TascBase lesson: analysing.
Mats Winther
The most important theory development technique is looking at games and
trying to find novelties. This is the technique which professional players
use. TascBase is very suitable for this technique - by key split; import
position from main board, etcetera.
However, there are many other techniques, as well. Here is one that suits
me very well. I have database icons on the TascBase Desktop (see elsewhere)
that are configured with Overwrite (in game score). Also they are
configured with Opening Book Window at Main screen, etcetera (naturally,
it´s not necessary to use this setup, but it´s practical).
While wandering about in the Opening Book (in Main mode) I can try new
alternatives and save these as new games in the empty Scratch gamefile (I
don´t want it in the repertoire-tree immediately). This is why it is
necessary to have Overwrite mode - otherwise every variation would be
inserted in one single game (although this is useful when using other
analysis techniques). I save these lines without the usual game
information, and continues analysing the variations.
In these analysis lines developed using the engine, I sometimes insert
statements like "With the the idea 14.Nf5", etcetera. I write this in the
white player position. This is very elegant since after compilation this
statement will show up when stepping on that position in the tree (under
which the game is stored).
Prior to saving every individual game I always set an evaluation symbol at
the end of the line (by pressing F5). These different evaluations will be
visible at the end of every branch in the tree, after compilation - Try
this! It´s very practical and elucidative.
If I change my mind about an analysis line (happens often) then I can
overwrite the latest line with the new one by simply pressing Ctrl-S.
Prior to compiling the Scratch file to create a tree from it, I let the
chess program (The King) analyse the final position of every game line
(with command Gamefile/Analyse Gamefile). This is to have every line
commented (these games are later automatically stored in the
repertoire-tree leaf nodes). Although it´s only a computer evaluation it
has some value, at least if you have given 10 minutes for every end
position. These engine-comments will be stored as variations from the end
positions. Sometimes I let the engine analyse overnight.
To set the King to analyse the end positions exclusively - set the Move
Range from whites 999:th move (or whatever). Whenever the startmove is
beyond the actual end move, The King will exclusively analyse the end nodes
(like Zarkov in Bookup). One can set this start move under Game/Analyse
Game. I have this 999-start move set as default (in the "Repertoire
database with Overwrite"-icon, by having saved the current settings in
TascBase).
(Note! One could set The King to analyse the entire game as well - but I
want to concentrate the analysis CPU-time to the important end positions.)
(Note! When The King is ready with the game file analysis, he shuts down
the Analysis window (in case you wonder)).
Then, if the computers evaluation is different from my evaluation symbol -
I simply remove this line so that it doesn´t end up in the tree. So this a
good double-check.
If you let the engine analyse for a long time on the positions you should
set the Keep n Analysis Moves to a higher value since a long analysis line
can better be trusted.
When compiling, set it to import the games also (if you have let The King
analyse and have inserted comments in the White-player position). If you
want to compile with game import, you should prior to this delete the first
half of the games in the Scratch file since The King adds the analysed
games to the bottom (if you don´t import the games this is not needed, of
course. The evaluation symbols will show up anyway in the branches). The
compilation is done in a few seconds.
After compilation I import this new tree to the corresponding
repertoire-tree. And there I have lines with symbolic evaluations visible
and also symbolic/textual comments that lights up when I step on the node
positions. Also, there lays computer evaluations under the end nodes. This
method is a surprisingly rapid and effective method of constructing theory
in an elucidative form.
Mats Winther