Hello All,

I'm a very bad chess player. Haven't played since childhood, and any moderately skilled kid would be me. This may be the reason I don't understand the following.

When I try to play training games and then do post game analysis, a lot of what is explained doesn't make any sense to me. The examples are numerous, and I'll give you just one, so you understand the type of problems I'm facing.

Look at this screenshot: http://brutsoft.com/tmp/cm2.jpg

As you can see the best rated move by the mentor lines is queen to f6. This immideatly results in the queen being taken with no forseeable retribution. So why the program recomends this silly move? Well maybe it's not the move that is silly, maybe that's me, but then, could you please explain why?

As I mentioned, I see examples like this all the time. Any help will be appreciated.

quote:

Originally posted by zespri:
Hello All,

I'm a very bad chess player. Haven't played since childhood, and any moderately skilled kid would be me. This may be the reason I don't understand the following.

When I try to play training games and then do post game analysis, a lot of what is explained doesn't make any sense to me. The examples are numerous, and I'll give you just one, so you understand the type of problems I'm facing.

Look at this screenshot: http://brutsoft.com/tmp/cm2.jpg

As you can see the best rated move by the mentor lines is queen to f6. This immideatly results in the queen being taken with no forseeable retribution. So why the program recomends this silly move? Well maybe it's not the move that is silly, maybe that's me, but then, could you please explain why?

As I mentioned, I see examples like this all the time. Any help will be appreciated.

It looks like you are only requesting one best line. However, CM is actually showing you 4 possible lines. If you look at the time in the first col you will see that only the fourth item is showing something other than 0:00. That appears to be the only active line and is the much more reasonable Bxf6.

If you choose to show more than one best move, then the Bxf6 moves up to the top of the list.

So it appears to be a bug in you only ask for one best line.

Dear Lenard2005,

thank you for you reply. Unfortunately I don't quite understand what you mean, could you please explain? Selecting one best line is supposed (as I read it) to return the best move. The move returned (as per you) is not the best. Why?

Thank you in advance,
Andrew.

quote:

Originally posted by zespri:
Dear Lenard2005,

thank you for you reply. Unfortunately I don't quite understand what you mean, could you please explain? Selecting one best line is supposed (as I read it) to return the best move. The move returned (as per you) is not the best. Why?

Thank you in advance,
Andrew.

In a bug free world, what you say is very true. However, CM seems to still have a few bugs running loose in it - and Mentor lines is one of the places where there are still some bugs!! Maybe they will get fixed in a future patch.

Anyway, this bug appears to show itself when you say you want to see only the one best move. Instead of the best move being at the top of the list, it ends up showing at the bottom of the list!!!

However, if you tell it you want to see the three best moves (for example), the 'bug' seems to go away and the program will show you what it thinks are the three best moves, with the best move at the top of the list. The other two moves are the next best moves moves that the compluter sees from the current position. So, it looks to me like if you want the one best move to show at the top of the list, you have to tell the program to show more than one best move. I know that this is a bit confusing, but it does seem to work.

When I review my games, I often have Mentor Lines set up to show the 5 best moves to see if my move is anywhere on the list. If it is not one of the five best, then I have to really think about why I made that move.

i cannot reproduce the results. however, you're reading the data a little wrong. what you circled is the initial choice it ptroposed. as you can see it changed it's mind a few times before the second was up... it's current choice is at the bottom of the list.

quote:

Originally posted by Lenard2005:
However, if you tell it you want to see the three best moves (for example), the 'bug' seems to go away and the program will show you what it thinks are the three best moves, with the best move at the top of the list.

Yeah, tried to set the value to 3 and the queen takes f6 move is not even present in the list anymore! Really wierd, but works. Thanks, I'll do a few more experiments and if I have more questions I'll be back.

quote:

Originally posted by Tomski1981:
i cannot reproduce the results. however, you're reading the data a little wrong. what you circled is the initial choice it ptroposed. as you can see it changed it's mind a few times before the second was up... it's current choice is at the bottom of the list.

What does the "score" column mean then?

the score column is how how good the move chessmaster thinks it is. at first, it rates it with 1.6+.. but in a fraction of a second it realizes it's NOT a good idea (probably giving it like -6 or more, which you don't see), and changes the proposed move altogether.. the score changes as chessmaster goes deeper, because it's learning more about the outcome. in fact, the line at the very very bottom (all by itself, that changes few times a second), that's the actual line that chessmaster is thinking about. in this example it's: 0:28 2/13 12759040 45 46 11.Bxf6 gxf6 12.Qe3....... however, the best one it's found so far is: 0:10 2/12 0.95 4553483 11.Bxf6 Qxf6 12Qxf6 gxf6.....

when 1 best move is selected, it's always at the bottom of the group... i guess they could have programed it to be at the top, but it doesn't matter when you know where to look for it.

quote:

Originally posted by Tomski1981:
the score column is how how good the move chessmaster thinks it is. at first, it rates it with 1.6+.. but in a fraction of a second it realizes it's NOT a good idea (probably giving it like -6 or more, which you don't see), and changes the proposed move altogether.. the score changes as chessmaster goes deeper, because it's learning more about the outcome. in fact, the line at the very very bottom (all by itself, that changes few times a second), that's the actual line that chessmaster is thinking about. in this example it's: 0:28 2/13 12759040 45 46 11.Bxf6 gxf6 12.Qe3....... however, the best one it's found so far is: 0:10 2/12 0.95 4553483 11.Bxf6 Qxf6 12Qxf6 gxf6.....

when 1 best move is selected, it's always at the bottom of the group... i guess they could have programed it to be at the top, but it doesn't matter when you know where to look for it.

So why the best line has the lowest score in this example then? 0.95 is less then 1.14. And 1.14 is the move I'm questioning (and circled).

The best line has the lowest score, because as chessmaster analysed the game, it 'realized' that the position wasn't as good as it originally thought.

let me see if i can more accuratetly explain it:
computers are dumb... they don't think like you and me. one of the things a computer chess program (like chessmaster) searches of is captures.

1.So here chessmaster sees a beautiful capture Qxf6. (keep in mind it first checks all of it's own moves).
2.Qxf6 at first looks better than Bxf6, because chessmaster is trying to accomplish two things at once: get the queen out of attack AND take a piece.
3.It reports this 'ingenious' move and gives it a score of about 1.14 (1 being equal to about a pawn)
HOWEVER, chessmaster sees a problem when it analyses possible responses of black:it's own queen can get captured! oh no! it realizes things don't look so easy and it quickly changes it's 'mind' and amends the score of it's position to 0.69.. then later to 0.79 then to 0.95.

Every time a new line pops up below the others, it means chessmaster found something and changed it's 'mind' and altered it's opinion of the strength of a given position.

therefore, to sum up:
at first, Qxf6 looked good... and if the game were to end after 11.Qxf6, then white would win... but since the game continues and black will no doubt capture the white queen, chessmaster tries to think of another plan. as you go wait longer and longer, you get a truer and truer 'score' (a more accurate one.. as chessmaster searches deeper and deeper).

No doubt i bet you yourself while playing wanted to do an 'awesome' move that would CRUSH your opponent, only to realize (after a little analysis), that it's actually a blunder and you'd lose badly!

i've done that many times.

i hope this clears it up.. if not, i'm out of ideas but i'm sure others will be willing to explain this futher

Hi, about the mentor lines...how long should I give the chessmaster to make the decision of what the best move was. I'am asking for the 5 best moves, I guess a couple of minutes?? Any other tips for using mentor lines would be greatly appreciated. Thanks!

really, it will never know exactly what the best move is unless it solves the game to the very end. so if you want the absolute truth, give it years. however, for practical purposes a couple of minutes should be fine.