Basically all WinBoard engines tied for place 2-5, the sum of opponent scores determining the ranking. (Which is basically a lottery.)
The tourney was 6 round Swiss; 1/128 Rigan only lost to Lima, and beat TJshogi and Nebiyu. Nebiyu beat Lima and Shokidoki, but lost to 1/128 Rigan and TJshogi. Lima beat Shokidoki and 1/128 Rigan, but lost to Nebiyu and the number 6. Shokidoki lost to Nebiyu and Lima, but beat TJshogi.
It seems Nebiyu had the toughest opponents, meeting all 4 programs from the top-5.
That is great! Indeed most engines seem to be of equal strength, and the fact that Rigan lost a game seem to also indicate that it is not far off the pack. I wonder how much stronger a minishogi version of the current state of the art of standard shogi (bonanza?).
I'm not an expert in japanese yet, but bear with me. The kanji in number 9 is 功(こう in hiragana) which means "success". I think the whole name means something like "success this time" but I'm not completely sure.
PS. In romaji the name was koro kou.
Functional programming combines the flexibility and power of abstract mathematics with the intuitive clarity of abstract mathematics. https://github.com/mAarnos
@Mikko: thanks for the translation. The problem is that the results were posted on the UEC minerva website as a jpg image. So there was no way to offer the names of the programs to Google translate.
This problem will solve itself when they start posting the game records on their website.
@Daniel: Indeed, all engines are very close. Last year and this summer during the ICGA Olympiad Shokidoki beat 1/128 Rigan. (Of course I don't know whether this is a spectacularly improved version compared to the Olympiad, but it seems unlikely, as the program already exists a long time.)
The low number of games basically make this a lottery (especially since draws are not possible). Nebiyu was unlucky in this: it had by far the toughest opponents, and in particular played against one more strong opponent than 1/128 Rigan. Of course the games themselves are also a kind of lottery; even a 100 Elo stronger engine scores only 65%, and can thus easily lose.
On Dec 6 Shokidoki will participate in the TAAI mini-Shogi competition in Taiwan, and presumably will meet 1/128 Rigan there again. (This tourney they managed to avoid each other, despite the fact that they both lost their first game. I don't really understand the pairings of the first two rounds.)
Basically all WinBoard engines tied for place 2-5, the sum of opponent scores determining the ranking. (Which is basically a lottery.)
The tourney was 6 round Swiss; 1/128 Rigan only lost to Lima, and beat TJshogi and Nebiyu. Nebiyu beat Lima and Shokidoki, but lost to 1/128 Rigan and TJshogi. Lima beat Shokidoki and 1/128 Rigan, but lost to Nebiyu and the number 6. Shokidoki lost to Nebiyu and Lima, but beat TJshogi.
It seems Nebiyu had the toughest opponents, meeting all 4 programs from the top-5.
That is already a nice result overall for the winboard engines gang , unfortunately they are verifying some game results, so perhaps the standings could be changed.
I have suggested to Takeshi Ito that next year WinBoard could broadcast the games live. So at least the game records of the WinBoard engines would all be available immediately from my website. With so many WinBoard engines participating, that should cover an appreciable fraction of the tournament.
It seems they inverted the results of Shokidoki vs Lima, in the latest cross table they mailed around, and are also complaining about the game 1/128 Rigan vs Lima, which they are replaying. (It was listed as a win for Lima before.)
My guess is they got confused by Lima's resign message: "(Black Resigns} 1-0". To the Japanese "black" means "sente", i.e. the side that moved first (thus white). So when Lima printed "Black Resigns" they must have accepted it as a resign on behalf of the opponent!
I tried to remove confusion about this black/white business by supplying them a WinBoard that was set to language "Shoginese", which replaces White by Sente and Black by Gote in all WinBoard messages. But in this case the result comment was coming from the engine.
I guess it would be better to also subject the result message of the engine to WinBoard's translation routine before displaying it on screen. Although there is no way of knowing what an engine could print, at least it could capture the most common messages. (In this case it would not have automatically recognized it, as the "Resign" was spelled with a capital, while WinBoard would itself print it with lower case, and the language file thus only contained a translation for the latter. But is could of course also be added with upper case to the shoginese.lng file.
Consequence seems to be that 1/128 Rigan won all its games, and Shokidoki only lost to Nebiyu. However, due to the wrong game results completely messing up the Swiss pairing, they will play all remaining games, to make it a round-robin. Which means that Shokidoki still has to play 1/128 Rigan.
For Shogi engines it should furthermore be avoided to ever print "black" or "white". It is good enough to just print "resign" as result comment; the 1-0 or 0-1 provides all necessary information on who resigned, in an unambiguous way. Same with checkmate (although WinBoard would usually pre-empt the engines in declaring that).
Do you have the final standings ? It seems I am getting a prize anyway, just been asked to provide information. I will probably get third place if no RR are played, since shokidoki is getting one more win. But this feels like a real tournament, with something to remember it with, a prize. Take note organizers.
, unfortunately they are verifying some game results, so perhaps the standings could be changed.
Organization there seems a bit chaotic.