How many moves?

[mohsart]

I read that a game with 20 moves or less is not going to affect one's rank and that this was standard for go servers.
I checked the KGS help pages (took me a while) and as I understand it KGS has a limit of 10 moves, but with the exception that a player can claim the win.

Ranked Game (R)
A ranked game counts towards players' ratings. Both players must be registered accounts and not guests, the game must be on a 19×19 board, and the handicap must be six stones or less. Pro accounts can not play rated games on KGS. In general, a game must have at least 10 moves before it is saved by KGS and used by the rating system, however, a ranked game with even less than 10 moves WILL count either if it is saved by the person who plays as white OR if the game was created by automatch. All ranked automatch games are used by the rating system. In both rated and free games, players cannot see comments made by observers until it has ended, and players cannot edit the game until the game is over. A "rated game" is the same thing as a "ranked game".

While this is not entirely bad, I see no reason for a game that lasts, say, three moves or longer not to be counted as a "real" game. You will notice that hc is wrong, you ended up with the wrong opponent or that you messed up with who's white/black well before three moves are made.

Thoughts?

/Mats

[Phelan]

I also think that ten moves is already an investment in the game. Maybe not much of a time investment, but you've already devoted time to think about your moves and reaction moves by the opponent. 3 moves would make more sense. I'm not that keen on changing the system now though, since that never really bothered me much.

[entropi]

Long time ago, maybe 10 years or so, I was working as a programmer on one of early gaming sites, games like chess, backgammon, checkers, etc. We didn't have Go but had the same limit for ranked games because:

Some rank/rating freaks were going to internet cafes, opening plural accounts and playing fake games against each other to increase their ranks artificially. I don't know if the reason of KGS 10 moves rule is the same, but at those times we had specifically implemented that feature for making this kind of cheating a bit more difficult.

[mohsart]

That's a very interesting aspect that I didn't think of at all.
I wonder, though, if it's that useful for go. I mean, just randomly clicking the board will get you to 20 moves quite fast.

/Mats

[Kaya.gs]

Im honestly not sure which move-number is the best or why. I have no feeling whatsoever for the situation, because i've never in my life felt that a game was lost after 20 moves, even with a terrible start.
Changing that move number is absolutely trivial, to digits in the code, but which number is best and why ? i don't know .

Maybe you can present a case mohsart. I do think 3 moves is too little. You could notice something wrong after 2 moves, becasue you started too fast. In any case, what is the meaning of a game won/lost at move 4? It doesnt say much.

I believe the same at move 20, altough some joseki-mistake or fuseki mistakes is quite likely to have happened already.

[daniel_the_smith]

Clearly, 19 is the correct number to use here...

[HermanHiddema]

(;SZ[19]FF[3]
PW[Kudo Norio]
WR[9d]
PB[Ono Nobuyuki]
BR[6d]
EV[53rd Honinbo]
RO[2nd Preliminary]
DT[1996-11-21]
PC[Nihon Ki-in]
KM[5.5]
RE[W+R]
GC[Shortest game in Japan]
US[GoGoD95]
;B[qd];W[dd];B[cp];W[pp];B[fc];W[id];B[oc];W[fd];B[gd];W[fe];B[dc];W[cc]
;B[eb];W[gc];B[hc];W[gb];B[hd];W[fb];B[ec];W[ed]
)

[mohsart]

Well, I really don't have any strong opinions, but what I was thinking of was not that the game would be lost after 3 moves but that some people resign because they don't want to play.
Maybe they only want to play as black and resigns all games as white, for example.
(Edit: bad example, since they would resign before the second move.)
Or say for example that they make a REALLY small mistake and resign after 10 moves. So they don't lose rank, no "cost", arguably they COULD have won the game in the end but this method could actually be used to boost the rank.

/Mats

[Marcus]

Personally, I think making rated games count no matter what is the correct way to go. There will always be those who seek to cheat and artificially boost (or drop) their rating. I don't believe that those people are more than a very small group.

Well, I really don't have any strong opinions, but what I was thinking of was not that the game would be lost after 3 moves but that some people resign because they don't want to play.
Maybe they only want to play as black and resigns all games as white, for example.
(Edit: bad example, since they would resign before the second move.)
Or say for example that they make a REALLY small mistake and resign after 10 moves. So they don't lose rank, no "cost", arguably they COULD have won the game in the end but this method could actually be used to boost the rank.

/Mats

A better example is a player who refuses to play against a certain opening, or only wants to play a certain opening, who resigns if the first few moves aren't according to his plan.

I must re-iterate, though, that I think this is really NOT a big issue, and doesn't warrant so much attention. YMMV

[Suji]

Personally, I think making rated games count no matter what is the correct way to go. There will always be those who seek to cheat and artificially boost (or drop) their rating. I don't believe that those people are more than a very small group.

Well, I really don't have any strong opinions, but what I was thinking of was not that the game would be lost after 3 moves but that some people resign because they don't want to play.
Maybe they only want to play as black and resigns all games as white, for example.
(Edit: bad example, since they would resign before the second move.)
Or say for example that they make a REALLY small mistake and resign after 10 moves. So they don't lose rank, no "cost", arguably they COULD have won the game in the end but this method could actually be used to boost the rank.

/Mats

A better example is a player who refuses to play against a certain opening, or only wants to play a certain opening, who resigns if the first few moves aren't according to his plan.

I must re-iterate, though, that I think this is really NOT a big issue, and doesn't warrant so much attention. YMMV

Personally, I agree. Rated games should count regardless.

[Mef]

Personally, I think making rated games count no matter what is the correct way to go. There will always be those who seek to cheat and artificially boost (or drop) their rating. I don't believe that those people are more than a very small group.

Well, I really don't have any strong opinions, but what I was thinking of was not that the game would be lost after 3 moves but that some people resign because they don't want to play.
Maybe they only want to play as black and resigns all games as white, for example.
(Edit: bad example, since they would resign before the second move.)
Or say for example that they make a REALLY small mistake and resign after 10 moves. So they don't lose rank, no "cost", arguably they COULD have won the game in the end but this method could actually be used to boost the rank.

/Mats

A better example is a player who refuses to play against a certain opening, or only wants to play a certain opening, who resigns if the first few moves aren't according to his plan.

I must re-iterate, though, that I think this is really NOT a big issue, and doesn't warrant so much attention. YMMV

From a server perspective, I think it depends on what you intend for ranks (And likewise ranked games) to be. If your goal is to make them as accurate a predictor of game results as you can (note: for many this isn't the purpose, but for some it is) then it is a signal to noise issue -- A game with only 1 move provides virtually no useful information and would be random noise if a result was assigned...at some point it's more likely the game is meaningful information for predicting future results (like Herman's example). I think having the buffer of a few moves is nice, of course I think we're all in agreement picking the right spot is pretty arbitrary and not worth stressing over.


Amusingly enough -- While I think your example of a person who doesn't want to play a certain opening is plausible (someone who hates taking white against the Chinese for instance), from my experience people who are convinced they don't like certain openings, etc don't always have their feelings validated by actual game results.