BLACK WON MORE OFTEN THAN WHITE!

Unfortunately, it uses a different formula, and classic chess turns out to be position #518.

(It has a difficult interface. Paste rnbqkbnr into the first box, click the 3rd button, then click the 2nd button. The 1st button chooses a random position.)

* Betsy follows Steinitz's principles: create a strong pawn center and develop the pieces toward the center. It lacks much positional knowledge, so it loses in the middlegame, despite its relatively high search speed.

* Chispa is a solid engine, capable of stealing a point from any of the others. It typically concentrates on pawn structure, then flounders in the middlegame. It reminds me of Tarrasch's style.

* The best thing about Gothmog is not its surprising strength at lightning chess (though it does not calculate many nodes per second) but rather its courageous attacks. (Gothmog was set on "Aggressive", which according to the author should weaken it. I think it becomes stronger in Fischerandom.) At its best, it plays like Morphy, leaving pawns (or even pieces) unprotected when their capture might lead to a quick victory. It also plays amazingly well in the endgame; I've watched the evaluations swing several points.

* Its nemesis is Hermann, which defends solidly but otherwise plays a rather tepid game. Gothmog was able to pull out several draws against Hermann only because of relative endgame strength. I liken it to Lasker.

* The Baron plays the most well-rounded game: fairly defensive, energetic middlegame, good endgame (but not quite as good as Gothmog in the games I watched). It is the antidote to Hermann. I expect it to be best at slower games. It reminds me of Capablanca.

* The other surprise is Frenzee, which had won the whole tournament handily until I added Gothmog into the mix. Frenzee reminds me of Nimzowich with its prophalactic style. It often leaves opponents burdened with a bishop in the corner, trapped behind its own pawns.

Presently, Chess960 or Fischer Random Chess seems to represent the leading edge of Baseline Chess in historical evolutionary terms. Chess960 imposes an additional constraint on the starting setup, namely, that the King be flanked by the Rooks. To this effect, the FIDE rules applicable to castling are suitably extended. Chess960 allows for 960 different starting setups. However, due to symmetry, only 480 of its starting setups would require separate study and analysis. The FIDE starting position is one of them. :)

I am of the humble opinion that, what with its 480 substantively distinct variants, Chess960 is still...overkill! Instead, Baseline Chess variant enthusiasts may want to focus on...Chess18. Chess18 comprises a small subset of Chess960 variants. This subset maintains the King and Rooks on their orthodox squares. Thus, no extension to the FIDE rules applicable to castling is necessary. To boot, all 18 variants are substantively distinct and would require separate study and analysis. The FIDE starting position is still one of them. :)

Chess18 would still make "booking up" quite a formidable challenge. Chess playing computer programs could easily accommodate it. Moreover, dedicated chess position databases could be easily modified to support Chess18 opening theory research and training.

Could Chess18 opening theory research and play galvanize enough interest over the long haul so as to create a hobbyist subculture of some permanency?

oblivion95: I'm searching for more chess960 games, do you have a collection of them as pgn files ? If yes, pls. contact me (info@chess-960.org).

Using cards you can avoid having to reshuffle, by laying out black and white squares separate, making sure that one bishop goes to each, like this:

Mix a queen, two jacks(=knights), and three kings(=rook+king+rook), and split in two piles of three. Mix one Ace(=bishop) in with each pile, and lay them out in rows for white and black squares respectively, eg:

(white square row) Ks Qh Ac Kh
(black square row) Jh Kc Ah Jc

yields: NRKQBBNR

It is easy to see that this generates all FR positions with equal probability. Considering the 6 non-bishop cards only, they are all treated equally, and should have equal probability of ending anywhere. Shuffling these 6 cards will yield 6! = 720 possible sequences, but with 2 possible J-J permutations and 6 possible K-K-K permutations that does not change the FR position - thus there are 720/12=60 distinct KKKJJQ shuffles.

- Throw a D6 to get the initial file for the Queen, and then roll twice to get the initial starting files for the knights, reroll if you hit an occupied file.

- Place QNN on the indicated files and RKR on the three leftmost files available.

- Throw the dice - and reroll 5's and 6's until you get a number n in the range 1-4. Place the black-squared bishop on the n'th black square, shifting other pieces to the right to make room.

- Repeat with the white-squared bishop.

Because two places will be taken up by the bishops, there are 6 possible places the queen can go to - so throw a 6-sided die (D6).

If you think it makes it easier to visualize, you can of course place the bishops first. Then after you throw a D6 for the queen you just count from the left, that many free files to place the queen. Eg. you start off by throwing a '1' for the b sq bishop (-> a1) and a '2' for the w sq bishop (-> d1), then a '4' for the queen, putting it on (b,c,e...) f1.

Set up the chess board as usual, but replace the queen with a grasshopper, which moves straight or diagonally like a queen, but must hop over another unit (of either color) and land one square beyond. Thus, White can play 1 Gf3, threatening both 2 Gxf8 and 2 Gxa8. (Black's best response is 1...Gf6!)