Author |
Topic: 2007 Postal Tournament (Read 4023 times) |
|
Fritzlein
Forum Guru
Arimaa player #706
Gender: 
Posts: 5928
|
 |
2007 Postal Tournament
« on: Feb 5th, 2007, 12:35pm » |
 Quote  Modify
|
Hey, Omar, I see you opened registration for the annual Postal Tournament. Thanks! Let me start a thread to discuss the format, as well as the tournament itself once it gets underway.
I really like the idea of allowing a variable number of games per participant. That will allow everyone to participate at the level their time commitments allow. Furthermore, it gets us further away from thinking of the tournament as a championship, and closer to the "just for fun" model.
One thing I'm not sure I like is that prize money is no longer contingent on how many moves you survive in your games. There used to be an incentive to fight to the death, but now as long as you don't time out or resign, you get your entry fee back. The new system is simpler, but it doesn't discourage someone from suiciding their pieces and pulling an enemy rabbit to goal when they get tired of playing. Also it is all-or-nothing. So for example Thorin, who last year won nine of his games and timed out in the tenth after a hard-fought 48 moves, would get nothing back.
An alternative is to score it exactly like last year, except to divide each player's score by the number of games he has played. That means someone who plays 20 games won't automatically rack up the points and take all the money. Your commitment will be measured relative to how many games you promised you would play.
Given that the pairing algorithm which allows everyone to specify a variable number of opponents is not yet tested, it might pay to run it when you generate the registered players page. So when you click on the link to see who has signed up so far, and for how many games, you would also see which players each person is going to be paired with if the registration doesn't change before the deadline. Right now, of course, there are only five of us signed up, and each willing to play at least four games, so it will look like a round-robin, but it will be interesting to see the pairings start to shake out as more and more folks sign up, and would serve as a test of the new algorithm.
What do you think of allowing a second game with reversed colors between two players who both have spare games after the pairing fills up? I expect that folks who sign up for 12 games will get 12 unique opponents, but folks who sign up for 20 games will not. Should those game-hungry folks play each other twice? Surely not more than twice, though.
Are you going to send out a mass e-mail including an announcement of the Postal Tournament? I know that you don't like to spam, and that you have limited yourself to two mass e-mails in years past, but before the postal tournament is a great time to send out a mailing. Both years of the postal tournament so far I heard several people complain, "I wish I hadn't started playing just after the tournament started!"
I notice the links on the right hand side of the game room still point to the 2006 Arimaa Challenge and the 2006 Postal tournament instead of the current events.
The time control still concerns me, since theoretically the game could last 300 days with only 70 moves played on each side, and thus be decided on score after only 70 moves. I know that didn't happen last year, and if it ain't broke, don't fix it, but even so the 80-day initial reserve is the culprit, and if people can specify fewer than ten games, then we don't need so much initial reserve. 40 days should be more than enough given that you can scale the number of games, so I propose changing the time control from 1d/80d/100/0/300d/21d to 1d/40d/100/0/300d/21d. That would also insure games go at least 110 moves before being decided on score.
That's my laundry list of concerns, but basically I expect another fantastic tournament this year, since last year went so well. Each year the infrastructure and traditions and community seem to get a little stronger.
|
|
 IP Logged |
|
|
|
omar
Forum Guru
Arimaa player #2
Gender:
Posts: 1003
|
|
Re: 2007 Postal Tournament
« Reply #1 on: Feb 9th, 2007, 5:11am » |
Quote Modify
|
Thanks for starting this tread Karl.
I've updated the postal tournament link in the gameroom; thanks for noticing that.
Redistributing the proper amount to each player was getting to be a bit combersome. So I wanted to keep that part simple this year for my sake  Hope it doesn't affect the players too much.
I'll try to add a link for viewing the current pairings.
I like the idea of allowing a second game between the game hungry players.
I will be sending out a mailing to notify players about the postal tournament.
I also like the idea of shortening the initial reserve now that players can choose the max number of games they want to play. But I would suggest something like 60 days. Would like to know what others think about this.
|
| « Last Edit: Feb 9th, 2007, 5:14am by omar » |
IP Logged |
|
|
|
woh
Forum Guru
Arimaa player #2128
Gender:
Posts: 254
|
|
Re: 2007 Postal Tournament
« Reply #2 on: Feb 9th, 2007, 6:25am » |
Quote Modify
|
on Feb 9th, 2007, 5:11am, omar wrote:I also like the idea of shortening the initial reserve now that players can choose the max number of games they want to play. But I would suggest something like 60 days. Would like to know what others think about this.
|
|
I wouldn't mind if the initial reserve is cut down to 20 days. That would still work for me. I think 40 or 50 days should be enough for every one.
I will join the tournament whatever the initial reserve is.
|
|
IP Logged |
|
|
|
arimaa_master
Forum Guru
Arimaa player #2010
Gender:
Posts: 358
|
|
Re: 2007 Postal Tournament
« Reply #3 on: Feb 9th, 2007, 9:31am » |
Quote Modify
|
I will be happy with 14 days. But I think 30 days could be the right choice.
|
|
IP Logged |
|
|
|
chessandgo
Forum Guru
Arimaa player #1889
Gender:
Posts: 1244
|
|
Re: 2007 Postal Tournament
« Reply #4 on: Feb 9th, 2007, 10:38am » |
Quote Modify
|
If the aim is to make the games the highest level possible then it seems logical to keep the time reserve as large as possible, in the limit of what is practical. 110 moves games are more than unlikely, so 40 days initial time reserve is more than reasonnable, isn't it ? 60 days seems good as well, and I wouldn't mind 80 days either.
|
|
IP Logged |
|
|
|
IdahoEv
Forum Guru
Arimaa player #1753
Gender:
Posts: 405
|
|
Re: 2007 Postal Tournament
« Reply #5 on: Feb 9th, 2007, 12:16pm » |
Quote Modify
|
I think 40 days reserve would be okay, but I would discourage making it too small. I don't know about others, but I have an awful lot of travel planned in the next 4 months.
|
|
IP Logged |
|
|
|
Fritzlein
Forum Guru
Arimaa player #706
Gender:
Posts: 5928
|
|
Re: 2007 Postal Tournament
« Reply #6 on: Feb 9th, 2007, 12:31pm » |
Quote Modify
|
Random note: we now have nine players signed up, one in each century. (i.e. one 23xx, one 22xx, one 21xx, etc.) I like how that underscores the "just for fun" tournament design. Everybody can play, and I hope everybody does!
|
|
IP Logged |
|
|
|
OLTI
Forum Full Member
Arimaa player #1034
Gender:
Posts: 25
|
|
Re: 2007 Postal Tournament
« Reply #7 on: Feb 9th, 2007, 2:46pm » |
Quote Modify
|
on Feb 5th, 2007, 12:35pm, Fritzlein wrote:
Furthermore, it gets us further away from thinking of the tournament as a championship, and closer to the "just for fun" model.
|
|
Actually I don't like so much the idea of "just for fun". We can play "just for fun" games all the time even without paying a registration fee.
If there is not going to be selected a winer this isn't a "Tournament" anymore.
I think it's more stimulating to play for wining the tournament.
|
|
IP Logged |
|
|
|
Fritzlein
Forum Guru
Arimaa player #706
Gender:
Posts: 5928
|
|
Re: 2007 Postal Tournament
« Reply #8 on: Feb 9th, 2007, 5:04pm » |
Quote Modify
|
on Feb 9th, 2007, 2:46pm, OLTI wrote:| If there is not going to be selected a winer this isn't a "Tournament" anymore. |
|
One thing that distinguishes a tournament from pickup games is that if you enter the tournament, you can't run away from someone else who enters. I've been avoiding chessandgo for a while, but now he is guaranteed a game against me.  In general everyone can count on getting a variety of opponents who aren't rated too far away, and who are serious (as insured by the entry fee). It's hard to get such a good profile of opponents casually.
Furthermore the results will still be collected and prominently displayed for all posterity. I may have a thousand games in my game history, but the ones anyone is likely to ever see are my tournament games, so there is still an element of glory in doing well that doesn't exist for casual games. Does anyone care that I beat robinson postally in game 19732? You would have to hunt for it if I didn't brag, but my win over robinson in the 2006 Postal Tournament was only two clicks away from the gameroom for the better part of a year.
Finally, Omar has mentioned to me the possibility of having Arimaa titles in the future, such as "Expert" and "Master", which would be based on winning tournament games rather than just based on ratings. Start collecting your Arimaa grandmaster norms now!
|
|
IP Logged |
|
|
|
chessandgo
Forum Guru
Arimaa player #1889
Gender:
Posts: 1244
|
|
Re: 2007 Postal Tournament
« Reply #9 on: Feb 10th, 2007, 3:15am » |
Quote Modify
|
on Feb 9th, 2007, 5:04pm, Fritzlein wrote:
Start collecting your Arimaa grandmaster norms now!
|
|
lol I agree with Olti, though : it wouldn't seem much more complicated to have the postal tournament select a winner, or a bunch of winners, and it would keep it more in the spirit of a tournament
|
|
IP Logged |
|
|
|
Fritzlein
Forum Guru
Arimaa player #706
Gender:
Posts: 5928
|
|
Re: 2007 Postal Tournament
« Reply #10 on: Feb 10th, 2007, 8:50am » |
Quote Modify
|
What's a good metric for selecting a winner if everyone can play a different number of games? Winning percentage? Wins minus losses? Performance rating?
|
|
IP Logged |
|
|
|
chessandgo
Forum Guru
Arimaa player #1889
Gender:
Posts: 1244
|
|
Re: 2007 Postal Tournament
« Reply #11 on: Feb 10th, 2007, 11:42am » |
Quote Modify
|
With the aim of keeping it of real tourney, I'd prefer last year's rules, where everyone played a fixed amount of games ...
|
|
IP Logged |
|
|
|
OLTI
Forum Full Member
Arimaa player #1034
Gender:
Posts: 25
|
|
Re: 2007 Postal Tournament
« Reply #12 on: Feb 10th, 2007, 6:19pm » |
Quote Modify
|
I agree with Jean 
|
|
IP Logged |
|
|
|
woh
Forum Guru
Arimaa player #2128
Gender:
Posts: 254
|
|
Re: 2007 Postal Tournament
« Reply #13 on: Feb 12th, 2007, 1:07pm » |
Quote Modify
|
on Feb 10th, 2007, 8:50am, Fritzlein wrote:| What's a good metric for selecting a winner if everyone can play a different number of games? Winning percentage? Wins minus losses? Performance rating? |
|
I agree with chessandgo and OLTI. A tournament should gave a winner. So I have been thinking how to select one.
I started with the scoring system of last year's tournament but to select a winner I didn't liked the idea that the loser can gain as much points as the winner. So I changed it to
- A player receives 100 points for winning
- A player loses 100 points if they forfeit, resign or lose on time
- If a player loses any other way they receive points equal to the number of moves in the game up to a limit of 60 points
It's easier to win all your games when you play 3 then when you play 12 games. So a player who wins 12 out of 12 should be ahead of a player that wins 3 out of 3. And also a player who loses 12 out of 12 should be behind a player that loses 3 of 3. Therefore each player gets 70 points for a fictitious game. 70 being the mean of a game he won and a game he lost in 40 moves.
So we get: score = (sum(Pt) + 70) / (Nt + 1)
Experimenting with this formula in different situations I found that a player who plays more games is by far less punished for forfeiting a game. So I left the games lost by forfeit, by resigning or lost on time out of the above calculation. Instead for each of those game 30 points is subtracted from the result.
Nt: total number of games played
Nf: number of games forfeited, resigned or lost on time
Nn=Nt-Nf: number of games won or otherwise lost
for each of those last games Pn points is awarded where Pn equals
- 100 points for winning
- when losing: the number of moves in the game up to a limit of 60 points
Now we get: score = (sum(Pn)+70) / (Nn + 1) - 30 * Nf
Any thoughts on this scoring system?
|
|
IP Logged |
|
|
|
Fritzlein
Forum Guru
Arimaa player #706
Gender:
Posts: 5928
|
|
Re: 2007 Postal Tournament
« Reply #14 on: Feb 12th, 2007, 3:42pm » |
Quote Modify
|
The Postal Tournament is currently paired by trying to give everyone as many even games as possible. There have been very few games where the players were more than 400 points apart in rating, because the biggest mismatches were simply not paired. As a result, higher-rated players get tougher opponents than lower-rated players. Does this create a problem for people who would like the tournament to have a clearly-defined winner?
For example, last year Adanac scored 8-2 against opponents with an average rating of 1985, while Thorin scored 9-1 against opponents with an average rating of 1791. Should Thorin place higher because he had more wins, even though it was against much weaker opposition? Maybe to really determine a winner in a fair manner, we not only have to have everyone play the same number of games, we also have to have everyone play across the entire spectrum of opponents, i.e. we need to pair more mismatches.
My hunch is that stronger players generally are more interested in determining a tournament winner, whereas weaker players are generally more interested in having good games. I say this as a weak chess player. When I used to go to chess tournaments, I would be very happy for the Swiss pairing. Every round I lost, I would get to play a weaker opponent, whereas every round I won I would get to play a stronger opponent. The method of determining the tournament winner was totally irrelevant to me, because I had no chance of being that winner. All I cared about was having good games.
If I had been told that the purpose of the tournament was just to determine a winner, and that I would have to get destroyed a few times by the top-rated players in order to help them determine a winner, I probably wouldn't have bothered to participate. I think our World Championship tournament has a similar deterrent effect on low-rated players. Why should they enter just to get beaten up? In order to enter, you have to enjoy taking a shot at someone rated way above yourself, and not everyone enjoys this.
I think it is totally to be expected that more low-rated players enter the Postal Tournament than enter the World Championship. Look right now: for the World Championship we had only one entrant rated below 1700. For the Postal Tournament we already have four entrants rated below 1700. I think this is entirely attributable to the "just for fun" tournament design. That's why I'm not in favor of focusing so much on determining a winner.
|
|
IP Logged |
|
|
|
|
Home
Help
Search
Members
Login
Register
Reply
Notify of replies
Send Topic
Print