|
||||
|
Title: Viewing registered players of a tournament Post by omar on Mar 22nd, 2012, 12:36pm The tournament management tool (TMT) allows an option to hide the registered players of the tournament and only show the number of players registered. I've been contemplating about using this feature in future tournaments where the games are to be considered event games. The key distinction of event games is that the player does not have a choice in selecting the opponents. The gameroom ratings and even WHR ratings (against human opponents) can be inflated by choosing opponents that are weaker. The ratings limited only to event games (WHRE) cannot be inflated since the opponent cannot be chosen and is assigned by the rules of the tournament format. However, if a players can view who has already registered for the tournament they can choose to register or not register based on this information and indirectly influence the strength of the opponents they play; thereby potentially being able to inflate their WHRE rating. A more strictly adherence to the definition of event games would mean that players should not be able to see who else is registered for the tournament before the tournament begins. Seeing who else is registered may motivate others to register and increase the number of players in tournaments such as the potal mixer where large number of players are desired. We could allow seeing who is registered in some tournaments and not in others, but this breaks consistency in what WHRE ratings mean. So I'm interested to hear what others think about this issue. In a separate but related issue; the Autopostal games are currently being counted as event games, but I am starting to think that in a more strict adherence they should not because the player can choose to not play the game against any opponent by resigning on the first move without any negative consequence. The player can also specify one number to select the rating range of the opponents, but I don't think this can be used to inflate ratings. Being able to decline playing games against stronger players can. So I'm interested to hear what people think about this issue as well. |
||||
|
Title: Re: Viewing registered players of a tournament Post by ginrunner on Mar 22nd, 2012, 1:19pm There were two things that kept me from registering for the WC this year. The first (and most important by far) was the entry fee. The second was I didn't think I stood a chance in hell of winning any games. Once I saw that there were people within my skill range I was tempted to sign up but again the fee kept me from it. The ratings system here is flawed slightly because you have people that bot bash (inflation), people that tend to play bots until they beat them (deflation), people who don't play as often as they used to play (inflated ratings), and then the small percentage of people like hanzack who simply don't care about their rating at all and purposely alter it. I don't care what someone's rating is when I play them because there are people above 2000 that I can consistently beat and those rated below me that I find very difficult. If you hide who has registered there is a possibility of less people registering. There tends to be a snowball effect of people registering for events (observation in general maybe not specific to this community). Once 1 or 2 known people have registered then it is psychologically easier to register. Once more people have registered you tend to get larger numbers to register. By hiding the participants I disagree umption is you will have smaller yet higher rated tournaments. This isn't necessarily a bad thing but in a community that is growing that would not be my preference. |
||||
|
Title: Re: Viewing registered players of a tournament Post by hyperpape on Mar 22nd, 2012, 1:59pm Forgive me for my ignorance, but how can you manipulate your rating with WHR? Is it inaccurate for large skill differences? |
||||
|
Title: Re: Viewing registered players of a tournament Post by Fritzlein on Mar 22nd, 2012, 3:30pm on 03/22/12 at 12:36:37, omar wrote:
I disagree with this statement. Yes, choosing your opponents does allow you to manipulate your rating, but not in this way. Choosing low-rated opponents doesn't allow you to inflate your rating except when the opponents are bots. Bots are a special case because they don't change or learn or adapt, so there isn't much learning curve between winning 90% of the time and winning 100% of the time. If I am rated 400 points above a bot, Elo says I will lose 1 out of 11, but in reality I will lose less than 1 out of 11 if I play in exactly the same mode every game. (Assume that I don't get bored and change things up for my own amusement.) The bot will never do something to break out of my winning pattern. Humans aren't like that. If I play a human rated 400 points below me, he is not going to let me win ten times in a row in exactly the same way. He is going to see my strategy coming and do something different to prevent it. I will indeed lose 1 out of 11 games as the ratings predict, because the strategy isn't entirely formulaic. Indeed, Mark Glickman compiled evidence from chess tournaments that extreme rating underdogs win slightly more than the pure Elo formula predicts. Apparently (at least in chess) picking on players rated much lower than yourself is a stupid way to try to inflate your rating. (Let me point out that the one game I lost in the 2011 Postal Mixer was not to chessandgo but to mistre.) The true way to inflate your rating by picking opponents is by (A) choosing to play overrated opponents and refusing to play underrated opponents. For example, since I am better at postal than live, people should want to play me right after the Postal Mixer ends and I have my highest rating of the year, but avoid playing me right after the World Championship ends and I have my lowest rating of the year. Also newcomers are generally overrated on their first day, so you can get an inflated rating by beating up on them, as ArifSyed did. Newcomers tend to be underrated after playing twenty games on the bot ladder, so that is a good time to avoid them. But this is in gameroom ratings only, and doesn't apply very much to WHRX ratings. Or (B) choosing to play opponents that do worse against you than their rating suggests and avoiding the opposite. For example, PMertens had my number. He always beat me more than our respective ratings suggested. But chessandgo owned PMertens, even back when chessandgo and I had about the same rating. So PMertens (if he had given a hoot about ratings, which he didn't) had an incentive to avoid chessandgo and play me. Showing the list of registered players would trivially allow people to jump on weaker opponents and avoid stronger opponents, but that doesn't give anyone an easy path to inflated ratings. Showing the list of players doesn't easily allow people to secure favorable pairings and avoid unfavorable ones, or easily allow people to secure overrated opponents and avoid underrated opponents. I think the dynamic ginrunner mentions is far more powerful and important. People sign up for tournaments when they see other people signing up for tournaments. The record participation in the 2011 World Championship owed a lot to the vocal early registration of Dave Dyer, rated 1387. A lot of lower-rated people thought, "Well, if he can play, so can I," and the more people that registered, the more other people registered as well. It was a snowball effect, a virtuous cycle. I would hate to short-circuit a very positive dynamic due to tenuous theoretical concern over rating manipulation. |
||||
|
Title: Re: Viewing registered players of a tournament Post by omar on Mar 22nd, 2012, 3:33pm on 03/22/12 at 13:19:37, ginrunner wrote:
Actually there are multiple rating systems now. We started out with just the gameroom rating system. But later added the Whole History Rating or WHR (thanks to woh for providing this service) which uses only human-human games in its calculations. Finally we added the Whole History Rating for Events (WHRE) which includes games with and between bots as well, but only includes games that are from events or marked as event games. WHRE is considered to be the most accurate and is not considered to be easily abused. Quote:
The number of players that are registered is shown. So is this not enough to provide the motivation for more players to register. Is it necessary to see who actually is registered? |
||||
|
Title: Re: Viewing registered players of a tournament Post by Fritzlein on Mar 22nd, 2012, 3:35pm on 03/22/12 at 15:33:13, omar wrote:
Yep, people need to see that there are opponents with a similar rating. The best games are even games. People need to see that other people like them are registered, not just the total number of registrants. That's exactly what ginrunner was getting at with the World Championship. If it hadn't been for the high entry fee, seeing that woh and ocmiente had signed up would have induced him to sign up in a way that two anonymous (possibly 2300-rated) signups wouldn't. Also this is what happened with Dave Dyer and the 2011 World Championships. His registration would have had a fraction of its actual impact if it had been just one more in the total. |
||||
|
Title: Re: Viewing registered players of a tournament Post by omar on Mar 22nd, 2012, 4:01pm on 03/22/12 at 15:30:29, Fritzlein wrote:
Good point. But one could still use the information to join the tournament if it has more over rated players and avoid it if it has more underrated players. Quote:
In addition to showing the total number of players that are registered it might be helpful to also show the lowest rating in the tournament. Do we really need to see who the registered players are? |
||||
|
Title: Re: Viewing registered players of a tournament Post by hyperpape on Mar 22nd, 2012, 4:10pm Isn't the point of WHR (or WHRE) that it counteracts the effects of transitory overrating? If Fritz is momentarily overrated after the postal tournament, but then loses a bunch of games, the system retroactively assumes he wasn't really as strong as he appeared at the time, and therefore his opponents also gain less than they would have. Granted, it's not perfect, because the ratings system can't perfectly distinguish between Fritz peaking after the postal tournament and Fritz being overrated, but aren't we talking about very small potentials for manipulation. |
||||
|
Title: Re: Viewing registered players of a tournament Post by omar on Mar 22nd, 2012, 4:27pm on 03/22/12 at 15:35:57, Fritzlein wrote:
I guess you posted while I was still writing my last post. Yes, seeing the lowest rating might not be enough. We could show just the ratings of all the players and not identify the usernames, but someone could probably still match it up from looking at the rating list. We could round the ratings to the nearest 100s or add some randomness to it to prevent this. But now it gets to the question of how far do we want to go to ensure that WHRE ratings are not inflated. Maybe it is not enough of a problem to go to this extent. |
||||
|
Title: Re: Viewing registered players of a tournament Post by ginrunner on Mar 22nd, 2012, 6:38pm on 03/22/12 at 16:27:19, omar wrote:
Agreed |
||||
|
Title: Re: Viewing registered players of a tournament Post by Dolus on Mar 27th, 2012, 2:01pm I like being able to view registered players. It does help me avoid tournaments where I'll get creamed so badly, it would feel like a waste of my time to play, as well as sign up for tournaments where I know there'll be people on the same level as me. Or encourage me to sign up for a tourney if I ever want to play against one of the Arimaa world champs. :D |
||||
|
Arimaa Forum » Powered by YaBB 1 Gold - SP 1.3.1! YaBB © 2000-2003. All Rights Reserved. |