Yes 700 thousand I meant. 700 million, egads!

But at least I can say I have won games against World Computer Champions in the game of chess (Deep Thought) and Checkers (Chinook) and I don't think anyone else will be able to say that ever again :)

This game is mentioned is Feng's book on Deep Blue, but he calls me a mere "expert" (I was over 2200 at the time and was a moderate master class player) and does not mention my name (cause I upset him with some of my tauntings.) He also mentions that Deep Thought just walks into an easy checkmate, which was not the case at all.

Feng was a bit of an elitist, and there was absolutely no piece of advice you could share with him that he would take. I tried to tell him that the play of Deep Thought in its Center Counter was not that strong. I said it rather plainly, just offering friendly advice, and he said something like (in broken English with a bad accent) "In your weak mind maybe, but machine is high performance."

He started to walk away when I yelled after him "I will crush your machine if it plays the Center Counter and I am given white!" About 20 people heard this, and there was about 2 hours before the last round was to start, and he was convinved of his superiority, so he see up Deep Thought to play Game in 30 with me.

At this time, the program was doing about 700,000 nodes per second, still very fast.

He typed something into his opening book to avoid some of this book play, so I got it out of a main line. Even with its 14 and 15-ply search, it saw the win of a pawn, and no way for me to capitlize on it.

You can see my moves were a tad tactically complex, the thing that drives a program crazy (having multiple hanging pieces and offering more) so it finally succummed to the momentum of the attack.

I meant it more like "computers can be beaten", there are not invincible. For example, you beating Vortex :)

Well it is still possible to outdo the silicon machines. here is my chess game against the Deep Thought computer from 1989:

Ed Trice vs. Deep Thought

You can replay this and enjoy!

1st article, about 2 Megs

2nd article, 224 K

3rd article, 284 K

George I don't know why you are equating file size with the program being the same. Even if I add 300 lines of code to the program, it won't change its size once I compile it.

This change was more subtle than that, a new "pruning mechanism" that generates fewer positions in the game tree, but also evaluates the positions better. It plays R + P vs. R + P 10,000 times better than the old Vortex program.

The version of vortex that was recently uploaded to "softlookup.com" accidentally has my super-strong version of the evaluation function, but it has the old opening book.

If anyone wants to try this one out, better get it fast!

click here to go to their download site

For those looking for "checkmate puzzles", take a look at the new Gothic Chess article on Chessville, now online at:

http://www.gothicchess.org/chessville/ches
sville_3.doc

or click here to download it as a word file.

It is still "vaporware" though... that project has not officially started yet.

I always thought people would prefer an engine to play against rather than a collection of games to browse with no opponent.

I am thinking if we do release such a version, it will be no more than $10.

OK, you pay my developer the $60/hour times the number of hours he works on it, and I will release that version for free.

Deal?

You mean just sell a gui that does not play Gothic Chess, but has a complete database of the games?

Guess I would have to call that Gothic Vortex Copper Edition :)

Yes. The Nuclear Version of Gothic Vortex will have it first, with over 4,000 games linked to the graphical user interface. As you look at any position during your game or just enter moves in human vs. human mode, it will build, on the fly, a dynamic list of every complete game that had the same position with the same side to move. You can then consult these games with one mouse click, and play through those games, with the list still being built dynamically as you do.

I think a "middle version" between Gold and Nuclear might have to be created. Perhaps a "Gold Plus" will be Gold + the database feature.

Vortex Gold had a page missing, as George reported. It is up now.

Instead of 16.Bxg4 if white plays 16.exf4, black has 16...j5! which seems to hold. White can continue to be coy and let the Archbishop hang even further, but the threat of ...jxi4 starts to take steam out of white's attack. If black gets in ...jxi4 then a follow-up ...h5 would be very strong. The reason is, if white takes the en passant capture with gxh6 e.p., black still has the knight to recapture with ...Nxh6 since white did not play Bxg4.

My version of Vortex shows 17...i6 through ply 5 (0 seconds) and switches to 17...Rg8 on the next ply. It finds a mate in 9 due to all of the check extensions at only ply 10, so it switches to 17...h5 and sticks with it. Through ply 12 it thinks it can force a repetition draw, but at ply 13 it sees white can still break through and win material, but the play is much more complex and would be hard to find.

It looks like 17...i6?? was the collosal game-ending blunder. I cannot get my version of Vortex to play this move when I let it play Black. 17...h5 offers the best shot at holding the draw, since after the white Bishop retreats to set up a check by eventually revealing the Rook, the black king has ...Ki8 and then the new retreat square ...Kh7 with the h-pawn pushed.

I will have a new version of the opening book ready by Monday which will allow the program to evade this danger, as well as play some of your outstanding moves for white in the event the positions ever come up.

Nice game George!

Email me your address Ted, I will send you all the information in the mail. My email is GothicChessInfo@aol.com.

Taking the Archbishop leads to a "Mate in 10" announcement when I let Vortex run overnight.

Vortex 1.0.7, which has not been released yet, will not take the Archbishop after only 20 seconds on my 2.8 Ghz computer.

The move 15...exf4! wins for black.

15...exf4 16. Bxg4 fxg4 and white's position is becoming exposed faster than any attack will come forth.

1. f2f4 d7d5 2. g2g3 g7g6 3. d2d4 Nb8c6 4. c2c3 Ni8h6 5. h2h3 Bh8f6?!

This is the so-called "Bishop Lunge" problem in the earlier versions of Vortex. The program gets a bonus for being able to reach more squares with its Bishops, so it deploys to the third rank almost in every case. The natural 5...Bg7 seems better.

6. Ag1h2 e7e6

Another shallow search error. 6...Nf5 is much better. The intention is to play ...Nd6 if provoked with pawn play later, perhaps hitting on a white Bishop on e4 if white traded pawns via a push to e4 and bishop recapture. Black also has a counterplay trap: 6...Nf5 7. g4? Bh4! hitting on the white Chancellor. No matter where the Chancellor moves next, 8...Ng3+ is deadly.

7. Ni1j3 Ag8e7?

I like making the exposed Black King a sore point after an e-pawn push. 7. b3 intending the flank Bishop check with 8. Ba3+ would work good for white. After 7. Nj3 Black should simply play 7...Be7 to clear off f6, prepare ...Nf5 and ...Ng7, deploy a better pawn perimeter, and only then, castle.

8. g3g4 Bf6g7?!

There is nothing wrong with white's 8th move, but black should have played 8...Bh4 for some slight counterplay.

9. Bh1f3 f7f6

On 9. g5 black gets the upper hand with 9...Nf5 10. Bf3 h5! After white's 9. Bf3, black has to be careful about castling. 9...0-0 10. g5! Ng8 11. e4! dxe4 12. Bxj7+ Kj8 13. Be2 does leave black with a tough choice. 10...Nf5 leads to the same attack for white. Black's 9...f6 is not best, as the odd-looking 9...Nj5 may do the trick! White can't provoke with 10. i4 since ...Nh4 trades knights favorably. If 10 g5 to cutoff the Queen/Archbishop combo on the dark diagonal, then ...h5 solves all of black's worries (10. i4? hxi4 11. hxi4? Rh8! 12. ixj5 Rxh2 13. Nxh2 Af5 looks odd, but black is better.)

The last really big OTB event was in Kansas City at Bartel Hall, over 3100 players.

Nuclear will probe its endgame databases in RAM. For example, imagine a R+P vs. R ending. Hit "at a leaf node" from a great distance, the evaluation routine cannot search it. The program makes basically a "0-ply guess" at the value of the position, then reports this score back down the tree.


R+P = +600, R = -500, so it will score the position as +100.

The positions might be a dead draw, so the "real score" should be 0. The program, if it is up a pawn many moves from this leaf node, should avoid swapping into it. Instead, the program will play for it, "thinking" it is up one pawn.

Now imagine your Pawn can be on any of 60 squares, your king on any of the 79 remaining, and your Rook on the 78 left from there. The opponent's Rook can be on any of the other 77 squares, and the king will be on no more than any of 76 squares, with about 64 on average due to all the checks that can take place. That would be about 1,821,980,160 positions, x 2 since either side can move gives you 3,643,960,320 positions with just R + P vs. R!

How many of those 3.6 billion positions do you think Vortex can "guess" right with a 0-ply search? Not many.

Now imagine instead of +100 for being up one pawn, Vortex could return +6320 for mate in 80 moves, or 0 for a dead draw, or -6285 for being mated in 115 moves!

This information is PERFECT, no EVALUATION has to be "guessed", and this will greatly effect the way the program will play, even from a great distance!

There are roughly another 3.6 billion Q+P vs. Q positions, and Arch + pawn vs. Arch, Arch + pawn vs. Chancellor, etc, etc.

What you have is about 300 billion positions in a huge database probed in RAM being delivered to the search engine.

There is no way to create an effective buffer with less than 8 GB of RAM, and I would even recommend 16 GB. Macintosh systems now ship with 8 GB on some of their boxes. 64-bit 8x opterons are at the same stage I believe.

This type of knowledge needs RAM, RAM, and more RAM. But it would be soooo cool.

This software is MONTHS away from completion, not YEARS. The databases are already computed and are disk resident. Writing routines to probe it in RAM has never been done, and this is the last step.

8 GB will be a very common configuration in time :) Just because the software is designed to exceed current capacities doesn't mean it will always be that way. Remember, this version of the software is still being written. When it is finished, it will be a monster.

Requirements: click here

I was a Mac programmer for decades, from October 1984 until Mac OS X made me a dinosaur in 2001. I am in the process of divorcing the gui from the engine so that it would be an easy port to OS X. If you know any Mac OS X gui developers, please let me know.

Well, you can download Gothic Vortex for free here:

Download Vortex

You can try everything you want for 30 days.

Sure, take over for white, you can mate in 268.

click here

I am working on some new graphics for the next version of Vortex to be released.

click here to see the proposed new board image.

I think it would be helpful if people post their own values on the board for the pieces. Here are the ones I put in Gothic Vortex

Pawn = 100

Knight = 275 with 20 pawns, 255 with 0 pawns.
Exact value = 250 + pawns.

Bishop = 310

Rook = 530 with 0 pawns, 440 with 20 pawns.
Exact value = 530 - (pawns * 3).

Archbishop = 690 with 20 pawns, 650 with 0 pawns.
Exact value = 650 + (pawns * 2).

Chancellor = 860, unchanging

Queen = 900 with 20 pawns, 960 with 0 pawns.
Exact value = 960 - (pawns * 3).

These values were chosen carefully, but I still think they can be changed.

For example, compare how Rook + Pawn will be traded for Bishop + Knight as the pawn count varies. At the beginning of the game, R + P = 540, and B + N is 585. You don't want to give up a B + N for R + P early on, because your two minors are developed an in play, while the Rook is most likely just sitting there after having castled.

A Bishop will not trade itself for 3 pawns unless there is some other form of compensation. Notice a Knight will always trade iteself for 3 pawns, or 2 pawns + "heavy positional damage". Some of you may have seen Vortex tossing its Knight to do this already. So far, I think this is strong behavior.

Also, 2 Knights cannot mate a lone king in the absence of pawns. So, a Rook is more valuable than 2 Knights with pawns = 0 (530 for Rook, 510 for 2 Knights).

This "simple" approach makes the program appear to have endgame intelligence. It will prefer to have a Rook over 2 Knights heading into the endgame, and it can direct this play from a distance with the properly chosen values.

If others want to post their "ideal values" I would be happy to review them.

A piece's value is not merely a function of its mobility. A board can be loaded with pawns that impede a Bishop of the same color, making it a "very bad" Bishop and a piece you would trade for a Knight immediately. One long series of pawn swaps, and now the Bishop might be poised to dominate the Knight in the ending.

As I discussed in my paper here there are values other than the "mobility/safe check" that also factor into the equation. On page 90 of this paper:



There are a few things worthy to note at this point. These are not exact ratios universally accepted by the chess-playing public. The work of Taylor in 1876 provided a foundation upon which players could experiment and adjust the relative merits of the pieces over the years. In this respect, Taylor’s equations will compute semiconditional values for the pieces (Katsenelinboigen, 1997, p. 53). In this fashion, we differentiate these data from conditional, unconditional, partially conditional, and positional values that are computed in a variety
of ways by the contemporary chess master at various stages of the game.

This is explained in more detail in the paper, and in my reference to Dr. Katsenelinboigen's book.

Oops!

I do not think a Bishop is worth a Knight and a pawn.

Thad, your post was too vague. Name names. Who was off topic, and what was posted that was off topic? We all know not to make off topic posts, so your post supplied no information.

Well I am not unhiding them to see what they said, I just suggest everyone hide them and pretend they don't even exist.

I have some people on "hide", so I can only assume there are other comments being made. As I have done my best to answer questions seriously, and I have seen some responses which don't make sense, I will leave this discussion for the time being.

There are a number of techniques we have to try and determine a piece's value. One thing I use to add more merit to the Archbishop is 'safe check' plus 'solo mate' summations. Since the Archbishop can not only deliver a check, but a mate as well, should that not increase its worth?

So, I sum over all the squares where the solo-mate occurs, add that to the sum of checks, then divide by the number of squares (multiplied by king arrangements first, of course.)

Basically, I found Archbishop is like Bishop + Knight plus 2 pawns, while Chancellor is like Rook + Knight plus about 1.25 pawns.

We can't really have "hard values" for the pieces for every scenario.

Two knights in regular chess are 300 + 300 = 600, which is greater than the Rook at 500. But what happens with 0 pawns? The Rook can mate, the 2 Knights can't!

So, you have to turn 2 Knights into "0" with 0 pawns and 0 other pieces on the board, since at best they can draw a lone king.

Clearly 2 knights are stronger than the Rook with most non-zero pawn counts on the board.

But we can debate this endlessly.

My question is, which is stronger?

Knight + Archbishop or Queen?
Archbishop + Rook or Bishop and Chancellor?
Bishop + Archbishop or Chancellor and Pawn?

These are compex issues.

You are allowed to disagree, of course.

But I gave mathematical backup.

On a larger board, a piece has to be weaker. It just make sense. How long do you think it would take a Queen to mate a King on a board of dimension 100x100?

Does not the amount of moves required to deliver mate correspond to piece "strength"?

If "no", then why does that harderst Q + K vs. K mate more quickly than the hardest R + K vs. K?

If "yes", then how can your larger board, with mates taking longer, back up your claim that a piece is stronger on it?

Put Ben Johnson, the fastest sprinter, on a 200 meter track, and he is deadly. Put him on a marathon course, is he much less likely to be a strong contender?

A larger board tends to "dilute" the strength of the pieces. As an exaggeration to make a point, imagine a 5x5 board, and a board 100x100.

Place a queen near the center of each "empty" board. On the 5x5 board, the queen has 4 horizontal, 4 vertical, and 4 diagonal (x 2) moves. It can reach 16 of the 25 squares. 16/25 means it can cover 64% of the board in one move.

You can see on the 100x100 board, with 10,000 squares, there is no way it is going to reach 6,400 (64%) of these squares. It will reach (100-1) x 4 = 396. You can see 396/10,000 is a very small fraction.

In this sense, pieces on a smaller board are stronger since they have a greater "density".

The new version of Vortex makes use of an "incremental hash table". It stores some of the "deep danger" from the previous search, so on its next move, it can find the "fail low" type of moves much sooner. Much of the play for Black is improved with this technology. It sees the danger, on average, 2 plies sooner, since "it" makes a move, "you" reply, and the "worse" positions for it saved from the previous search are hit without having to evaluate the positions or call the move generator. This makes the nodes/second creep up gradually as you play the game.

It is a very impressive win, combining strategy and tactics in the perfect way to defeat the software. Bascially, George got into a strong position, sacrificed material, then created a position where the program had so many legal moves, the game tree basically exploded, and it could not search very far at all.

A subtle way to win, and very hard to do!

Just ask the person who wants to play white to play 1. d4 2. Nh3 3. g4 4. g5 5. Cf3 6. g6 7. gxf7 8. Cxf7+ and 9. Ng5+ and the player with black makes the moves 1...Nh6 2...i6 3...Bi7 4...Nj5 5...Bj6 6...h6 7...Axf7 8...Kxf7 then you can each play whatever you want from there. If there are any questions about the notation, just play through my game with WhiteShark and make those same moves, up to white's 9th.

I would like to see players volunteer to play from this starting position:

1. d4 Nh6 2. Nh3 i6 3. g4! Bi7?! 4. g5! Nj5 5. Cf3!! Bj6
6. g6 h6 7. gxf7 Axf7 8. Cxf7+! Kxf7 9. Ng5+ !?! ∞


EdTrice vs. WhiteShark

I sacrificed a good amount of material early on to force a win from a great distance. The question is, can black find a draw from here?

Replay this game, see if it is adventerous enough for you, pair up, and play it out. Let me know if any of you are interested in doing so.

If anyone has analysis on this position, I would appreciate seeing it.

1. f4 e5?? is a loss for Black. 1. f4 e5?? 2. d4?! gives Black to move and draw. I researched one line about a year ago, another about 4 months ago. All of my analysis is in the opening book that will be distributed with Gothic Vortex 1.2, which will come out after the Computer World Championship tournament this November.

Just ban me and everything will be fine.

Volle.

exitus acta probat :)

quis custodiet ipsos custodes is who will guard the guardians

That was not the post bwildman, he used the word sh*t and p*ss in the post in question.

bwildman left ugh's post with the curses aimed at me sit on this board in plain view for 2 hours 45 minutes. bwildman deleted MY post, which was a fully delineated, well worded response to a concern Caissus had about international law, in about 7 minutes.

The reason bwildman is under examination is because of this. Curses are OK as long as it's his friend, but any response from me over something serious is clearly in need of being deleted immediately.

The poster of the text at http://www.symmetryperfect.com/gothic click here is Derk Nalls. He lives in Ardmore, Oklahoma, and he is now being served papers as I am suing him.

Let's see how long his website displays the slander, shall we?

I like this game, anyone else want to play?

If you read his own Wikipedia entry:

http://en.wikipedia.org/wiki/User%3AOme
gaMan or click here you will see:

I must confess ...

I am an ex-editor of "Chess Variants" at the Google directory. I was blacklisted for flaming a senior editor of the games category, I infer. To be sure, this person [whom I will only refer to as "she"] did not know a damned thing about chess variants. Yes, they were hanging a "help wanted" sign for a reason. Yet astonishingly, she would dare to arrogantly, relentlessly countermand most of my work for extremely petty (and often, erroneous) or unstated reasons. More importantly, she did some things which I strictly regard as socially unconscionable.

...



which is then followed by a full-page rant against the Senior Editors of the ChessVariants.org website.

I know who this person is, I just got him removed from that place, so he is carrying out his little "posting war" somewhere else. It does not matter, his entry has now been removed from Wikipedia.