Mind blowing moves made by the Star Opal

[Monsieur Plastique]

In finishing up a long study of this chess program (found in the Novag Star Opal, Star Aquamarine and Carnelian II), it made two moves in two games that really take the cake. I mean moves so astonishingly bad, that you would think the program was an experimental amateur one from the 1970s.

Here they are. Be prepared to be stunned. Perhaps this is an interesting test. Does anyone own any other dedicated chess computer that can make such horrific blunders as indicated below on their slow tournament (3 minutes) level?


HORRIFIC BLUNDER # 1.

After 10 minutes thought, Star Opal as black plays ....Qxc4 here, losing instantly to Bf1

[fen]r4rk1/pp2bppp/3p1n2/2pPp3/1qPn1P2/2N3P1/P2B1PBP/1R1QR1K1 b - - 0 1[/fen]

And before you ask, it makes this move on every level up to 10 minutes per move. It needs 15 minutes thinking time before it realises it cannot take the c pawn with it's Queen.



HORRIFIC BLUNDER # 2

In the position below, Star Opal as white thought for 10 minutes then came up with the scintillating Nf3??, instantly losing the knight or suffering mate.

[fen]rnbqk1nr/ppp2ppp/3b4/8/8/8/PPPPP1PP/RNBQKBNR w KQkq - 0 4[/fen]


There is no doubt about it. This program has very serious issues and Novag have been packaging it in various models for years now. I could give you many more examples, but these are the worst I have come across.

If you want a competant Novag program from their current range, Citrine and Obsidian are the only choice.

[The Wizard]

Hi,
Horrific Blunder 2 ... Is that not just the basic position from the `Froms Gambit` against the Birds Opening?

1. f4 e5 2. fxe5 d6 3. exd6 Bxd6 and then 4.Nf3 is the standard move, black tends to reply with g5 or Nf6 (there are of course other moves) .. I cannot see where white instantly loses the knight or suffers mate. The Knight covers h4 ... OR did you actually make a typo and meant 4.Nc3 is played by the comp of course then 4... Qh4+ 5.g3 Bxg3+ wins

Regards

[Monsieur Plastique]

Hi,
Horrific Blunder 2 ... Is that not just the basic position from the `Froms Gambit` against the Birds Opening?

1. f4 e5 2. fxe5 d6 3. exd6 Bxd6 and then 4.Nf3 is the standard move, black tends to reply with g5 or Nf6 (there are of course other moves) .. I cannot see where white instantly loses the knight or suffers mate. The Knight covers h4 ... OR did you actually make a typo and meant 4.Nc3 is played by the comp of course then 4... Qh4+ 5.g3 Bxg3+ wins

Regards

Sorry, I meant Nh3?? I will edit the original post.

[Monsieur Plastique]

Too late to edit it, but the correction is clear from your post above.

[The Wizard]

Hi,
Yes 4.Nh3 is horrific lol although 4.Nc3 is worse since it is a forced mate whereas 4.Nh3 Bxh3 5.g3 and others is not ... It makes you wonder if your electricity would be better spent making a cup of tea as these chess comps seem to be dire. I wonder if it is some kind of bug? as these comps must be able to go 10ply with 10 minutes thought, I think my old commodore 64 could manage that

Regards
Tony

[Monsieur Plastique]

I have never really gotton to the bottom of why this program behaves the way it does. Sometimes it can play a really fun game with perhaps the worst thing being a dubious pawn sacrifice. Then just as you think you have found a level that "works", it starts making these horrific blunders again. The wierdest thing about that From Gambit is that it sometimes plays the correct Nf3 and will usually only play Nh3 when it is given much more time to consider it's response!

But that Queen move in # 1 for example, it makes that on every single level except for 15 minutes or higher for average time and 10 minutes or higher for fixed time. So it was a good test to confirm the same program is in the Star Aquamarine.

There was a third example just as bad from the weekend, but sadly I accidently hit the "new game" button on Chess Genius and I lost the record of the game before saving it to the database. But it went along the lines that the opponent (Constellation) checked the Opal's King, forcing it behind the Opal's Queen (nowhere else for the King to go) upon which the Constellation immediately pins the Queen to the King with it's Rook. So instant loss of the Opal's Queen. Actually, blunders with it's Queen plus a particular issue relating to pin and skewer avoidance seems to be a big problem with this program.

I have tried the average time levels, running it with a 24 Mhz crystal so it is three times faster, the fixed time levels and even the extended ply levels.

The only conclusion I have been able to draw is that it needs at least 20 minutes per move (at the stock speed of 8 Mhz) to avoid these terrible blunders, but even then, that does not stop it from making some extremely dubious pawn sacrifices which generally cost it the game. Actualy, I think it makes that Nh3?? move even with 20 minutes thinking time!!

To give you an example of how inconsistent the program is, my brother played a match between an 8 Mhz Star Opal and the Saitek Express 16K at slow tournament time controls. The Opal got 8.5 out of 20. Not bad at all.

I then played the Star Opal against the Saitek Cavalier (which is meant to be almost identical to the Express 16K) and Star Opal did not win a single game. That was when I started putting in faster crystals. At 24 Mhz it still only won 1 game out of 10. I played it against my Mephisto Maestro (handheld) and it could not beat that either. And it's record against me is one draw and 29 losses across about 30 games.

I have just completed an informal match between the original Novag Constellation from 1983 versus Star Opal, with the Star Opal running at 24 Mhz on level White G6 (just over 3 minutesd per move, or 10 minutes per move on the stock speed Star Opal), versus Constellation on a mere 30 seconds per move. This - by hardware equivalents, gives the Star Opal an 18 fold speed advantage, since the 24 Mhz Star Opal is at least equivalent to a 6 Mhz 6502 - and the Constellation was not only running at 2 Mhz, but also it had only one sixth the time per move that Star Opal had!

In the above match the Star Opal - despite all the massive hardware / time advantages - only scored 2 points out of 10. And yet it is reported that the same author (Kittinger) wrote both programs and the Star Opal is a couple of decades younger!

All this from a program that I have seen advertised at 2200 USCF ELO playing strength at two US based shops!

This is actually why I have trouble accepting that Kittinger even had a hand in this program, but by the same token, I understand the Star Opal is a derivative (improvement?) of the original Mentor 16 program. And my understanding is that Kittinger was Novag's only programmer from 1982 through to 2003 which of course covers all the relevant years.

[Fernando]

In finishing up a long study of this chess program (found in the Novag Star Opal, Star Aquamarine and Carnelian II), it made two moves in two games that really take the cake. I mean moves so astonishingly bad, that you would think the program was an experimental amateur one from the 1970s.

Here they are. Be prepared to be stunned. Perhaps this is an interesting test. Does anyone own any other dedicated chess computer that can make such horrific blunders as indicated below on their slow tournament (3 minutes) level?


HORRIFIC BLUNDER # 1.

After 10 minutes thought, Star Opal as black plays ....Qxc4 here, losing instantly to Bf1

[fen]r4rk1/pp2bppp/3p1n2/2pPp3/1qPn1P2/2N3P1/P2B1PBP/1R1QR1K1 b - - 0 1[/fen]

And before you ask, it makes this move on every level up to 10 minutes per move. It needs 15 minutes thinking time before it realises it cannot take the c pawn with it's Queen.



HORRIFIC BLUNDER # 2

In the position below, Star Opal as white thought for 10 minutes then came up with the scintillating Nf3??, instantly losing the knight or suffering mate.

[fen]rnbqk1nr/ppp2ppp/3b4/8/8/8/PPPPP1PP/RNBQKBNR w KQkq - 0 4[/fen]


There is no doubt about it. This program has very serious issues and Novag have been packaging it in various models for years now. I could give you many more examples, but these are the worst I have come across.

If you want a competant Novag program from their current range, Citrine and Obsidian are the only choice.

As much as I could not find the blunder of Nf3 in the second example, in fact because there is not , I thought i was a worst player than Opal..... Next time you make me suffer that way i will sue you,.
Fern

[Cubeman]

I am glad I scroll down when confused about a particular issue.Because in the 2nd example I was scratching my head how the move Nf3 was blunder.

[Steve B]

Too late to edit it, but the correction is clear from your post above.

[fen]rnbqk1nr/ppp2ppp/3b4/8/8/8/PPPPP1PP/RNBQKBNR w KQkq - 0 4[/fen]

Hi Jon
i set up the position on my "off the shelf" Star Opal
on the 3minute per move level (Level wf6) ..SO immediately chooses Nf3(not Nh3) and sticks with it for the entire 3 minutes

i wonder did you perhaps fiddle around with your Star Opal ..over clocking it or modifying it in some way??

Concerned Regards
Steve

[The Wizard]

Too late to edit it, but the correction is clear from your post above.

[fen]rnbqk1nr/ppp2ppp/3b4/8/8/8/PPPPP1PP/RNBQKBNR w KQkq - 0 4[/fen]

Hi Jon
i set up the position on my "off the shelf" Star Opal
on the 10 minute per move level (Level wg6) ..SO immediately chooses Nf3(not Nh3) and sticks with it for the entire 10 minutes

i wonder did you perhaps fiddle around with your Star Opal ..over clocking it or modifying it in some way??

Concerned Regards
Steve

Hi Steve,
What does your Opal play in the Horrific blunder 1 example?

Regards

[Steve B]

Too late to edit it, but the correction is clear from your post above.

[fen]rnbqk1nr/ppp2ppp/3b4/8/8/8/PPPPP1PP/RNBQKBNR w KQkq - 0 4[/fen]

Hi Jon
i set up the position on my "off the shelf" Star Opal
on the 10 minute per move level (Level wg6) ..SO immediately chooses Nf3(not Nh3) and sticks with it for the entire 10 minutes

i wonder did you perhaps fiddle around with your Star Opal ..over clocking it or modifying it in some way??

Concerned Regards
Steve

Hi Steve,
What does your Opal play in the Horrific blunder 1 example?

Regards

hi Wizard
i edited my first post
i used the 3 min per move level (not 10 min)
will try the second position later
staying with this position i am now bumbing up the level to 5 minutes per move(level wg1) ..
Stay Tuned Regards
Steve

[Steve B]

Too late to edit it, but the correction is clear from your post above.

[fen]rnbqk1nr/ppp2ppp/3b4/8/8/8/PPPPP1PP/RNBQKBNR w KQkq - 0 4[/fen]

Hi Jon
i set up the position on my "off the shelf" Star Opal
on the 10 minute per move level (Level wg6) ..SO immediately chooses Nf3(not Nh3) and sticks with it for the entire 10 minutes

i wonder did you perhaps fiddle around with your Star Opal ..over clocking it or modifying it in some way??

Concerned Regards
Steve

Hi Steve,
What does your Opal play in the Horrific blunder 1 example?

Regards

hi Wizard
i edited my first post
i used the 3 min per move level (not 10 min)
will try the second position later
staying with this position i am now bumping up the level to 5 minutes per move(level wg1) ..
Stay Tuned Regards
Steve

At 5 min per move the (unmodified ..8 Mhz ..out of the box )"Shining" Star Opal ..again immediately chooses Nf3 ..sticks with it for the entire 5 minutes and plays it

Hmmm Regards
Steve

[Monsieur Plastique]

Hi Steve,

I would not have been so amiss as to not verify these findings on a stock untouched 8 Mhz version. Both versions play precisely the same, albeit the 24 Mhz version is 3 times as fast.

As I mentioned in the original post, it takes 10 minutes thought for the stock 8 Mhz Star Opal to play Nh3, so you will need to set it to levels WG6 or higher, or WC6 or higher. Additionally, it plays Nf3 about 30% of the time on these levels and Nh3 about 70% of the time, so you still might get an Nf3. But it seems to think Nh3 is better, since the longer the thinkig time, the higher the proportion of Nh3 responses.

But my point is that this program has no consistency. Half the time it plays OK chess, then the next time it inexplaicably plays horrible blunders like these.

At the fast levels it does play Nf3 most of time, but plays Nh3 about 10% of the time.

As for the first test position, you will find your 8 Mhz Star Opal will play
Qxc4 on all fixed and average time levels up to WG6 for average and WC4 for fixed. From WC5 upwards or WG7 upwards, it will find the saving Qa5 instead.

As I say, I have a huge number of these sorts of Opal blunders - these are just two of the worst examples.

[Steve B]

Far More importantly Jon..
i have finished the CCC forum game which ended in a draw

you mentioned a few games on PlayChess against My Scisys Mark VI(Philidor)?
i can play this Saturday Or Sunday
i imagine Harvey or Rob will grant us safe harbour if we play in the Engine room

let me know if any of those two days are good for you and what time(server time) you can play

Challenge Up Regards
Steve

[Monsieur Plastique]

Hi Steve,

I am thinking of a best of three gamesmatch between the SciSys Mark V and my 24 Mhz Star Opal, both sides 30 seconds per move.

It will probably be a complete walk-over for the SciSys machine, but I am at the point now where I laugh rather than cry At any rate, a clumsy mixed martial arts brawl between one of the first ever decent dedicated machines and (likely) one of the very last to ever be produced should be amusing to watch - if only to show how we have gone absolutely nowhere in commercial terms over the last three decades

Anyway, my proposal would be that I run Star Opal at the fixed time level and will force it to move immediately if there is only one legal move (otherwise it will think for the full 30 seconds, despite what the user manual says).

I can't recall how the Mark V levels work, but if you have a casual 30 second per move level, that would be good, since otherwise the Opal will lose on time due to my and yours operator time.

Also, I will resign on behalf of Opal if I feel it extremely unlikely that it will draw or win from a lost position. Although I always play my own games to mate, this is a casual encounter, so I am happy to apply that rule to expedite things. As for Mark V, it can resign on it's own

As for Chessbase server time, two time windows would be OK. They are between 11.30 PM and 3.30 AM Chessbase server time, or 4.30 AM and 8.30 AM Chessbase server time. If those times are difficult then I can try and figure out something else.

As for access, I have premium membership but my account is only at pawn status because I only watch games with Chessbase, having no interest in playing them. So I don't think I have any access to their engine room by default unfortunately.