Author |
Topic: 2012 World Championship format (Read 15773 times) |
|
omar
Forum Guru
Arimaa player #2
Gender: 
Posts: 1003
|
 |
Re: 2012 World Championship format
« Reply #15 on: Jan 21st, 2011, 5:47pm » |
 Quote  Modify
|
aaaa, I saw the explanation you gave of the advantages in the other thread.
I'd like to see the simulation results from this.
|
|
 IP Logged |
|
|
|
aaaa
Forum Guru
Arimaa player #958
Posts: 768
|
|
Re: 2012 World Championship format
« Reply #16 on: Jan 26th, 2011, 6:21pm » |
Quote Modify
|
I now realize that the influence of seeding can be further decreased by only having the initial seeds serve as a tiebreaker after calculating the performance ratings using equal priors. This will also have the advantage that one would only need a relative ordering and no absolute ratings, allowing the scheduling algorithm to be used for the computer championship as well (if the bots will continue to be seeded in advance).
|
|
IP Logged |
|
|
|
Hippo
Forum Guru
Arimaa player #4450
Gender:
Posts: 883
|
|
Re: 2012 World Championship format
« Reply #17 on: Jan 27th, 2011, 9:33am » |
Quote Modify
|
I also dislike the WHR filtering. On previous year I have played 1 HvH normal speed game before the turnament started (with Tuks who was disconnected). I have finished 2 autopostals till that (one started on GMR around 1300 the other around 1600). The participation on WC was big opportunity for me.
I have played several funny matches with PMertens losing a lot of rating points to him during the WC period and then I played HvH only occasionally. I have played postals on mixer, autopostal and causal.
I had time on my trip to play the blitz/fast turnament, but I had such bad connection that I lost with Trevor game which could not be lost with good connection.
Actually there was lack of players there so not playing would be worse for the comunity ... I don't care the loss, it was fun, but it wouldn't be in the case of WHR filtering. I still have very few HvH matches played and each one has big influence on my WHR.
I am not sure how much WHR changed, but as I read it long time ago, the player was considered to have a fixed rating (old games were as important as the new ones) ... if that remains, players would be encouradged not to play HvH till they got high skill level and start after that ... Is it really the incentive we want to create? Or may be to create duplicate accounts whenever the skills jump up such that the historical WHR pulls you down too much?
In case of duplicate accounts one could use one for causal games the other for building his position ...
I have looked at my statistics now ... I have played 18 rated HvH games in 1 minute per move. 7 of them were highly important. 4 were training games where I didn't care the final result 6 were just causal games and losing in one game was almost planned. (The highly important games were AWL game and the WC games not ending on disconnection).
Games played during WC period are mostly to train some special kind of the game ...
|
| « Last Edit: Jan 27th, 2011, 10:06am by Hippo » |
IP Logged |
|
|
|
aaaa
Forum Guru
Arimaa player #958
Posts: 768
|
|
Re: 2012 World Championship format
« Reply #18 on: Feb 17th, 2011, 4:38pm » |
Quote Modify
|
I have come to the conclusion that ignoring one-sided forfeits for the purpose of recalculating seeds gives too much of an incentive for a weaker player in a lopsided match not to show up in order to improve his or her seeding in the next round. For example, if exactly one player were to forfeit in the first round, then in the next round, that player would always become the best-seeded player amongst those with a loss.
I now propose that one-sided forfeits are handled completely identical to a regular loss by the tournament rules (except for the required appeal for continuing one's participation).
For the purpose of calculating performance ratings, doubly forfeited games would still continue to be treated as not having taken place at all (while otherwise still being registered as a loss for both players of course).
|
|
IP Logged |
|
|
|
aaaa
Forum Guru
Arimaa player #958
Posts: 768
|
|
Re: 2012 World Championship format
« Reply #19 on: Feb 24th, 2011, 7:35pm » |
Quote Modify
|
on Jan 19th, 2011, 8:19pm, aaaa wrote:| Consolation playoff games as usual (will the increase of number of lives cause any problems here?). |
|
By running simulations of floating triple elimination tournaments, I have been able to determine that, of the losers, as many as 5 players can share the most number of wins, and in case it's only one player, as many as 6 players can have the second-most number of wins.
Omar, how many rounds of playoff games would you be willing to accept? Of course, you could also just abolish playoff games altogether and allow multiple players to share second and third place.
|
|
IP Logged |
|
|
|
omar
Forum Guru
Arimaa player #2
Gender:
Posts: 1003
|
|
Re: 2012 World Championship format
« Reply #20 on: Feb 25th, 2011, 7:45pm » |
Quote Modify
|
on Feb 24th, 2011, 7:35pm, aaaa wrote:
By running simulations of floating triple elimination tournaments, I have been able to determine that, of the losers, as many as 5 players can share the most number of wins, and in case it's only one player, as many as 6 players can have the second-most number of wins.
Omar, how many rounds of playoff games would you be willing to accept? Of course, you could also just abolish playoff games altogether and allow multiple players to share second and third place. |
|
Thanks for looking into this aaaa. I'm just wondering how many players were in tournament. Basically you are saying that the second place could potentially have 5 players and 3rd place could potentially have 6 players. Wow, it would extend the tournament several more rounds to break these ties. I think I might opt to just allow multiple players win second and third place and split the prizes.
|
|
IP Logged |
|
|
|
aaaa
Forum Guru
Arimaa player #958
Posts: 768
|
|
Re: 2012 World Championship format
« Reply #21 on: Feb 28th, 2011, 3:43pm » |
Quote Modify
|
on Feb 25th, 2011, 7:45pm, omar wrote:| I'm just wondering how many players were in tournament. Basically you are saying that the second place could potentially have 5 players and 3rd place could potentially have 6 players. |
|
I had these many ties for second and third place occur with simulated tournaments starting out with as few as 11 and 19 players, respectively.
Quote:| Wow, it would extend the tournament several more rounds to break these ties. I think I might opt to just allow multiple players win second and third place and split the prizes. |
|
That would seem particularly appropriate given that everyone starts out with 3 lives. I say this, because, if I'm not mistaken, the number of lives should correspond to how many players will be strictly ranked at the top if each of them would always beat anyone weaker. In other words, if a top 3 of players were to manifest itself perfectly, then this should likewise be reflected in the final standings, and that without having any playoff games.
On a different subject, given that the championship is to become a ruthless tournament to determine the best player, the whole idea of changing time control settings mid-tournament doesn't seem to make sense in the first place. I assume then, that it's going to be 90 seconds per move throughout the entire championship?
Also, I think it's worth noting that having surplus time be completely banked could be construed as actually being a compromise position in light of some even wanting the possibility of using up the entire reserve regardless of total turn time. If one were to accept such a spectrum of positions, then it follows that, at minimum, the idea of partial banking is being overwhelmingly rejected here.
|
|
IP Logged |
|
|
|
omar
Forum Guru
Arimaa player #2
Gender:
Posts: 1003
|
|
Re: 2012 World Championship format
« Reply #22 on: Mar 2nd, 2011, 12:04pm » |
Quote Modify
|
I think one of the reasons for using a faster time control in the preliminaries was to allow more people to be able to participate. So if we limit the number of people in the WC championship then I think it would be best to have the same time control throughout the event. If we want to allow more people to participate then using a faster time control in the earlier rounds would be better.
We also should standardize on some simple time controls and start using these on the site and in events games. To simplify the time controls I suppose we could always have 100% of unused time added to reserve and no limit on building the reserve. That eliminates needing to specify two of the six parameters. If the initial reserve, max time per turn and total game time are specified in terms of the time per move then we only have one parameter that needs to be specified. We kind of already have names for various speed games based on time per move; like Blitz=15 sec/move, Fast=30 sec/move, etc.
|
| « Last Edit: Mar 2nd, 2011, 12:44pm by omar » |
IP Logged |
|
|
|
omar
Forum Guru
Arimaa player #2
Gender:
Posts: 1003
|
|
Re: 2012 World Championship format
« Reply #23 on: Mar 2nd, 2011, 12:39pm » |
Quote Modify
|
Here are the most common time controls that we currently use. These are for non-postal game where one of the players is a human.
Code:
select count(*) as n, timecontrol from game where postal=0 and (wtype='h' or btype='h') group by timecontrol order by n desc limit 20;
+-------+------------------------+
| n | timecontrol |
+-------+------------------------+
| 44991 | 2/2/100/10/8 |
| 21653 | 1/1/100/5/0 |
| 14366 | 2/2/100/2/0 |
| 12304 | 2/2/100/10/0 |
| 10251 | 2/2/100/5/0 |
| 9144 | 15s/1/100/2/2 |
| 7108 | 30s/3/100/5/8 |
| 6938 | 0:30/3/100/5/8 |
| 5674 | 15s/1:30/100/2/2 |
| 2982 | 3/3/100/15/8 |
| 2211 | 15s/1:30/100/2/4 |
| 2018 | 15s/90s/100/2m/4h |
| 1756 | 2/2/100/2/8 |
| 1495 | 60s/5m/75/0/4h/4m |
| 1123 | 15s/90s/100/2/4h |
| 1076 | 8s/60s/100/2/2 |
| 1002 | 1/2/100/3/0 |
| 871 | 2/2/100/10/5 |
| 844 | 45s/3/100/5/0 |
| 790 | 0:45/4:30/100/0/3:30/5 |
+-------+------------------------+
|
|
Here it is for event games:
Code:
select count(*) as n, timecontrol from game where postal=0 and eventgame=1 and (wtype='h' or btype='h') group by timecontrol order by n desc limit 20;
+-----+------------------------+
| n | timecontrol |
+-----+------------------------+
| 424 | 60s/5m/75/0/4h/4m |
| 62 | 90s/90s/100/7m/6h |
| 54 | 90s/5m/75/0/6h/5m |
| 39 | 15s/90s/100/120s/50m |
| 39 | 2m/2m/100/10m/8h |
| 25 | 30s/1m/100/3m/90m |
| 18 | 2/2/100/10/8 |
| 15 | 30s/3m/100/5m/8h |
| 14 | 1/1/100/3/3 |
| 10 | 1:30/1:30/100/10/5 |
| 8 | 3/3/100/15/8 |
| 7 | 1:15/1:15/100/3/5 |
| 4 | 1:30/1:30/100/3/7 |
| 3 | 30s/1m/100/3m/1:30h |
| 3 | 2/2/100/15/7 |
| 3 | 0:30/3:00/100/0/2:30/5 |
| 1 | 1/1/100/5/0 |
| 1 | 1:30/1:30/100/7/0 |
| 1 | 60s/2m/75/0/4h/4m |
+-----+------------------------+
|
|
For faster games it seems we want the initial reserve to be a much higher multiple of time per move than for slower games. So it might be hard to have one multiple fit all cases, but maybe something like this will do:
initial reserve (R) = 4 x time per move (M)
max turn time (T) = 4 x time per move (M)
max game time (G) = 240 x time per move (M)
So for a 15 second per move (Blitz) game this would be:
15s/1m/100/0/1h/1m
For a 30 second per move (Fast) game this would be:
30s/2m/100/0/2h/2m
For a 60 second per move game this would be:
1m/4m/100/0/4h/4m
For a 90 second per move game this would be:
1m30s/6m/100/0/6h/6m
For a 2 minute per move game this would be:
2m/8m/100/0/8h/8m
|
| « Last Edit: Mar 2nd, 2011, 12:46pm by omar » |
IP Logged |
|
|
|
Janzert
Forum Guru
Arimaa player #247
Gender:
Posts: 1016
|
|
Re: 2012 World Championship format
« Reply #24 on: Mar 2nd, 2011, 12:58pm » |
Quote Modify
|
I think one of the great things about Arimaa is its very powerful and flexible time control system. I certainly understand why we want a few standard, named controls (e.g. blitz, fast, etc.) but I think essentially restricting the time control system to a much less flexible set is a detriment not a benefit.
I would much rather when a player creates a game, in addition to a drop down with the standard controls they also be allowed to enter any arbitrary control.
Janzert
|
|
IP Logged |
|
|
|
omar
Forum Guru
Arimaa player #2
Gender:
Posts: 1003
|
|
Re: 2012 World Championship format
« Reply #25 on: Mar 2nd, 2011, 1:15pm » |
Quote Modify
|
I'm not proposing that we reduce the Arimaa time control system to a simplified one. No, the current flexible system would still be there. The ability to enter your own time control has been on my todo list for awhile. Although this is a neat feature for power users, I don't think this is something that would commonly be used or is critical needed to improve the site, so I never get around to it.
|
|
IP Logged |
|
|
|
Fritzlein
Forum Guru
Arimaa player #706
Gender:
Posts: 5928
|
|
Re: 2012 World Championship format
« Reply #26 on: Mar 2nd, 2011, 5:12pm » |
Quote Modify
|
I agree with both Janzert and Omar. It's nice to have all the parameters to play with, but in practice people respond well to things that have a few, easy, prominent choices. Think of the one-button I-pod. So we should decide on behalf of users which standard time controls are best and offer only those rather than the huge list we have now. Also I would like a single option to enter an arbitrary time control.
I think 45s/move should remain a standard option, leaving six standard choices instead of five.
Also, let me point out that reducing the number of standard choices doesn't force us to bank 100% of the unused time in every case. We could instead choose to bank 75% of the unused time in every case. Either way there would just be six standard time controls.
Recall the case for partial banking of unused time: it rewards thinking for approximately the same time on every move relative to blitzing some moves and thinking a long time on others. What is one of the most annoying things for spectators of chess games? The uneven pace of moves. Maybe the opening twenty moves are played in ten minutes and then the next move by itself takes twenty minutes. And then after dragging their heels throughout the middlegame, the players often have a time scramble to make it to forty moves and get more time. Then the game screeches to a halt and the spectators have another long wait while the players try to sort out what just happened.
By all means, let's have a uniform standard for simplicity, but let it be uniform in a spectator-friendly way. The main argument in favor of banking 100% of unused time is that people are used to it. Anything new and different can upset people. And certainly it is new and unusual to organize a board game partially for the convenience of the spectators rather than entirely for the convenience of the players.
And even in terms of the convenience of the players, is banking 75% a burden? It requires only an adjustment towards a steady move pace, namely thinking a little longer on the reflex moves and moving a little faster on the analysis paralysis moves. Keep that in the back of your head, and that's it. It's not like the players are required to compute for themselves how many seconds of reserve they are gaining/losing at any given time. We've been banking 75% reserve throughout this World Championship and the previous, without it causing any problems. If it really made the players uncomfortable, you can be sure they would be squawking. On the upside, spectator enjoyment of games has been taking off like wildfire. Let's keep a good thing going.
|
| « Last Edit: Mar 2nd, 2011, 5:26pm by Fritzlein » |
IP Logged |
|
|
|
mistre
Forum Guru
Gender:
Posts: 553
|
|
Re: 2012 World Championship format
« Reply #27 on: Mar 4th, 2011, 7:47am » |
Quote Modify
|
I agree that a simplified time control system would make the site more user friendly. I also agree with Fritzlein that we should keep the 45s option. If you count just human-human non-event games, you will find it is a popular option.
I am a little concerned about the banking of only 75% and shorter initial reserve times in fast and blitz games. I think these changes could potentially cause some inflation in those bot ratings due to even more increased time pressure. However, for games of 45s or longer, I am in favor of the proposed changes.
We should also look at reducing the number of postal options. It appears that 95%+ of postal games are on the 1d/60d/100/0/300d/21d setting anyways. I actually prefer the 14d reserve, but most players don't like it, so I stopped using it. Maybe a 30d reserve option?
Lastly, this time control topic should be moved to a new thread as it has gotten off track from talking about the original subject.
|
|
IP Logged |
|
|
|
Fritzlein
Forum Guru
Arimaa player #706
Gender:
Posts: 5928
|
|
Re: 2012 World Championship format
« Reply #28 on: Mar 4th, 2011, 9:31am » |
Quote Modify
|
I guess blitz and fast games are not very conducive to spectators anyway. Maybe there could be a standard breakpoint; games at 60s/move and longer bank 75% of unused time, while games at 45s/move and faster bank 100% of unused time.
|
|
IP Logged |
|
|
|
rbarreira
Forum Guru
Arimaa player #1621
Gender:
Posts: 605
|
|
Re: 2012 World Championship format
« Reply #29 on: Mar 4th, 2011, 9:43am » |
Quote Modify
|
Since bot ratings were mentioned, I should say that I expect at least some of the bots to not be ready for 75% reserve accumulation (I know mine isn't).
So if that becomes the standard for human games it will feel different when playing bots.
|
|
IP Logged |
|
|
|
|
Home
Help
Search
Members
Login
Register
Reply
Notify of replies
Send Topic
Print