Sep 23 RussCon Report

It was a normal-sized turnout (15 of us), but we played fewer games than normal (only 4!), mostly thanks to an unprecedented 9-player game of Outpost! No player played more than 2 games this evening!

JP won Outpost, the TimJim classic where the rich always get richer but it's still fun.

Tim won Ursuppe, the amoeba game.

RussW won Priceless, to my amazement -- I would have sworn I was losing hideously. Very interesting strange little financial/auction game, brought to us courtesy of the evening's Devil (KevinH). Check it out.

Doug won Elfenland, Marty's most excellent new German import.


Evening's Soundtrack:

...And You Will Know Us by the Trail of Dead (industrial surrealism)
The Black Blood (part 1 of the X and Y Trilogy, a cybernetic opera)
The Horse I Ride Has Wings (opera songs accompanied by piano)
Ingrid Karklins, A Darker Passion (Latvian-influenced pop)
Rockbusters, S.O.B.GYN. (funny crude rock)

Nobody deduced the subtle & mysterious common theme linking these disparate CDs. The answer appears below.


Rank ratings:
 1.0000 Tim (1 game played)
 0.7500 Doug (2)
 0.6000 Brady (1)
 0.4545 Ken (2)
 0.3077 JP (2)
 0.2727 RussW (2)
 0.2308 Marty (2)
-0.0909 JeffF (2)
-0.0909 KevinH (2)
-0.2500 James (2)
-0.3333 Alfred (1)
-0.5000 Jay (1)
-0.7500 JonathanB (1)
-0.8000 William (1)
-1.0000 Clayton (1)

Win ratings:
 1.0000 Tim (1)
 0.7419 RussW (2)
 0.6842 JP (2)
 0.5385 Doug (2)
-1.0000 Brady (1)
-1.0000 Ken (2)
-1.0000 Marty (2)
-1.0000 JeffF (2)
-1.0000 KevinH (2)
-1.0000 James (2)
-1.0000 Alfred (1)
-1.0000 Jay (1)
-1.0000 JonathanB (1)
-1.0000 William (1)
-1.0000 Clayton (1)


Digression on Win Ratings:

This give a simple test of your intuition on win ratings, and whether or not you agree with the RussCon formulas. In general, I think we would all agree that winning a 6 player game is more glorious than winning a 4 player game, but how much more? And similarly, losing a 4 player game is more ignominious than losing a 6 player game. But how should these get quantified and combined?

Tim played and won 1 game. Surely most people would agree Tim should get the highest win rating since he won a game and lost no games.

RussW won a 4-player game and lost a 9-player game.

JP won a 9-player game and lost a 6-player game.

Doug won a 6-player game and lost a 4-player game.

Comparing JP & Doug's results, I think everyone would agree JP should be higher than Doug, since JP's win was more glorious than Doug's, and Doug's loss was more ignominious.

But what about RussW & JP? The formula used by the RussCon Statistical Judging Committee makes RussW have a higher win rating than JP. Does this agree with your intuition? I'd suggest that most people's intuition is a bit murky here. You might argue JP should be higher than Russ since winning a 9-player game is more glorious than winning a 4-player game. But on the other hand, JP lost a 6-player game, which is more ignominious than losing a 9-player game. So how do these balance out in your intuition?

The actual calculations are here:

Russ = (3/4 - 1/9) / (3/4 + 1/9) = 23/31
JP = (8/9 - 1/6) / (8/9 + 1/6) = 13/19
Doug = (5/6 - 1/4) / (5/6 + 1/4) = 7/13

In general, from each game you get win points:
Winner gets #losers / #players
Loser gets -#winners / #players
This has the nice interesting property that the total points awarded to players in a given game sums to 0. It also handles joint victories. The win points from a game are bounded between -1 and 1.

Then your cumulative win rating for a set of games is:
(sum game points) / (sum abs(game points))

The cumulative win rating is also bounded between -1 and 1.

E.g. if Squeaky & I win a joint victory in a 4-player game, then I lose a 5 player game (to a sole victor), my combined win rating is:
(2/4 - 1/5) / (2/4 + 1/5) = 3/7.

If I'd lost that 5-player game to a coalition of 2 winners, then my win rating is:
(2/4 - 2/5) / (2/4 + 2/5) = 1/9.

Thus it is more ignominious to lose a game when there are more winners and you aren't among them. This certainly agrees with my intuition. In the extreme case, everybody except me wins the game, and that is surely a maximally ignominious fate.

Note that ties can include the degenerate case of "Everybody wins", which is equivalent to "Everybody loses" -- either way, everyone gets 0 win points from that game. And such a game has no effect on your combined win rating. (Such a game WOULD affect your combined rank rating, however.)


The statistics are meant only to indicate how well you performed on a given occasion. They do not purport to indicate how good a player you are overall. (Although there would presumably be some loose correlation there.)

Various people have argued that somehow the system should factor in the complexity and luck-vs-skill of the game. These are, alas, subjective and nebulous. And since the point is simply measuring how well your games went, as opposed to how great a player you are, I could argue that it's not even theoretically necessary to take into account such things.

Former RussCon attendee Randy (now banished to the Middle East where he teaches mathematics to nubile young college women) suggested that it might make sense to weight the points from individual games according to how long the game took to play. E.g. a quick game of Fight would carry much less weight than a long game of Titan. This seems like an interesting and promising idea, but would require people to note how long the game lasted when they record the results. In reality, I am convinced that missing and inaccurate data would plague such a scheme. As to whether it's a theoretically sound idea... I'm still not sure.


Who is the Devil?

Obviously having fewer games played gives a much simpler partial ordering, since most people got a win rating of 0:

        / | \
       / Doug\
      /   |   \
     JP Brady RussW
      \   |   /
       \ Ken /
        \ | /
    JeffF = KevinH

Tim would be the Devil, except he commited that faux pas known as the Brady Maneuver! (I.e., playing only 1 game and winning it.) Recall that to qualify as the Devil, you must have played more than 1 game.

Therefore Doug is the Devil! Then JP, Brady, and RussW have a 3-way tie.

Clayton commited that faux pas known as the Reverse Brady Maneuver.


Answer to the Evening's Soundtrack Theme:

This one was harder than the numerological and Shakespearean themes... The common thread is simply that all the music was made and played in Austin!


Coney Island White Fish!

See you next wednesday,