Dominion Strategy Forum

Please login or register.

Login with username, password and session length
Pages: [1]

Author Topic: strange rating again  (Read 3773 times)

0 Members and 1 Guest are viewing this topic.

papadjango

  • Baron
  • ****
  • Offline Offline
  • Posts: 50
  • Respect: +18
    • View Profile
strange rating again
« on: April 09, 2015, 01:01:58 pm »
0

Can someone explain this rating?
Just had a game, where the player with about 4100 goko points quit just before finish. I was in front (my score is 3990) and the second player (about 3850) was second.
Player:    Rank: VPs:    Turns:    Quit?
eellzzss    2nd    31    19    
papadjango    1st    39    19    
Instant Moniker    3rd    27    19    Quit

I only got 2(two!) points for my victory!?
Because of the quit? Or why?
Logged

SCSN

  • Mountebank
  • *****
  • Offline Offline
  • Posts: 2227
  • Respect: +7140
    • View Profile
Re: strange rating again
« Reply #1 on: April 09, 2015, 02:32:20 pm »
+3

Of the many roads leading to insanity, wondering why, whether, when or how Goko does something is the shortest.

Because prevention is better than cure, I recommend sticking to the isotropish leaderboard. If you use the Salvager browser extension you can also display those within the Goko client if you so desire.
Logged

WanderingWinder

  • Adventurer
  • ******
  • Offline Offline
  • Posts: 5275
  • ...doesn't really matter to me
  • Respect: +4381
    • View Profile
    • WanderingWinder YouTube Page
Re: strange rating again
« Reply #2 on: April 09, 2015, 02:49:34 pm »
0

Of the many roads leading to insanity, wondering why, whether, when or how Goko does something is the shortest.

Because prevention is better than cure, I recommend sticking to the isotropish leaderboard. If you use the Salvager browser extension you can also display those within the Goko client if you so desire.

Did you even read his post? He's playing 3-player. isotropish doesn't rate that.

SCSN

  • Mountebank
  • *****
  • Offline Offline
  • Posts: 2227
  • Respect: +7140
    • View Profile
Re: strange rating again
« Reply #3 on: April 09, 2015, 03:15:20 pm »
0

Why the caustic tone? I wasn't aware isotropish didn't rate 3p games because I never play them myself.
Logged

WanderingWinder

  • Adventurer
  • ******
  • Offline Offline
  • Posts: 5275
  • ...doesn't really matter to me
  • Respect: +4381
    • View Profile
    • WanderingWinder YouTube Page
Re: strange rating again
« Reply #4 on: April 09, 2015, 03:18:09 pm »
+1

Why the caustic tone? I wasn't aware isotropish didn't rate 3p games because I never play them myself.

I don't think my post is any more caustically-toned than yours.

SCSN

  • Mountebank
  • *****
  • Offline Offline
  • Posts: 2227
  • Respect: +7140
    • View Profile
Re: strange rating again
« Reply #5 on: April 09, 2015, 03:26:43 pm »
0

Really not getting what your beef is here.

My original post is only slightly caustic if your name is Goko and you can't handle being made fun of in a lighthearted manner.
Logged

Goko

  • Pawn
  • **
  • Offline Offline
  • Posts: 1
  • I wish people would stop Making Fun of me
  • Respect: +26
    • View Profile
Re: strange rating again
« Reply #6 on: April 09, 2015, 03:37:17 pm »
+26

My original post is only slightly caustic if your name is Goko and you can't handle being made fun of in a lighthearted manner.
Nah, it's fine.
Logged

Arctic Penguin

  • Chancellor
  • ***
  • Offline Offline
  • Posts: 23
  • Respect: +54
    • View Profile
Re: strange rating again
« Reply #7 on: April 09, 2015, 03:44:30 pm »
+1

If you're going to be making fun of the online version, you should at least get the name right. They even changed it into a mnemonic, and still no one remembers...

Can someone explain this rating?
Just had a game, where the player with about 4100 goko points quit just before finish. I was in front (my score is 3990) and the second player (about 3850) was second.
Player:    Rank: VPs:    Turns:    Quit?
eellzzss    2nd    31    19    
papadjango    1st    39    19    
Instant Moniker    3rd    27    19    Quit

I only got 2(two!) points for my victory!?
Because of the quit? Or why?
According to this thread it looks like you were awarded a win over the quitting player, Instant Moniker, but shared 1st with eellzzss. Perhaps the rating system counted it as the sum of a win over Instant Moniker and a draw/loss to eellzzss because he/she started after you?
« Last Edit: April 09, 2015, 03:51:43 pm by Arctic Penguin »
Logged

papadjango

  • Baron
  • ****
  • Offline Offline
  • Posts: 50
  • Respect: +18
    • View Profile
Re: strange rating again
« Reply #8 on: April 09, 2015, 05:26:51 pm »
0

thanks arctic pinguin for at least one serious answer. So you mean that although I won by points, both "not quitters" (eellzzss and me) are treated equal, because Instant Moniker quit? Should be more points for both of us nevertheless, because he has the highest rating. But as Shecantsayno correctly mentioned: Of the many roads leading to insanity....  :o
Logged

WanderingWinder

  • Adventurer
  • ******
  • Offline Offline
  • Posts: 5275
  • ...doesn't really matter to me
  • Respect: +4381
    • View Profile
    • WanderingWinder YouTube Page
Re: strange rating again
« Reply #9 on: April 09, 2015, 05:53:05 pm »
+3

thanks arctic pinguin for at least one serious answer. So you mean that although I won by points, both "not quitters" (eellzzss and me) are treated equal, because Instant Moniker quit? Should be more points for both of us nevertheless, because he has the highest rating. But as Shecantsayno correctly mentioned: Of the many roads leading to insanity....  :o

Yeah, and it has to be this way - you can very often be in a much player position than another player even if they have more points than you - think about a player who opens Estate/Estate/Estate/Estate. You will beat them essentially every time, but if a third player quits on turn 4, they will have more points than you....

As for why it's only 2 points... well, one thing is that it's shared 1st, which I think it is treating like you tied with the other guy, which it probably shouldn't. So my 'it has to be this way' is maybe not entirely right - the thing that "has to be this way" is that you can't get awarded a win over the guy you were left in the game with just for having more points on game-end. But you probably also shouldn't naturally have it thrown at you that you did equally well as him, because who knows, maybe you really were crushing him. The software doesn't exist to be smart enough to know that though; it's beyond reasonable programming capacities, I think, because it requires an actually good AI, which hasn't really been made for this game, and beyond that actually, it would need to know how you would play the game out, and well, that's obviously unrealistic.

The last thing I would note is that what MF displays is not actually what they have down as your rating (in terms of how strong they think you are), but that number MINUS some uncertainty number. So there are a few things in that which can hurt you here. (I'm going to simplify this a bit because most people don't want a small lecture on Graphical Models). If you don't play often, your uncertainty number might be really high. In that case, and if your opponents' uncertainty numbers were much lower, then it might actually think you're much better than the other players (despite you having similar displayed ratings), thus giving you not so much credit (and having your tie with the other guy penalize you more than you might think). The other thing would be if at least one of those players had really high uncertainty, your rating might not move much at all. Did you get before-and-after rating numbers for both of them? Probably not - it's pretty hard to get afters sometimes, and you could easily not get befores in some cases, too.

papadjango

  • Baron
  • ****
  • Offline Offline
  • Posts: 50
  • Respect: +18
    • View Profile
Re: strange rating again
« Reply #10 on: April 11, 2015, 07:25:42 am »
0

thnks you for detailed answer. I appreciate that!  :D
Logged

Throwaway_bicycling

  • Moneylender
  • ****
  • Offline Offline
  • Posts: 151
  • Respect: +140
    • View Profile
Re: strange rating again
« Reply #11 on: April 15, 2015, 11:28:46 pm »
0

thanks arctic pinguin for at least one serious answer. So you mean that although I won by points, both "not quitters" (eellzzss and me) are treated equal, because Instant Moniker quit? Should be more points for both of us nevertheless, because he has the highest rating. But as Shecantsayno correctly mentioned: Of the many roads leading to insanity....  :o

So I have a slightly different theory. One of the cool things about TrueSkill is that it can in principle adjust the ratings of multiple players in a single game...under normal circumstances. But in a situation where there is a premature end to the game, the algorithm must decide how to assign points to the other players. What I am guessing is happening here is that both you and eellzzss are not being treated as *equal* but, rather, are being treated *the same*. So Instant Moniker clearly loses points; let's say for the argument 30 points. Neither of you beat each other, so neither of you can gain or lose points to each other. So the points that are at stake need to be divided between the two players (ratings gains and losses are zero sum). Because eellzzss has a lower rating than you, she or he will gain more points than you, ceteris paribus. Now, as to why you get just two, I think that Wandering Winder's explanation probably contains most of the salient bits.

So I am probably not in the majority here, but at some point I think it would be interesting to see if there could be Isotropish-ish ratings boards for 3P or 4P games. A lot of people play that way IRL, and a number of in-person tournaments have multi-player rounds, so as long as you don't just collapse ratings over numbers of opponents, there is some useful and interesting information to be gotten.
Logged
Pages: [1]
 

Page created in 0.116 seconds with 20 queries.