Simple idea for EGTBs

[jwes]

Create dtm tables for all piece combinations where the 50 move will never come into play with best play. This should be the large majority of tables. Programs would play perfectly in these cases and nearly perfectly in dtz50 since bad moves would cost the win or draw.

[bob]

Create dtm tables for all piece combinations where the 50 move will never come into play with best play. This should be the large majority of tables. Programs would play perfectly in these cases and nearly perfectly in dtz50 since bad moves would cost the win or draw.

This is not so easy because you can start off in a position where N reversible moves have already been played. The DTM tables can then take you directly to a drawn path.

In a game, this should not happen, but with analysis it will since humans don't play these things optimally anyway.

[jwes]

Create dtm tables for all piece combinations where the 50 move will never come into play with best play. This should be the large majority of tables. Programs would play perfectly in these cases and nearly perfectly in dtz50 since bad moves would cost the win or draw.

This is not so easy because you can start off in a position where N reversible moves have already been played. The DTM tables can then take you directly to a drawn path.

In a game, this should not happen, but with analysis it will since humans don't play these things optimally anyway.

So you are saying that you really need both dtm and dtz tables to do good analysis.

[bob]

Create dtm tables for all piece combinations where the 50 move will never come into play with best play. This should be the large majority of tables. Programs would play perfectly in these cases and nearly perfectly in dtz50 since bad moves would cost the win or draw.

This is not so easy because you can start off in a position where N reversible moves have already been played. The DTM tables can then take you directly to a drawn path.

In a game, this should not happen, but with analysis it will since humans don't play these things optimally anyway.

So you are saying that you really need both dtm and dtz tables to do good analysis.

It depends. A simple example is the WDL case where the basic assumption is that upon entry to that specific file, the 50 move counter is zero. Then it gives correct answers. But if you call it with a 50 move counter value that is not zero, it can be wrong.

Humans are bad about repeating a position, such as when near a time control, and this causes problems for these kinds of files.

The DTZ files address this by telling how far to the next irreversible move, and you can add that to the 50 move counter and discover whether you can win or not. But then many are complaining that the distance to the irreversible move is not as good as the shortest distance to a mate.

Bottom line for me is if I can win all winnable games, and draw the ones I am supposed to draw, I'm happy.

BTW the idea of having WDL + DTZ50 + DTM seems a bit wasteful, particularly when you venture beyond 5 pieces.