Author |
Topic: One versus the Mob (Read 10226 times) |
|
Fritzlein
Forum Guru
Arimaa player #706
Gender: 
Posts: 5928
|
 |
Re: One versus the Mob
« Reply #15 on: Mar 22nd, 2007, 8:50am » |
 Quote  Modify
|
Should the teams be allowed to consult computers? I don't think it would change the playing strength of a team very much, but it would be interesting to hear what Bomb thinks as we go, so I would vote for allowing computer assistance.
|
|
 IP Logged |
|
|
|
arimaa_master
Forum Guru
Arimaa player #2010
Gender:
Posts: 358
|
|
Re: One versus the Mob
« Reply #16 on: Mar 22nd, 2007, 10:07am » |
Quote Modify
|
on Mar 22nd, 2007, 8:50am, Fritzlein wrote:| Should the teams be allowed to consult computers? I don't think it would change the playing strength of a team very much, but it would be interesting to hear what Bomb thinks as we go, so I would vote for allowing computer assistance. |
|
I am for allowing computer assistence too.
|
|
IP Logged |
|
|
|
chessandgo
Forum Guru
Arimaa player #1889
Gender:
Posts: 1244
|
|
Re: One versus the Mob
« Reply #17 on: Mar 22nd, 2007, 10:08am » |
Quote Modify
|
on Mar 21st, 2007, 4:27pm, OLTI wrote:
Why not have both, then ? 
|
|
IP Logged |
|
|
|
aaaa
Forum Guru
Arimaa player #958
Posts: 768
|
|
Re: One versus the Mob
« Reply #18 on: Mar 22nd, 2007, 4:52pm » |
Quote Modify
|
You need to have the move determination well thought out to prevent the same restrictiveness that, for example, plagued "Kasparov versus The World". Bringing up Condorcet is a good thing, but there are two problems. One is that you need a 100% decisive method, which Condorcet isn't (not only because of symmetry). The second more important point, which was already pointed out, is that even if you rank moves by preference, there are still way too many possibilities to list them all in order and thus it often will be the case that because many moves aren't being ranked by more than a few voters, this would result in a de facto approval vote of moves based simply on how many voters have ranked them at all. This would allow a small minority of people to (possibly inadvertently) collude in determining the resulting collective vote.
To prevent this from happening, but still not restrict people in their choice of moves, whenever it's the turn of a group of voters, they should continuously maintain a preferential ballot during the discussion. This should go on until a majority is satisfied or time has run out, after which the final choice of move is determined by the chosen voting method (Schulze would be a good choice for this), with any possible ties broken by a random vote or by a (pre-game or per-turn) designated team captain.
With the members being made aware of each other's current preferences and the intermediary collective choice that would result from it, a feedback loop should result in these becoming more elaborate over time and that a move that is a good reflection of the collective opinion is being settled on as soon as possible; if one were to be unsatisfied with the current hypothetical result, one could attempt to change it by introducing new compromise moves in one's ranking, which in effect would be proposals for consideration by others. Those could in turn consider ranking them as well and/or introduce counterproposals of their own in the same manner. This negotiation by repeated polling is all to make sure a collective choice is (relatively) well-supported and not an unsatisfactory surprise in the end.
Obviously, moves here should be considered to be identical based on whether their resulting board positions are.
|
|
IP Logged |
|
|
|
NIC1138
Forum Guru
Arimaa player #65536
Gender:
Posts: 149
|
|
Re: One versus the Mob
« Reply #19 on: Mar 22nd, 2007, 10:42pm » |
Quote Modify
|
As you can read in that article, the Schulze method *is* a Condorcet method!!... 
I don't imagine we might run into terrible paradoxal election situations... I don't even think we will need this kind of thing badly... I imagine that (at least a few steps inside the game) there won't be a big number of moves to choose from... If it happens that the debate over a certain move heats up, we just follow the martial arts tradition: the guys with highest rank and age decide! 
I can even envision a small web page where the users can enter the moves, give arguments pro and con, and organize their votes, to be seen by everyone... Has anybody started to think abou that already??  We can even use many plugins, to see the outome of the elections using multiple schemes, from dictatorship to anarchy! 
|
|
IP Logged |
|
|
|
The_Jeh
Forum Guru
Arimaa player #634
Gender:
Posts: 460
|
|
Re: One versus the Mob
« Reply #20 on: Mar 23rd, 2007, 12:53am » |
Quote Modify
|
With such a large number of prospective candidates, this voting situation is very difficult indeed. My struggles to come up with a simple, foolproof solution have failed. Nevertheless, here are two ideas to ponder:
1. Should proposals receive a certain number of "signatures" from other mob members before being considered "candidates?" (similar to the political election process)
2. Should candidates receive votes, be ordered by preference, or receive satisfaction ratings?
|
| « Last Edit: Mar 23rd, 2007, 1:09am by The_Jeh » |
IP Logged |
|
|
|
aaaa
Forum Guru
Arimaa player #958
Posts: 768
|
|
Re: One versus the Mob
« Reply #21 on: Mar 23rd, 2007, 4:20am » |
Quote Modify
|
on Mar 22nd, 2007, 10:42pm, NIC1138 wrote:As you can read in that article, the Schulze method *is* a Condorcet method!!...
|
|
I never implied it wasn't. I just want to make clear that when one speaks of "Condorcet method", one shouldn't interpret this as an explicit method (like plurality) which would be a somewhat indecisive one, but rather as an instantiation of a generic class of methods.
|
|
IP Logged |
|
|
|
IdahoEv
Forum Guru
Arimaa player #1753
Gender:
Posts: 405
|
|
Re: One versus the Mob
« Reply #22 on: Mar 23rd, 2007, 10:59am » |
Quote Modify
|
Try this system:
1) Have a coordinator for the group.
2) Initial discussion is completely free-form; anyone can suggest moves and/or indicate approval for suggested moves
3) On a certain day of the week-long process, the coordinator picks the top N suggested moves, based on his subjective impression of which N have the most support from the community. N is a pre-chosen integer; I think anything from 4 to 7 would work well.
3.5) (Optional) Coordinator may have the ability to increase N by 1 or 2 for any particular move if it seems necessary to include all well-supported candidates.
4) Coordinator publishes these N moves as the official candidates for the group.
5) The whole group votes on those N moves. Ballots must rank all N candidates to be a legitimate ballots.
6) Ballots are scored using Schultze, the coordinator publishes the results and submits the winning move in the gameroom.
The additional criterion that voters must rank all N candidates prevents problems with moves winning just because they were ranked at all. Having a coordinator select the appropriate candidates from the discussion avoids the problem of having eighty candidates for each move.
|
|
IP Logged |
|
|
|
JacquesB
Forum Senior Member
Arimaa player #2380
Gender:
Posts: 26
|
|
Re: One versus the Mob
« Reply #23 on: Mar 24th, 2007, 1:10pm » |
Quote Modify
|
1. I like that all players may propose moves, but at the end some difference should be made. Strategical evaluation is based on suppositions because it cannot be solved by search. These suppositions are grounded on experience. The only way to prove one has that successful experience is: rating.
I think something like:
rating >= 2200 -> vote weight x 4
2000..2199 -> vote weight x 3
1800..1999 -> vote weight x 2
below -> vote weight x 1
would be nice. Of course, strong players don't have 4 votes. They vote only one move but that counts four times.
I am a weak player and will learn a lot from the discussions, but my vote should be considered with care. It is also a good reason for improving our own rating if we want to become "heavier".
2. It is more complicated, I know, but how about a double round system. It could be done with fixed dates: First round until Friday, second round Saturdays 24 hours so all timezones can do it. The second round with only two moves to choose, the first open.
|
|
IP Logged |
|
|
|
The_Jeh
Forum Guru
Arimaa player #634
Gender:
Posts: 460
|
|
Re: One versus the Mob
« Reply #24 on: Mar 24th, 2007, 2:15pm » |
Quote Modify
|
Of course better players will be able to make better choices, but they will also be able to persuade more inexperienced players that their reasoning is good. A bad move proposed by a younger player will quickly but courteously be eliminated by a better player. The end result should be that the move chosen is better.
|
|
IP Logged |
|
|
|
Fritzlein
Forum Guru
Arimaa player #706
Gender:
Posts: 5928
|
|
Re: One versus the Mob
« Reply #25 on: Mar 24th, 2007, 2:38pm » |
Quote Modify
|
Jacques, I agree that higher rating roughly corresponds to better strategic judgment, but I would prefer a system where each player's vote counts equally, as a democratic experiment. I am curious whether a team might actually play worse than its strongest player. We would have a chance to test this, particularly in the One vs. the Mob scenario rather than the team vs. team scenario. Folks who scoff at democracy will expect a team to do worse and worse as it has more members, and would expect the One to win against the Mob every time.
Yet my experience from team chess is that the team can easily play better than its strongest player, even with one vote per person. Furthermore, every additional member can make a contribution of analysis. The key is not is the voting (although incidentally the Schulze method sounds great) but rather the key is a productive discussion. Weaker players can often refute the moves proposed by stronger players, but this isn't any benefit unless they can propose the refutation in the discussion, and everyone can have a chance to hash it out.
The point of one vote per player is that it forces people to lobby for their votes. If my vote counted 4x, and I knew that a few other strong players were on my side, I would not have to bother explaining why it was a good move, because I would know that it would win the vote anyway. But if each vote counts 1x, and lots of players seem to be attracted to a weaker move, then I must not only propose a better move and vote for a better move, but also campaign for the better move. The more democratic the system is, the more it forces public discussion.
The free exchange of ideas helps everyone because, of course, the experts are often wrong. I make fewer mistakes than the average Arimaa player, but I still make a ton of mistakes. When I'm on a team, I want my ideas to have to compete on a level playing field, without any undue respect for my supposed insight. Let the power of my ideas win the day, not the power of my reputation.
|
|
IP Logged |
|
|
|
The_Jeh
Forum Guru
Arimaa player #634
Gender:
Posts: 460
|
|
Re: One versus the Mob
« Reply #26 on: Mar 25th, 2007, 12:39am » |
Quote Modify
|
I think IdahoEv's suggestion is the most sensible. The coordinator would function sort of like the Speaker of the House. If the game is played on the server, he could also be logged in against the One and input moves on behalf of the Mob, if necessary (although I would like to see "The_Mob" on the screen). If the Mob don't like the job he is doing, they can vent their feelings, but I don't think that will be a problem in this amiable community.
If the One and the Mob both take a week to move, the game could last two years. I guess this would be all right, but I'd like to see a more efficient way of doing things. Here is a suggestion:
1. From the time the Mob officially makes a move, the One has until 11:59 P.M. CDT/CST seven days later to make a move. He may make a move ahead of time if he so wishes.
2. From the time the One makes a move, the Mob has until 11:59 P.M. six days later to produce candidates. The Mob then has until 11:59 P.M. the following day to vote and move. The Mob schedule is slightly flexible, but as much of the time should be used as possible.
3. Neither the One nor the Mob shall have a time reserve.
I think the One will move ahead of time quite often, so this will speed things up. The Mob members will have to keep their heads up as to when it's their turn, but that's okay. It's a continuous process, and they'll be speculating even before the One moves.
|
| « Last Edit: Mar 25th, 2007, 12:53am by The_Jeh » |
IP Logged |
|
|
|
JacquesB
Forum Senior Member
Arimaa player #2380
Gender:
Posts: 26
|
|
Re: One versus the Mob
« Reply #27 on: Mar 25th, 2007, 3:10pm » |
Quote Modify
|
1. One person one vote: Its Ok for me. Your reasons sound good. Let's see what it gives. I am eager to start. 
2. Reading the previous post I see that you mean one week per move. I had misunderstood as one week per round (=two moves). If the player agrees, I think it makes sense. I think routine helps. Having always the same days for the discussion and the election can make people more committed.
The player would have 3.5 days, e.g. full-Sunday to mid-Wednesday. Of course the admin could add some extra time if he cannot play one week. I think it is enough time.
|
|
IP Logged |
|
|
|
NIC1138
Forum Guru
Arimaa player #65536
Gender:
Posts: 149
|
|
Re: One versus the Mob
« Reply #28 on: Mar 28th, 2007, 12:13am » |
Quote Modify
|
on Mar 23rd, 2007, 4:20am, aaaa wrote:| I never implied it wasn't. I just want to make clear that when one speaks of "Condorcet method", one shouldn't interpret this as an explicit method (like plurality) which would be a somewhat indecisive one, but rather as an instantiation of a generic class of methods. |
|
You are right, sorry if I was rude. I only wanted to give some credit to Condorcet because I read a book about old french mathematicians just the other day! 
I actually came to this through the Schulze method, reading about the ongoing Debian elections...
|
|
IP Logged |
|
|
|
NIC1138
Forum Guru
Arimaa player #65536
Gender:
Posts: 149
|
|
Re: One versus the Mob
« Reply #29 on: Mar 28th, 2007, 12:32am » |
Quote Modify
|
on Mar 23rd, 2007, 10:59am, IdahoEv wrote:
I also think that's pretty much it... Plus I also agree with Fritz on the need for debate. Or else it would be more fun to have the 3 top players on the Mob plus the One playing 6 games against each other!... There must be politics and rhetoric in a game like this!
And I still envision this site where players will enter their suggestions, that will pile up in a box, where you can click on the links, and see the board... Then you say wether you like that move or not, and the moderator will select the moves for the final voting... Then we can keep the final votes open for a day in the site, for everyone to see each other's votes, and let people modify theirs!...
I'm pro a week-long complete cycle... I prefer to play during the week, since my girlfriend doesn't let me use my computer during the weekend (that's because *she* uses it ) . The eligible Ones should tell how many days they would find good enough for them to play.
I ask again: will we try something like this?... Who wants to work on the program?... (of course IRC / MSN and the coordinator taking care of all with pencil and paper is just fine too.)
|
|
IP Logged |
|
|
|
|
Home
Help
Search
Members
Login
Register
Reply
Notify of replies
Send Topic
Print