January 13 (First Anniversary) RussCon Report
First Anniversary Report!
Upcoming Settlers Tournament!
The Unexpected Return of Devil Point Theorizing!
...and More Fun Than You Can Shake a Stick at!
There were 20 of us on the First Anniversary RussCon! As part of the ongoing worldwide celebration, there will be a Settlers Tournament at my house, Sunday January 24. It will be run RussCon-style, with the winner determined by devil points. See below for particulars.
At the first RussCon (January 14, 1998) the gamers present were: Eric, Peter, Marty, Ken who played Loewenherz, won by Eric Tom, Doug, RussD, RussW, Brady, who played Material World, won by Tom
All these original folks were here except Tom. (Who would have been, Doug tells me, except he kept Tom up all night playing games the night before...)
Midway through the evening we held a champagne toast to commemorate this joyous occasion. After an exhaustive and lengthy search for a cork screw, we eventually opened a bottle of wine as well.
William brought Nickelodeon Green Slime Frozen Pops. The cover showed bright cheery colors. JonathanR tried one; it was red on the outside and brown on the inside, and basically looked like a frozen bloody dog turd on a stick.
There were tentative efforts to finally give JonathanR (aka The Third Jonathan) a nickname: Slimy. It's not clear if that's going to stick (so to speak).
Tutanchamun 5 JeffF 4 Laurie 2 William 0 JP -2 Marty -4 Tutanchamun 6 Marty 5 JeffF 3 William 0 JP 0 JonathanC -3 Laurie -5 TrainsportAustria 5 RussW 4 Jay 1 Clayton 1 JonathanR -2 Daniel -4 Guillotine 5 RussW 4 Clayton 2 Daniel 0 Jay -2 JonathanR -4 Elfengold 5 Steve 4 William 2 Marty 0 JeffF -2 JonathanC -4 Settlers 4 JP 3 Laurie 1 Peter -2 EricH -2 QuoVadis 5 JonathanR 4 Daniel 1 RussW 1 Tim -2 Jay -4 Fight 3 RussW 2 JonathanR 0 Daniel -2 LastChance 3 JP 2 Steve 0 Laurie -2 Can'tStop 4 JeffF 3 Marty -1 William -1 JonathanC -1 Medici 5 Tim 3 Steve 3 William 0 JeffF -2 Marty -4 Kahuna 2 William 1 Brady -1 Bohnanza 4 JeffF 3 Tim 1 Marty -1 Steve -3 MedievalMerchant 6 Daniel 5 JP 3 JonathanC 1 JonathanR -1 RussW -3 Laurie -5 Samurai 3 Doug 2 JonathanB 0 JeffF -2 Raj 5 JP 4 Doug 2 JonathanB 0 RussW -2 JeffF -4 Settlers 4 William 3 Marty 1 Brady -1 James -3 Successors 4 James 3 JonathanB 1 Doug -1 Ken -3 Can'tStop 3 William 2 Doug -1 RussW -1 Can'tStop 3 William 2 Doug 0 RussW -2 Samurai 3 JP 1 James 1 JonathanB -2
I note with pride that my Trainsport Austria dominance is firmly established now that I'm past my initial slow learning period.
There were tie victories in Medici and Samurai!
The theme? Identically initialed music makers! (Alliterative artists. Paired performers.) JonathanR & JonathanC said so. Good, guys! Slimy saw first, 'fore Sea Biscuit began to think.
Big Black, The Rich Man's Eight Track Tape
George Gershwin, Rhapsody in Blue
Herbie Hancock, Maiden Voyage
Monster Magnet, Dopes to Infinity
Shakespear's Sister, Sacred Heart
Rank ratings: 0.4400 JP (7 games played) 0.3750 Clayton (2) 0.3214 William (9) 0.3077 Steve (4) 0.1818 Tim (3) 0.1538 Doug (5) 0.1250 James (3) 0.1111 RussW (8) 0.1034 JeffF (8) 0.0000 Daniel (5) -0.0909 JonathanB (4) -0.1538 Marty (7) -0.1579 JonathanR (5) -0.4118 JonathanC (4) -0.4167 Jay (3) -0.4737 Laurie (5) -0.5000 Brady (2) -0.6667 EricH (1) -0.6667 Peter (1) -1.0000 Ken (1) Win ratings: 0.6541 JP (7) 0.6250 James (3) 0.4118 Steve (4) 0.3596 William (9) 0.2952 RussW (8) 0.2778 JeffF (8) 0.1429 Tim (3) -0.0566 Daniel (5) -0.0588 JonathanR (5) -0.2523 Doug (5) -0.3007 Marty (7) -1.0000 Clayton (2) -1.0000 JonathanB (4) -1.0000 JonathanC (4) -1.0000 Jay (3) -1.0000 Laurie (5) -1.0000 Brady (2) -1.0000 EricH (1) -1.0000 Peter (1) -1.0000 Ken (1)
JP is the Devil! William and Steve are Vice Devils!
Ken gets the Generous Angel Award for pulling off the rare and dangerous Reverse Brady Maneuver (which would have been received by Eric or Peter as well, but they cooperated to tie for last place in their Settlers game). William gets the Dedicated Award.
A New Proposal for Win Ratings:
A year ago I was intensely brainstorming rating systems. It seems appropriate that a year later I suddenly get a new ratings brainstorm...
The current win rating formula is more complex than the rank rating formula. It also gives a strangely skewed distribution, with lots of people having an equal rating of -1. Note that if you win no games, your rating is -1. This is regardless of the number of opponents or presence of ties, e.g. losing an 8-player game and losing a 3-player game both give you -1. Once you've won a game, then the differences show up, but otherwise, your rating is -1 regardless of the circumstances of your losses. This might be undesirable and unfair.
Analogously, if you win all your games, then your rating is +1, regardless of whether any were ties. Once you lose some, then those differences show up, but otherwise it's +1 regardless of the circumstances of your wins.
To address these issues, a simple modification to the rank rating formula may be what we want. Recall that the rank rating formula is simply:
(#opponents you came ahead of - #opponents who came ahead of you) / #opponents.
That gives a result between -1 and 1.
Thus a unique winner in a n-player game gets n-1 points from that game, and if that's the only game played, his rating is (n-1)/(n-1), which is 1. If 2 players win a joint victory, they each get a rating of (n-2)/(n-1), which is less than 1.
If we simply treat all the nonwinners as having tied (as losers), then this gives an elegant measure of winningness, I believe. So in a n-player game with a unique winner, all the losers would get -1 points, and if that's their only game, they each get a rating of -1/(n-1). Losing an 8 player game gives you a rating of -1/7. Losing a 3-player game gives you a rating of -1/2, which is worse than -1/7, which seems fair.
The intuitive motivation for it is quite simple: rank ratings care about whether you came in 2nd, 3rd, etc., whereas win ratings don't -- so treat all the losers equally. Essentially it's as if there are just 2 bands of tying players: the winners (normally 1 player) and the losers. None of the losers beat any of the other losers. Simple.
A player's win rating would be -1 only if all the other players tied for victory. Thus -1 would be a rare and ignominious event.
Here would be the new-style win ratings for this evenings results. Note as a sanity check that the resulting player order is roughly the same as the old formula, i.e., both formulas are plausible in some sense.
0.3846 Steve (4) 0.3750 James (3) 0.2800 JP (7) 0.1852 RussW (8) 0.1379 JeffF (8) 0.0909 Tim (3) 0.0714 William (9) 0.0526 Daniel (5) 0.0000 JonathanR (5) -0.0769 Marty (7) -0.1538 Doug (5) -0.2353 JonathanC (4) -0.2500 Clayton (2) -0.2500 Jay (3) -0.2632 Laurie (5) -0.3333 EricH (1) -0.3333 Peter (1) -0.3333 Ken (1) -0.4545 JonathanB (4) -0.5000 Brady (2)
If nothing else, this gives a less wacky distribution which is more directly comparable to the rank ratings distribution. I would appreciate any feedback or thoughts on this. I plan to switch to this method unless someone pokes holes in it.
A particularly interesting question is whether a joint victory should be potentially harmful. Under the old win rating system, winning with a tie could never hurt your rating. Under this system, winning a tie can actually hurt your rating if your rating is sufficiently high. This makes settling for a tie less of a no-brainer. Which seems more correct to you?
Also there is the unusual case of Everybody Wins (or Everybody Loses). The old system would make this result have no effect on your win rating. The new system would push your rating toward 0, i.e., the positive ratings are hurt and the negative ratings are helped. This seems potentially reasonable.
A historical footnote:
In an email sent to the RussCon mailing list on Feb 9, 1998, I first analyzed the possibility of win ratings as distinct from rank ratings, and ALMOST came up with this method. But there was an error in my analysis which no one noticed making it appear that this penalized you more for losing versus more opponents, instead of the correct penalizing you less for losing versus more opponents! Also, I took it as axiomatic that winning a joint victory should never hurt your rating, but now I question that. In fact one of the primary motivations for the more complex old formula was the treatment of ties.
Just imagine how the course of history would have been altered had I recognized this formula a year ago. Life as we know would be completely different!
Settlers Tournament Details:
It shall start at noon Sunday January 24 1999. No entry fee, yet there will be prizes! How is this possible? Life is a mystery!
It's a continuous self-grouped tournament, so you can come and leave when you want between games. You just need to play 3 or more games at some point during the day. I will collect the results of all games and determine the winner. We'll try to wrap it up in the evening, perhaps around 8 or 9pm, to allow some computation time in case the winner is not readily apparent. Play boardgames and talk amongst yourselves while the judges deliberate.
Important: you need to play a variety of people, not just the same group over and over! If two recorded games have the same set of players then the later game will not count! So mix it up between games. This might mean a small wait between games. Such is life. Quit your whining. I will play as needed to get a game started.
3 and 4 player games are both allowed and encouraged.
Unless all players in a game agree beforehand, we use the usual optional rule that victory point cards work just like other special cards, i.e., you may only play one per turn, and not the turn you buy them.
Unless all players agree beforehand, sevens are rerolled in the first 2 rounds.
And let's keep this pure Settlers: no Seafarers, Stadt und Ritter, Brady variants, etc.
To be eligible to win, you must have played at least 3 games. Of course if you want to come and just play one or two, that's fine too!
The Champion will be the player at the top of the total ordering induced by the usual method on the partial ordering of rank ratings and win ratings. (Ties broken by whoever played the most games.)
There will a Dedicated Award for whoever played the most games. (Ties broken by devil points.)
There may be other awards and zaniness; I'm still making this up.
Please bring your own Settlers set if you have one! Otherwise there might not be enough!
Daniel brought Blood Bowl to show people, since he is interested in starting up a league! It looks rather fun. You've got your team of orcs or dwarves or whatever, and the players gain experience between games, unless they get killed. Insane weirdness. Contact Daniel at firstname.lastname@example.org for more info.
Tim forwards the following game, author unknown:
The Golden Strider
Each person plays a runner in a race. To start, each player chooses five starting "cards" with value from one to ten each; the total of the five must be 30 points. These cards will be signified like so: 8/0, meaning "value 8, received turn 0" (at game start).
On a turn, each player simultaneously plays one card, dividing that card's value up into three categories, pretty much as desired. Those categories are movement, stamina, and overtaking cost.
For every point spent on movement, the player advances one square toward the finish line. Everyone starts on square zero, and the finish is at square 60.
The card received as a replacement to the one just played is determined by stamina and position. The base value of the replacement is twice the amount just allocated to stamina. Added to this is a bonus of one to those runners now in second or third position, and a bonus of two to those runners now in fourth, fifth, or sixth. The value of the replacement will be between a minimum of zero and a maximum of ten. For example, if I play 8/0 on the first turn, allocate 4 to movement and 4 to stamina, and end up in second place after playing this card, I would advance 4 squares and receive 9/1 as my replacement.
Overtaking costs *must* be paid from the card played on the next turn. For every position a player moves up during one turn, an overtaking cost of one point is incurred. For example, if I start a turn in seventh position, move 3 squares and end up in a four-way tie for third, then I must pay 4 points in overtaking costs out of the next card I play. Only positions are taken into account: if I am passed by 2 players but pass another 3, thus moving up one position, I pay one point of overtaking cost the next turn. I must play a card with enough value to pay the cost. EXCEPTION: No overtaking costs need be paid for passes made during turns one and two.
Finally, cards must be played after a certain time interval (overexerting will catch up to you). No x/0 (starting cards) may be held at the end of turn seven. This implies that no more than one x/0 card may be held at the end of turn six, and so on backwards in time. Likewise, no x/1 card may be held at the end of turn eight, and so on forward in time. ***This rule overrides the rule that a card with enough value to pay overtaking costs must be played.*** This rule may be summarized as follows:
At end of turn: You must hold: 4 3 or fewer x/0 5 2 or fewer x/0 6 1 or fewer x/0 7 no x/0 8 no x/1 9 no x/2 10 no x/3 11 no x/4 12 no x/5 13 no x/6 14 no x/7
The first player to cross the finish (square 60) is the winner. If two or more players cross during the same turn, then the one who was ahead at the beginning of the turn is the winner. If the players were tied, then the one who ends up on the largest-numbered square is the winner. If the players ended up on the same square, then the race is a dead heat.
SLIME in the ice machine!