong solution every time. So I'm taking a page out of the BDK and Temposchlucker play books and I'm going to write some narratives about these positions.

CT-Art 734:
White to Move

In this position there are two possible squares for white to invade with a knight, f6 and f5. The defenders of these squares are removed by 1. Rxh5 gxh5 and then either
a) Nd5 followed by Nf5 or
b) Nf5 followed by Nd5 and Nf6.

I'm always attracted to option (a), which is not correct. It appears correct because 2. Nd5 exd5 3. Nf5 attacks the queen and the queen cannot defend the mate on g7. However, this idea overlooks 3. ... Be3+ followed by Qxg5 (or Qe6 if Nxe3).

Option (b) is a winner because the bishop check on e3 doesn't interfere with the knight on f6 or the attack on h7. The queen is also unable to defend h7. So, 2. Nf5 exf5 3. Nd5 Be3+ 4. Kb1! Qxg5 5. Nf6+ Qxf6 6. Qxf6

One of the bottom lines is that h7 is weaker than g7 due to black's ability to get the queen to g5. This makes f6 the right square for a knight rather than f5.

I've got a few more of these to write...

Labels: , ,

Saturday, October 20, 2007

Tough losses

I had some tough losses at the coffeehouse this week.

In the first game I played against the Benko gambit. I don't typically have a lot of success against this opening, but I think I came out of the opening just fine. I've posted the full game in the analysis forum at, here, feel free to comment on the game here or there. I've also posted the game at chesslog just in case.


Wednesday, October 17, 2007

Progress on rep 5

I finished level 20 at 97%. This is down from 98% in rep 4. I think the decrease is due to complacency as I wanted to get through the level as quickly as possible and often did problems when I was tired and not fully focused. Though interestingly, I did focus on problems I had gotten wrong in the past and got many problems right that I got wrong previously.

With the disappointing setback on level 20, I decided to focus properly on level 30. The result: 97%, up from 92% on the previous rep. This is a big move up on a level that used to give me such problems.

I also project that on my current schedule, I'll get to the level 50 problems this time through.

I almost forgot to mention that I've stopped reporting my elo from CT-Art because it doesn't seem like a relevant measure anymore. Incidentally, the max rating achievable on level 20 is 2125 and on level 30 2275.


Tuesday, October 16, 2007

Housekeeping; Meta Post

When this blog started I had no imagination for what it would become and as a result, it was not very organized. Until just recently it has lacked even the simple blog element of tags (or categories). I have now gone back and tagged every blog post. Now if you want to just see posts that include tactics problems, simply click on the "Tactics" category on the sidebar.

I think this is a major upgrade to the useability of the blog. For example, I've made a number of posts where I criticize or point out flaws in CT-Art. These are now all easily found by clicking on the "CT-Art errors" category. So if you want a rather disorganized snapshot of a user's experience with CT-Art, there you go. If you'd rather see what I've done with CTS, there's a link for that too. If you're interested in seeing what kind of chess is being played by a USCF Class B player, check out the tactics/diagrams posts or the Full games posts (note: full games posts contain links to games posted at chesslog).

The increase in useability is good for me personally (yes, I do read my own blog -- I was thinking what?!), but a secondary reason for it is the recent revelation that there are in fact people who visit this blog. Which is the subject of the next part of the post.

The remainder of this post is mildly embarrassing as I didn't realize until somewhat recently how many people come to visit this blog. Don't get me wrong, the numbers aren't huge, but I always figured it was only a couple people who ever saw it. A few months ago Blue Devil Knight made a post about blog cred and a couple weeks later I had installed Google Analytics to see if anyone comes here. The results have been enlightening.

First, not many people visit this blog as a result of search engines. About 10% of all visits were as a result of a search engine. The largest sources of traffic are links from other blogs. This blog is linked on the sidebars of a couple very popular blogs and this results in a large fraction of visits. An even larger driver of traffic is being linked to in a post rather than on the sidebar. There was a big spike in referrals from Likesforests blog when he linked to one of my king and pawn endgame posts. The bottom line is that people visit by following links from other blogs.

The second interesting thing that I notice is that most visitors are first time visitors and only 1 in 5 come back for a second visit. There also seems to be a core of visitors that come frequently enough to see nearly every post (but I probably know these people from the comments already). The large majority of people who are new to the site is one of the reasons for the increased useability. If the site is potentially interesting to them, I might as well make it as easy as possible to navigate to the interesting content.

Next is something I saw someone else write about on their blog -- the geography of visitors. Chess blogs attract visitors from all over the world. Over 59% of my visits are from the US and 30% from Europe (90% of those from Western and Northern Europe), but there are also visits from the rest of the Americas, Africa, Asia, and Australia. I am the only visitor from my current city of residence. But I was very happy to see a few visits from Mexico City during the World Championship there. A side effect, based on a conjecture, is that I now believe there are search engine spiders operating out of Los Altos, California and Rock Island, Illinois.

Being from academia, the following excerpt from the list of network locations intrigues me: Duke University, universitaet trier trier, technical university of crete, university of chicago, williams college campus, bucknell university, dartmouth college, indiana univerity-purdue university at indianapolis, massachusettes institute of technology, rutgers university, southern illinois university, university of kansas, university of lausanne, and western iowa tech cc.

Saturday, October 13, 2007

Adventures in Coffeehouse Chess

I have a decent memory from my trip to the coffeehouse. I think the tactics I found today are a decent representation of the kinds of things I can find much better now than I could 6 months ago.

Black to Move

Due to a mating net on the back rank with Bf3, black picks up two pieces for his rook with 1. ... Rxf4 where 2. Bxf4 Bxf4 3. gxf4 is met by Bf3 and mate next. The game continued 2. Rxd6 Rh1+ 3. Ke2 Bf3+ and white resigned in view of Rxd1+.

Black to Move
(hopefully it's clear that the labels (a-h,1-8) on this diagram are backwards. oops!)

Material is unbalanced with black having two minor pieces for a rook. Black's long range plan should be to increase the activity of his two minor pieces. It turns out the unfortunate placement of white's pieces makes these positional ideas moot. 1. ... Qf6! and there's nowhere to run with the rook on a1 hanging in the backround. Protection doesn't help either, for example 2. Qf2 Re2 3. Qxe2 Qxd4+ also picking up the rook on a1. The game continued 2. Qf4 Re4 Fritz says Ne6 is better, but Re4 is so beautiful, it's why we play chess.

White to Move

Once again, just a lucky unfortunate placement of the pieces. The knight on b6 has had it's squares taken away by the queen and bishop trying to line up on h3. The placement of the queen is also important for finishing off the combo. The game continued 1. c5 Na4 2. c6 Bxc6 3. Nxc6 Nxc3 4. Nxe7+ Black resigned.

White to Move

In this position I engaged in dangerous pawn hunting on the queenside. My opponent has made inroads in my king position, but missed a chance to finish me off, instead going for a tactic that was refutable. There are dual threats of Nxa1 and Bd4+. White comes roaring back with 1. Qb8+ Bf8 2. Bh6 Qc5+ 3. Kg2 Qc6 4. Kh3 Qd7 5. Bxf8 f4+ 6. g4 Rxf8 7. Nxf6+ and black resigned since Kg7 Qxf8+ Kxf8 Nxd7+ leaves white a full rook ahead.

Labels: , ,

Sunday, October 07, 2007

Level 40, rep 4 done at 90%

I finished the 4th rep of level 40 today at 90%. I'm at the end of 3 weeks, so I'm going back to level 10. Since I was out of town for about 4.5 days during these 3 weeks, I don't feel bad about the time taken. However, I need to speed up a bit if I'm going to move beyond level 40 in the 3 weeks.
| Level 1 2 3 4 |
| 10 97% 99% 99% 100% |
| 20 92% 94% 95% 98% |
| 30 86% 86% 86% 92% |
| 40 77% 77% 84% 90% |

Scoring over 90% on the first 4 levels feels like an accomplishment. But I also feel so far away from being able to see the variations fly through my mind for many of the problems. This is what IM Ziatdinov says it should be like after really learning a position.

Does this software exist?

Given that everyone who attacks a set of chess problems seems to find some errors in the set, I wonder about a software that allows the user to edit the problems. Even better would be a program that would accept an entire problem set and score you according to the definitions in the set. That way, someone could work on producing a very good problem set where the user interface and progress tracking are already taken care of. If the length of the problems is unrestricted, you could even make solitaire chess games for it.

Does this exist and I just haven't found it yet?