playing first Amazons forum

20 replies. Last post: 2005-07-31

Reply to this topic Return to forum

playing first
  • Tim Shih at 2005-02-20 Am I alone in deeming that playing first holds an advantage over playing 2nd in Amazons games? :)

    If I am alone, then given an option to choose between the first and the 2nd, will you not prefer to play first?

    If I am not alone, may I hear your opinions on this issue? Should we consider adopting the komi system as Go players should, or the swap system as Hex players should?

    Look forward to hearing your opinions. In mine, I think that it is fair for the first player to give one point to the 2nd. That is, the 2nd player will win if both players have run out of squares (even if it is his turn to play).
  • Tasmanian Devil at 2005-02-20

    You are not alone. It is mentioned early in this article that the first player probably has an advantage for which one might want to introduce a compensation.

  • Tim Shih at 2005-02-20 thanks, Tasmanian Devil, but I am computer-clumsy (many-other-things clumsy, too), and failed to open your pdf file. :(

    But nice to know that I am not alone. What is your opinion aside from that article? :)
  • Tasmanian Devil at 2005-02-20

    Well, I am not a strong enough player that my personal opinion should matter. But I observe that in the LG games between the two strongest players I know of, Ludohex (aka mantonow on PBeM) and GoldenBear (aka rgsnatzke on PBeM), of which they have won two each, white has won three and black one. But as a mathematician I also know, of course, that this sample is too small to be very reliable. ;-)

    Regarding how the advantage should be carried out, I think the komi-rule is fine in both Go and Amazons (and mine and Watkins' game Pleiadis, hehe) - or a given number of passes for Black in the case of Amazons, as suggested in the article – as opposed to the swap rule of Hex and TwixT. Having to play a weak first move would distort the elegance of Go and Amazons I think.

  • Tim Shih at 2005-02-20 “ Having to play a weak first move would distort the elegance of Go and Amazons I think.”, well said, T. D. :)

    Wondering if there is anybody who may oppose to letting the black player have one point compensation?
  • Barry at 2005-02-21

    Looking simply at the games in the last 2 championship league 1 groups, it is clear that there is an advantage in playing first, but surely you would find an advantage to the first player in many games, chess and checkers amongst others.

    Would it not be better rather than to try and change the rules, simply create a tournement where each player played 2 games against each opponent, one as white and one as black. It would make for a longer tournement, but would test both players in both positions, a much fairer way, as strategies would be different if starting or replying to a move made by your opponent.

  • Tim Shih at 2005-02-22 Hi, Barry, thank you for sharing with us your bright idea. :) There is a small drawback in it, though. That is, ending up with a draw between two players will become a possibility. (Each player wins one of the two games)
  • Barry at 2005-02-22

    Tim, was is a draw in a series a bad thing, some people are better at offensive strageties and some better as defence, surely a player going first who plays better defensively, and is better at counter attack is at a disadvantage, if each player has the opportunity to play both as black and white, it is a better test.

    It is very unlikely that all the contests in a tournement will be draws, and so a winner will be declared on the majority of occasions.

    In my opinion it would be a better way of solving the first player advantage, than by manupliating the rules.

  • Tasmanian Devil at 2005-02-22

    I agree with Tim. If a game if biased in favour of the first player (or the second, for that matter) it is better to do something about it than to just let people play each other twice in each tournament. A komi-rule, ideally, doesn’t change “perfect play” at all (but in practice it may make the players play more or less offensively).

  • Barry at 2005-02-22

    Not a scientific study, but I’ve checked a few game statistics for championship leagues currently been played or in the case of go 13*13 the last championship top division,and we find for example :
    4 in a row, the 2nd player wins nearly twice as many games as the 1st player.
    Chess 1st player has an advantage.
    Reverse 1st player has an advantage.
    Go significant advantage to 1st player especially in 9*9.

    Now would you suggest evening up these game advantages as well.

  • Tasmanian Devil at 2005-02-22

    According to WZebra, perfect play in Reversi gives a draw. (It has not solved it, but with the high number of high quality games in the database, it is considered extremely unlikely that the conclusion is wrong.) So no, I don’t see any reason to change anything there. On what basis do you think that the first player has an advantage?

    In Go, there already is a komi-rule, so if one colour has an advantage, it just means that the komi value is wrong. Removing the komi rule would of course make it far easier for the first player to win.

    In most other games, the winning conditions are of such a nature that a komi-rule would not work anyway.

  • Tim Shih at 2005-02-23 See, even for Go of 19x19 size, the komi system was not invented/adopted before 1940s or 1950s, which suggests that the entire human race had been playing Go for hundreds of years as a game that strongly favors the first player.

    Why don’t we LG Amazons players establish first a one-point komi system (for the human history) to stop the unfairness of such an elegant game? :)
  • Robin at 2005-02-23

    I’ve been told that the game of Amazons was invented by computer programmers who wanted it to be a good AI-testsuite to improve the strength of Go-playing programs.

  • Tim Shih at 2005-02-24 That might have been true, Robin. :) If we extend the Amazons board to 12x12 with 5 queens (or even 14x14 with 6 queens each), the game can become as complicated as Go 19x19 is.

    Personally, I would love to see one day that Amazons is extended to 12x12 with 5 queens for each side.

    On 10x10, it seems that securing corners is the best way to win (if not the only way). If we have a 12x12 board, securing the center to win may also become a possibility. :)
    This added possibility will make the game more intriguing and attractive.
  • Tasmanian Devil at 2005-02-24

    Tim, I think that a higher number of Amazons will make the game too messy, and that four (for each player) is a suitable number even on much bigger boards. But that is just my opinion.

  • Bram Cohen at 2005-03-27

    Barry, the bias towards the second player at 4 in a row is funny, because with perfect play (on the standard smaller board) the game is a win for the first player.

  • Tim Shih at 2005-04-08 One point komi for Amazons games!! Whoever favors this minor revision, please say, “I”. :)
  • klaashaas at 2005-04-08

    Who says it’s not 2 points? Or 0.4 points?:-) I’m not sure if it’s a good idea to change the rules if you’re not sure the new rules are good.

    There is a way to find out: collect results of preferably high quality finished games and find out which komi comes the closest to an equal winning percentage for both colors. One problem... Where can we get a large set of amazons results (high quality and finished games)?

  • Tim Shih at 2005-04-08 Collecting data was what Go community did back in 1940s or 1950s. There were thousands of documented games available. (they do not need to be of high quality, however. The advantage exists for games of any quality) First, the komi was 4.5 points. Then, according to the statistics, first-move players still won more games than 2nd-move players. So, the komi was further increased to 5.5 points.

    As many of you know, the current komi 6.5 points system seems to have become the most widely-accepted one.

    My simple mind interprets the situation as follows:

    It is possible that 1-point komi is too much. But the likelihood is slim. And 1 point seems to be the minimum that is meaningful for Amazons games. So, shall we start with 1-point komi? It will take too long to wait for statistics. Imagine how many 2nd-move Go players in 19th century and earlier were buried in their graves along with the unfairness (associated with fame, honor, disgrace, etc. ) imposed on them. :)

    Another possibility is to implement “swap”, as is done in Hex.
  • wiske at 2005-07-31

    A 1 point komi would be unfair to the first player, since she would then have to win with a difference of at least 3 (three) empty spaces.

    Let’s suppose both players continue playing until no more moves can be made. See what happens:

    Game ends after 92 moves: Black wins.
    Game ends after 91 moves: Draw (due to komi rule).
    Game ends after 90 moves: Black wins.
    Game ends after 89 moves: White wins with 3 empty spaces in her territory.

    On larger boards with even sides (e.g. 12x12), I think it is Black who has the advantage, since White’s initial initiative will then be worth less than 0.5 territory points (in my opinion).

Return to forum

Reply to this topic




Include game board: [game;id:123456] or [game;id:123456;move:20] or [game;id:123456;move:20;title:some text]