Monday, May 19, 2008

A tribute to Mike Anderson

Speeding up

Glenn Wilson drawed our attention to the nifty program of Mike Anderson. Since I experienced trouble with my double attack-scan I decided to use his program.

The scan for double attacks is quite time consuming and so far I haven't seen hardly any habit forming to speed things up or to make it easier. It is a compound scan and maybe it works better to train the constituent parts seperately. To that end I'm experimenting with Mike's great little program.

The program is based on the microdrills of DLM. Let me demonstrate.

White to place a queen.

The problem is where you should place a white queen in order to attack both targets simultaneously in a way that you win the rook. To solve this you need to scan for the following:
  • where do the diagonals which radiate from the rook intersect with the lines that radiate from the king?
  • can you skewer?
  • can the rook intercept the check? (that's why the queen can't stand on b3 in the diagram)
  • can the king protect the rook?
Even such simple scan has 4 constituent parts already which have to be checked seperately. I strive for a speed of 2.6 seconds per mouseclick. When I try to do it faster the mouse handling begins to interfere. Any slower means that I'm thinking in stead of applying a habit.

In stead of a rook and a king the program enables you to do the same exercise with a bishop and a king, which is a comparable exercise, except that you scan for the horizontal and vertical lines that go out from the bishop in stead of the diagonals.

The program give you the possibility to use a knight and a king, which combines the exercises of rook-king and bishop-king. That is to say, you scan for all the lines that radiate from the knight for intersection with the lines radiating from the king. Be it horizontal-, vertical- or diagonal lines. Besides that, an extra element is added:

White to place a queen.

With a knight close to the rim of the board you have an extra possibility to win it. You can place the queen on the squares with the blue edge where she attacks both the knight and all of it's escape squares. That's a different kind of double attack. It doesn't make a difference where the black king is, as long as it is far enough away to prevent the knight from jumping into safety.

The exercise with the knight takes me 15 minutes (5.3 seconds per move). I want to be able to do it in under 12 minutes (2.6 seconds per move).

Thanks for sharing your program, Mike!

What is not adressed.
The program gives me an opportunity to improve on the constituent parts of the double attack. There are two parts of the double attack that are not adressed by the program though:
  • knightforks
  • replacement of targets
The movement of a knight is quite alien to the human mind. The knightvision exercises of DLM are not really making me happy. I haven't found a good alternative though. The checks for knightforks consumes a considerable amount of time of the total double attack-scan.

Replacement of targets.
You don't want to scan for double attacks that are manifest on the board solely. You want to know if there are any potential double attacks too. There are a few techniques and prelimanry moves to place targets on the spots that are under attack:
  • Trade. You can trade a target piece with a low value, thus forcing a takeback with a piece of a higher value.
  • Annahilation of the defender of a target. Thus making the target ready to pluck.
  • Deflection of the defender of a target.
  • Decoying a target to the right spot.
  • Etc..
Especially difficult to see is if the target is an empty square.

If I am able to adress these omissions and if I could proof that you can learn a habit by training it's constituent parts, that would be a major leap forward in the theory of chess improvement with the aid of habit forming.


  1. Caq,
    do you find those schemes to be of help by playing this program?

  2. Yes, sure. In the beginning it will take some time. But after a while when the figures are cemented in your memory you will instantly know if there is a fork or skewer, this will reduce your calculating time. It will just pup-up in your mind and vision

  3. I just scored 10m09s in the knight-department, thus averaging at 2.3 seconds per move. That is fast enough. Now I must find a way to exercise knightforks in a pleasant way.

  4. Does the program tell you when you get it wrong? It seems to let me click on any square and doesn't tell me 'wrong'!

  5. Blue,
    there are two possibilities: after the session you walk through the session in review mode (the best)or you can press show answers during the session. Since I learned to do the concentric circles fast in only 3 days with this program I consider it a very minor glitch that you don't get realtime feedback without extra mouseclicks. The review mode was sufficient to me.