Saitek Mephisto Travel Expert vs Saitek Travel Champion 2100

This forum is for general discussions and questions, including Collectors Corner and anything to do with Computer chess.

Moderators: Harvey Williamson, Steve B, Watchman

Forum rules
This textbox is used to restore diagrams posted with the fen tag before the upgrade.
User avatar
Monsieur Plastique
Senior Member
Posts: 1014
Joined: Thu Jul 03, 2008 9:53 am
Location: On top of a hill in eastern Australia

Post by Monsieur Plastique »

That actually adds to the mystery. I no longer have the Cosmos I borrowed but it seems that according to the levels for the TC2100 shown at schachcompuer.info, the level layouts are the same for TC2100, Cosmos and Expert Travel. All have the same fun levels ("e" file) and adaptive levels ("h") file.

I am wondering if Saitek somehow "tweaked" the fun levels from Cosmos onwards and this might account for the behaviour. I say this because the fun levels on the Cosmos are literally suicidal. It is not that the machine plays carelessly. It is literally suicidal, taking every possible opportunity to throw pieces away as fast as it can. Advance a pawn to attack a knight and it will not move the knight. Offer a low value piece to the Queen and it will capture that piece and lose the Queen. So to me that goes beyond careless and ventures into a program that is making every possible effort to lose as quickly as possible at all costs. This does not seem to describe your experiences with the fun levels on the TC2100, especially when I know you are a reasonably decent player (so careless to you would surely just mean weak in the traditional sense, not suicide chess).
Chess is like painting the Mona Lisa whilst walking through a minefield.
SirDave
Full Member
Posts: 610
Joined: Thu Sep 09, 2010 4:59 am
Location: Southern California USA

Post by SirDave »

Monsieur Plastique wrote:That actually adds to the mystery. I no longer have the Cosmos I borrowed but it seems that according to the levels for the TC2100 shown at schachcompuer.info, the level layouts are the same for TC2100, Cosmos and Expert Travel. All have the same fun levels ("e" file) and adaptive levels ("h") file.

I am wondering if Saitek somehow "tweaked" the fun levels from Cosmos onwards and this might account for the behaviour. I say this because the fun levels on the Cosmos are literally suicidal. It is not that the machine plays carelessly. It is literally suicidal, taking every possible opportunity to throw pieces away as fast as it can. Advance a pawn to attack a knight and it will not move the knight. Offer a low value piece to the Queen and it will capture that piece and lose the Queen. So to me that goes beyond careless and ventures into a program that is making every possible effort to lose as quickly as possible at all costs. This does not seem to describe your experiences with the fun levels on the TC2100, especially when I know you are a reasonably decent player (so careless to you would surely just mean weak in the traditional sense, not suicide chess).
Hi Jon,
In view of your experience, I'll have to try Fun Level 5 on my Travel Expert. I should notice pretty quickly if it plays at a suicidal level because when I played the 2100 at that level twice, it played more what I would call amateurish chess, rather than suicidal. Actually, it was rather fun to play strategies that don't always work on regular levels and have them actually work the way I planned.
david martin
Posts: 1
Joined: Sun Oct 27, 2019 3:59 pm

Post by david martin »

Maybe I'm wrong but if the H8 bug rears it's head, wouldnt it be possible to take back the offending move and press the info button to see it's analysis. This can then be entered as a move using referee mode.
User avatar
paulwise3
Senior Member
Posts: 1505
Joined: Tue Jan 06, 2015 10:56 am
Location: Eindhoven, Netherlands

Post by paulwise3 »

Hoping that I am not telling old news: apart from the H8 bug the most significant difference is the size of the opening library. The 2100 has a 30.000 ply book, where the Expert and other clones only have 6000 ply. The spare room is used for famous training games, described in the user manual.

Regards, paul
2024 Special thread: viewtopic.php?f=3&t=12741
2024 Special results and standings: https://schaakcomputers.nl/paul_w/Tourn ... 25_06.html
If I am mistaken, it must be caused by a horizon effect...
User avatar
appleshampogal
Member
Posts: 126
Joined: Sun Jan 12, 2014 8:53 am

Re:

Post by appleshampogal »

david martin wrote: Sun Oct 27, 2019 4:06 pm Maybe I'm wrong but if the H8 bug rears it's head, wouldnt it be possible to take back the offending move and press the info button to see it's analysis. This can then be entered as a move using referee mode.
I've verified this to be the case on a particular instance on my Mephisto Chess Challenger! Pressing the info button shows the sensible move the computer wants to make, in contrast to the bug move.
User avatar
Tibono2
Full Member
Posts: 695
Joined: Mon Jan 16, 2017 7:55 pm
Location: France
Contact:

Re: Saitek Mephisto Travel Expert vs Saitek Travel Champion 2100

Post by Tibono2 »

Part of the problem is the H8-bug move not only blunders, but also it is (usually, if not always) played instantly, regardless of the level set on the computer. So, the "info" display maybe suggests another move, but this alternate cannot result from a fair enough thinking time. If I remember correctly, you can see the depth reached as part of the info data; worth checking...

There is a way to identify the preferred move (apart from the H8-bug one, of course) after a decent enough thinking time, but it is rather cumbersome.
As far as I know the H8 bug affects mainly (if not only!) the ply1 move, so the idea is to get the computer to think about the very same position but at ply2; this can be done taking back two half moves (the H8-bug move by the computer, and the previous move by the opponent) and slightly updating the position to enforce the computer to plan for an obvious "must move" (ply1) leading to the expected position (where the H8-bug previously happened). You hand over the move to the computer and after the expected thinking time, you can check the ply2 displayed in the preferred variation, that's the move.
User avatar
appleshampogal
Member
Posts: 126
Joined: Sun Jan 12, 2014 8:53 am

Re: Saitek Mephisto Travel Expert vs Saitek Travel Champion 2100

Post by appleshampogal »

Tibono2 wrote: Mon Jul 11, 2022 3:23 pm Part of the problem is the H8-bug move not only blunders, but also it is (usually, if not always) played instantly, regardless of the level set on the computer. So, the "info" display maybe suggests another move, but this alternate cannot result from a fair enough thinking time. If I remember correctly, you can see the depth reached as part of the info data; worth checking...

There is a way to identify the preferred move (apart from the H8-bug one, of course) after a decent enough thinking time, but it is rather cumbersome.
As far as I know the H8 bug affects mainly (if not only!) the ply1 move, so the idea is to get the computer to think about the very same position but at ply2; this can be done taking back two half moves (the H8-bug move by the computer, and the previous move by the opponent) and slightly updating the position to enforce the computer to plan for an obvious "must move" (ply1) leading to the expected position (where the H8-bug previously happened). You hand over the move to the computer and after the expected thinking time, you can check the ply2 displayed in the preferred variation, that's the move.
I'll need to test that sometime by finding a game, a sequence of moves to enter so I can try out your idea. Although, a little painstaking, at least there is a potential workaround as you described. If you know of any games that I can input, I'd like to try it out. I'll look on my own too.

On another note... I used the test position 1 for the schachcomputer wiki that talks about the h8 bug. Interestingly, and I'm not sure if anyone else has mentioned this... But when selecting the mate search levels - it doesn't make the blunder. So that was neat to figure out as well. I wonder if there are statistics that detail the incidence of how often the h8 bug rears its head. I'm kinda worried 😫
User avatar
appleshampogal
Member
Posts: 126
Joined: Sun Jan 12, 2014 8:53 am

Re: Saitek Mephisto Travel Expert vs Saitek Travel Champion 2100

Post by appleshampogal »

I also tried the adaptive weak level 8, and the computer finds the mate in 1 instead of blundering too! Very very intriguing indeed...
Post Reply