Author |
Topic: Postal Tournament (Read 6761 times) |
|
omar
Forum Guru
Arimaa player #2
Gender: 
Posts: 1003
|
 |
Postal Tournament
« on: Dec 27th, 2004, 2:04am » |
 Quote  Modify
|
There seems to be quite a bit of interest in having a postal tournament. I've discussed this with several people in the chat room and we tossed around some ideas. I've written up some pages for it. Some things I've written are a little different than what was discussed in the chat room. Would like to get some discussion on the details of this before we go forward with it.
http://arimaa.com/arimaa/tourn/postal/2005/
Omar
|
|
 IP Logged |
|
|
|
Fritzlein
Forum Guru
Arimaa player #706
Gender:
Posts: 5928
|
|
Re: Postal Tournament
« Reply #1 on: Dec 27th, 2004, 12:39pm » |
Quote Modify
|
Omar,
This looks fantastic. I think the time control of 24 hours per move and 5 days in reserve is fine, although I can see an argument for a 10 day maximum reserve. (You wouldn't have to increase the starting reserve, so it wouldn't lengthen the tournament any to increase the maximum reserve.)
The 180 day cutoff is fine. That insures a minimum of 85 moves, even counting the initial reserve. There has never yet been a human vs. human game lasting 85 moves. If this happens to be the first time, we'll know that future postal tournaments, but I don't expect it to happen. Postal chess games don't necessarily go more moves than regular chess games, and until proven otherwise we can assume the same for postal Arimaa.
I like having a $10 entry fee. One of my fears was that people would sign up without taking it at all seriously, and then abandon all their games after a week or two. It is still possible that this will happen, but less likely when there's an entry fee.
Here's a suggestion for the prize money to take commitment one notch higher: rather than giving it back for first, second, and third, return it for committed play. If you win a game you get 100 points and if you lose you get the number of moves the game lasted (to a maximum of 90). When it is over, the entry money is given back proportional to the points accumulated. My idea is that an entry fee is mostly to ensure commitment, not to create a prize fund for top players to win. Giving back the entry fee in the way I suggest will encourage people to fight it out, and not resign or abandon games.
Anyway, where do I sign up?
P.S. This reminds me to ask for a feature for the server. When you click on "Current Games" it would be nice if each game had the move number and player to move displayed by it, e.g. "S23" to indicate that it is Silver's turn to move on the 23rd move. That way when I log in to check on my postal games, I won't have to open each one to see whether my opponent has moved or not.
|
| « Last Edit: Dec 27th, 2004, 12:44pm by Fritzlein » |
IP Logged |
|
|
|
omar
Forum Guru
Arimaa player #2
Gender:
Posts: 1003
|
|
Re: Postal Tournament
« Reply #2 on: Dec 27th, 2004, 4:58pm » |
Quote Modify
|
Thanks for the feedback Karl. I really like the suggestion for giving back the entry fee based on committed play. Lets see what others think.
I'll add the sign up page once we finalize the rules.
I will be changing the 'Current Games' so that it only shows interactive games and there will be a seperate page for postal games. I will add the move status to the postal games.
I can also make the max reserve time longer. Lets see what others think.
|
|
IP Logged |
|
|
|
MrBrain
Forum Guru
Arimaa player #344
Gender:
Posts: 148
|
|
Re: Postal Tournament
« Reply #3 on: Dec 27th, 2004, 5:52pm » |
Quote Modify
|
I think 7 days reserve would be the most that anyone will need. For me, 7 would be better than 5. I don't think starting with the maximum reserve is necessary.
As far as the 180 day limit, I like it. But rather than looking at it as an 85-move-ish limit, let's use the postal tournament to implement the idea of not using the scoring function to decide the game, but rather who's used less time. Then you won't really limit the number of moves at all, but rather how long you can continue to think about each move in a very long game.
|
|
IP Logged |
|
|
|
jdb
Forum Guru
Arimaa player #214
Gender:
Posts: 682
|
|
Re: Postal Tournament
« Reply #4 on: Dec 27th, 2004, 6:36pm » |
Quote Modify
|
I suggest the max reserve time be somewhere in the 7-10 day range. The tourny could take 6 months. This would give enough time for someone to take a break. (ie vacation or business trip)
I like the idea of scoring based on moves. How about rewarding a quicker win. Say, the winner gets 200 pts -number of moves. The loser gets a point for each move. If the game is not over when the time limit is elapsed, BOTH players lose. For scoring purposes, the move limit is capped at 90. (but the game can of course continue!)
|
|
IP Logged |
|
|
|
Fritzlein
Forum Guru
Arimaa player #706
Gender:
Posts: 5928
|
|
Re: Postal Tournament
« Reply #5 on: Dec 27th, 2004, 10:37pm » |
Quote Modify
|
I'm not sure how much I like the idea of rewarding quicker wins more than slower wins if the scoring system is primarily to reward committed play. When there was a prediction contest going on (with money on the line) it made sense to reward quick wins, so that the players would have a disincentive to mess with the prediction contest. In the World Championship, if I had gotten the same number of prediction points for winning quickly or slowly, I might have purposely dragged out a win to ninety moves just to stop other people from getting so many prediction points.
In the postal tourney, where there is no prediction contest we have to protect the integrity of, I'm not sure that we need to reward quicker wins more than slower ones. Besides, if a thirty-move win would score 170 to 30, it gives the losing player pretty poor compensation for being committed. I say 100 to 30 is quite enough of a swing, and maybe too much. On second thought I might prefer giving the winner a flat 75 and the loser the number of moves, capped at 65.
As for Mr. Brain's suggestion of deciding a time-cutoff game on the basis of who used less total time, I don't like it much, but I favor it over the present scoring system. In fact, I'm in favor of almost all the proposed alternatives to the present scoring system. The criterion of using less total time might have the desirable effect of bringing the game to a conclusion on the board before the 180 days are over: if it looks like the time cutoff might be in play, then the players will each try to move faster, and by the same token if one player is consistently moving in less than a day to get ahead on time early, that will mean that more than 90 moves get played in the 180 days, also tending to bring a conclusion on the board. The main disadvantage I see in the new rule compared to the present rule is that it requires extra programming on Omar's part to display total time used, whereas the wacky score function is good to go at present.
|
|
IP Logged |
|
|
|
Paul
Forum Junior Member
Arimaa player #888
Gender:
Posts: 8
|
|
Re: Postal Tournament
« Reply #6 on: Dec 28th, 2004, 2:13am » |
Quote Modify
|
I don't really like the idea of rewarding long games, for two reasons :
- using Fritzlein suggestion, one can imagine winning the tournament with 10 losses in 65+ moves. 650 points will not be easy to beat
- this method of scoring risks to conduct to very conservative and defensive games.
So I propose to keep things simple : 1 point for a win, and 1/2 for a draw.
Also, I propose that colours are determined to insure each player to have an egal number of time silver and gold. I don't see the point of letting the lowest rated player choose his colour.
For the prize money, do we really need one ? Why not simply ask for an entry fee refundable if all games are played ? I seriously doubt any player will participate for the prize.
Paul
|
|
IP Logged |
|
|
|
MrBrain
Forum Guru
Arimaa player #344
Gender:
Posts: 148
|
|
Re: Postal Tournament
« Reply #7 on: Dec 28th, 2004, 9:37am » |
Quote Modify
|
I think people are really missing the point of deciding a game by who's used less time. All it really does is to put a limit on the total time you can think for a game. This solves all the problems that we had for the last championship, and frankly the main problem with the arimaa time controls.
I complain, "Why can't we have 2 minutes a move?". Answer: "Because we need to have people be able to fit the game into their schedule." If we did 2 minutes a move, then to accomodate a really long game (say 100 moves) we need to allocate 7 hours. So why can't we have 2 minutes a move with a 5 hour limitation? Reason: The awful scoring function will kick in.
Now imagine that the game is decided on who uses less time. What is really the end result? It is that nobody will dare use more than 2.5 hours (150 minutes) for all moves, regardless of how long the game goes. In effect, we preserve the 2 minutes per move early in the game, and then when the most fantastic game of arimaa ever happens (goes 85 moves perhaps), the players just start moving quicker toward the end, and we have a real winner. In the rare case that someone continues to use 2 minutes per move, they will effectively lose on time (because they used more than 150 minutes).
This really is the solution to our problems with arguing about fast/slow, fit-into-schedule/can't-fit-it-in. You can play at a reasonable pace (2 minutes for championship, 1 day for correspondence), but still have a way to end the game reasonably. Ending a game because BOTH players have used a certain amount of time is really very bad. That's why chess clocks exist. Each player has a certain amount of time. You don't just have one big clock.
|
|
IP Logged |
|
|
|
MrBrain
Forum Guru
Arimaa player #344
Gender:
Posts: 148
|
|
Re: Postal Tournament
« Reply #8 on: Dec 28th, 2004, 9:47am » |
Quote Modify
|
Again, I emphasize... please think about what would happen really if we had the least-time decision mechanism. Would the game really end on time? Go through a game in your mind and figure out how people would really play. If I'm playing with this new decision mechanism, I know that I should not go over 150 minutes total in a 5-hour-limit game. The game is getting up to 60 moves. We've used 3 hours, 48 minutes total. I've used 1 hour and 52 minutes. I know that I should not use more than 38 minutes for my remaining moves, otherwise, I will be in danger of losing the game by having used more time than my opponent. Solution: Move a little quicker. Easy!
So would the game really end at 5 hours? Very unlikely. What is more likely to happen is that the people will speed up their play (or will have played faster earlier in the game) to avoid any chance of running into the 5 hours.
|
| « Last Edit: Dec 28th, 2004, 9:48am by MrBrain » |
IP Logged |
|
|
|
MrBrain
Forum Guru
Arimaa player #344
Gender:
Posts: 148
|
|
Re: Postal Tournament
« Reply #9 on: Dec 28th, 2004, 9:53am » |
Quote Modify
|
Now as for the postal tournament, I believe that Omar said at one point that the amount of time used/left is recorded somewhere in the game log for each game. Is this correct? If so, I don't think it's necessary for it to be displayed since, as people have already pointed out, it's very unlikely that the 180 day limit will come into play. If someone does happen to get close to 180 days, one could simply access, or ask Omar to access, the game log and calculate the total amount of time used for each player thus far. Also, it may be likely that this feature will be in some sort of interface by the end of such a game (more than a half year away!)
|
|
IP Logged |
|
|
|
Fritzlein
Forum Guru
Arimaa player #706
Gender:
Posts: 5928
|
|
Re: Postal Tournament
« Reply #10 on: Dec 28th, 2004, 11:18am » |
Quote Modify
|
Paul is right. The simplest thing is to keep winning the tournament totally separate from the prize money. Winning should be based on 1 point for a win and 0.5 points for a draw. Then we can refund the entry fee to people who don't abandon their games. That's what the entry fees are for, after all.
The reason I wanted to reward number of moves, though, is that it is possible to abandon games without resigning or losing on time. If I wanted to quit after ten moves, but I still wanted my prize money back, I could just start suiciding my pieces, or even dragging the opposing rabbits to the goal.
We could still use the regular formula for determining the tournament winner, and my formula for giving out the prize money. It would be very unlikely (although mathematically possible) for the tournament winner to not get the most money. It is more of a concern to me that someone might purposely choose a move with no chance of winning and a good chance of losing slowly over a move with some chance of winning and a good chance of losing quickly. That would be too bad; I see it as a tradeoff: in order to discourage suiciding we risk encouraging poor play in some sense.
|
|
IP Logged |
|
|
|
Fritzlein
Forum Guru
Arimaa player #706
Gender:
Posts: 5928
|
|
Re: Postal Tournament
« Reply #11 on: Dec 28th, 2004, 12:02pm » |
Quote Modify
|
on Dec 28th, 2004, 9:37am, MrBrain wrote:| I think people are really missing the point of deciding a game by who's used less time. All it really does is to put a limit on the total time you can think for a game. This solves all the problems that we had for the last championship, and frankly the main problem with the arimaa time controls. |
|
I do agree that this time control solves many of our problems with the Arimaa time control, but I'm still slightly leery of adopting it for interactive games.
The reason I'm in favor of your suggestion for the postal tournament is that
a) It's better than the current scoring function
b) 85 moves is long enough that almost all games will end naturally without the rule coming into play at all
c) 24 hours per move is long enough that a "time scramble" may mean trying to move in half a day each time rather than in a full day.
But I maintain my distaste for sudden death time controls in interactive games. I like to take away a player's ability to leave himself, say, one minute for all remaining moves.
The proposed time cutoff isn't so bad because it is only half-way sudden death. The players can't dawdle at the beginning: they have to move at basically the time-per-move pace, because the reserve is less than the sudden death control. I don't mind letting players move at faster than the time-per-move pace if they want to, even when they have a full reserve. But the remaining drawback is that players may be unaware of the time pressure from the time cutoff in a long game, and shoot themselves in the foot that way. To avoid that scenario, I personally would prefer time controls that got faster gradually but always left a certain time (say minimum 30 seconds) per move.
|
| « Last Edit: Dec 28th, 2004, 12:06pm by Fritzlein » |
IP Logged |
|
|
|
MrBrain
Forum Guru
Arimaa player #344
Gender:
Posts: 148
|
|
Re: Postal Tournament
« Reply #12 on: Dec 28th, 2004, 12:36pm » |
Quote Modify
|
on Dec 28th, 2004, 12:02pm, Fritzlein wrote:
The proposed time cutoff isn't so bad because it is only half-way sudden death. The players can't dawdle at the beginning: they have to move at basically the time-per-move pace, because the reserve is less than the sudden death control. I don't mind letting players move at faster than the time-per-move pace if they want to, even when they have a full reserve. But the remaining drawback is that players may be unaware of the time pressure from the time cutoff in a long game, and shoot themselves in the foot that way. To avoid that scenario, I personally would prefer time controls that got faster gradually but always left a certain time (say minimum 30 seconds) per move.
|
|
Yes, that would be better, but that would require a major modification to the arimaa time controls. My idea (really Omar's idea) is easy to implement (we don't need to do it right now), and practically will have almost no effect on arimaa games (as someone noted, the longest-ever game between humans was 75 moves). It's really a correction to what I believe was an error originally, which is having a deadline that's for both players, instead of each player individually.
|
|
IP Logged |
|
|
|
omar
Forum Guru
Arimaa player #2
Gender:
Posts: 1003
|
|
Re: Postal Tournament
« Reply #13 on: Dec 31st, 2004, 12:57am » |
Quote Modify
|
Brian I think I finally see your point about changing the meaning of the game time limit. The purpose of the game time limit parameter (G) was to accomodate situations in which a game must be completed within a certian amount of time. It can still acheive this purpose if we define it to mean the maximum time that can be used by each player for the whole game. So if a game should be finished in 4 hours we currently set G=4, but with the new definition we would just set G=2; meaning that each player can use a maximum of 2 hours for the whole game. The advantage of this is that it eliminates the need for an alternate method such as a score to determine the outcome of the game. I will make this change after the computer championship and challenge match events are over.
For the postal tournament I think we can increase the game limit parameter to 300 days. I don't think any game will ever hit that limit with a 1 day per move pace. The game would have to be about 150 moves before it hits that limit. So I don't think we need to worry about a game being decided by score. I think a much bigger concern is a game being decided by time running out. So I don't mind having a much larger starting reserve and reserve limit to try and prevent that from happening. So I think maybe even having a reserve and max reserve limit of 14 days might be OK. As JDB mentioned it would give people a chance to go on vacation and not worry about losing the game on time. So the new proposed time control is:
1d/14d/100/14d/300d
Paul suggested fixing the color assignment so that everyone plays an equal number of games with each color instead of letting the lower rated player select the color. I think we can accomodate this. How about fixing it so that against the first opponent on your left (in the ring) you play as gold, the second one you play as siver and so on alternating. This fixes the colors for the player on your right. This can also be applied for the round robin case by creating a similar ring (as described in the tournament rules). But in the case when there are an even number of players and you are playing the opponent on the opposite side of the ring, the lower rated player plays gold.
For determining the winners we will use the simple 1 point for win and 0.5 point for draw. But I still like Fritzlein's idea of putting the registration fees into a shared prize pool and redistributing it after the tournament based on the performance formula he suggested. I think it adds a little more fun to the tournament 
|
| « Last Edit: Dec 31st, 2004, 12:57am by omar » |
IP Logged |
|
|
|
MrBrain
Forum Guru
Arimaa player #344
Gender:
Posts: 148
|
|
Re: Postal Tournament
« Reply #14 on: Dec 31st, 2004, 10:06am » |
Quote Modify
|
Yes, Omar, that was my initial proposal to change (or enhance) the arimaa time controls. But after thinking about it, your suggestion of simply changing the winner-deciding mechanism is actually sufficient. If I know there's a 4-hour limit for both players total, I'll make sure to make all my moves in under 2 hours. But even if I don't, with the current 2-person-total limit, I may be able to win the game by goal before total time reaches 4 hours. But again, either way is very close to the same, so I'm fine with either one.
I like the proposed time control. That should be sufficient.
Now, here's another question: How do we prevent a player from "parroting" another player's moves in order to win 1 of 2 tough games? In other words, how can we prevent player A from relaying opponent B's moves onto the board against opponent C? We should have some rule against this, but how?
|
| « Last Edit: Dec 31st, 2004, 10:07am by MrBrain » |
IP Logged |
|
|
|
|
Home
Help
Search
Members
Login
Register
Reply
Notify of replies
Send Topic
Print