News
|
| Carcassonne | 5 | Daniel 4 RussW 2 MarkY 0 Whendy -3 Steve -3 |
| Cloud9 | 4 | MarkY 3 Steve 1 Daniel -1 Carly -3 |
| Bohnanza | 5 | BradSmith 4 JP 2 Clay 0 Clayton -2 Ben -4 |
| Tichu | 2 | ( Zack MarkH ) 1 ( William Dan ) -1 |
| Settlers | 4 | JimG 3 RussW 1 BradS -1 RussD -3 |
| WebOfPower | 5 | Daniel 4 Zack 2 Carly -1 Steve -1 MarkY -4 |
| Carcassonne | 4 | JP 3 Clay 1 BradSmith -1 Clayton -3 |
| Tichu | 2 | ( Dan Ben ) 1 ( MarkH William ) -1 |
| TitanArena | 5 | Carly 4 MarkY 2 Zack 0 Daniel -2 Steve -4 |
| SchottenTotten | 2 | JimG 1 BradS -1 |
| Chrononauts | 2 | PaulM 1 Fina -1 |
| Chrononauts | 2 | Fina 1 PaulM -1 |
| Outpost | 5 | JeffF 4 PJ 2 Marty 0 Fina -2 PaulM -4 |
| Carcassonne | 4 | JP 3 RussD 1 RussW -1 BradSmith -3 |
| WasSticht | 4 | Steve 3 Daniel 1 Dan -2 MarkY -2 |
| Tichu | 2 | ( Zack Ben ) 1 ( JimG William ) -1 |
| Settlers | 3 | BradSmith 2 RussW 0 RussD -2 |
| Can'tStop | 3 | William 2 MarkY -1 Steve -1 |
| Gess | 2 | JP 1 PJ -1 |
| Gess | 2 | JP 1 PJ -1 |
| CosmicCoasters | 2 | Steve 1 Daniel -1 |
| Exxtra | 5 | Ben 4 Zack 2 William 0 Dan -2 MarkY -4 |
| Bluff | 6 | RussW 5 PJ 2 JP 2 Steve -1 MarkY -4 BradSmith -4 |
| Settlers | 4 | Carly 3 JeffF 1 Marty -1 BradS -3 |
Good to see that old favorite Settlers getting played several times! Gotta fight the evil Tichu tide.
JP & PJ played Gess (linked to last week), and it looked quite interesting.
Also cool to see someone else besides me & JP playing the clever game Cosmic Coasters.
I have started calling BradS BradS instead of just Brad so he won't be confused with the New Zealand-bound BradSmith. Confused?
Meta Party GameJeffles was the Meta Devil. Here is his Meta Party Game:Alright, well here is my idea.Whendy played Carcassonne (and did place a road dude, thus entering the Meta Game) and left afterwards, so she is the Meta Devil. List subscribers will have already seen the Meta Game she emailed out before leaving on a trip. |
Kirk Smith, Despair's Book of Dreams and the Sometimes Radio
Clandestine, Music From Home
Rockbusters, SOBGYN
Susan McDonald, tico tico
Pain Teens, Beast of Dreams
Diverse music from Texas (in commemoration of March 6 being the anniversary of the battle of the Alamo).
Rank ratings: 0.7143 JeffF (2 games played) 0.7059 JP (6) 0.6000 JimG (3) 0.4286 Zack (5) 0.4118 RussW (5) 0.2632 Daniel (6) 0.2143 Carly (4) 0.2000 Ben (4) 0.1818 PJ (4) 0.1429 Clay (2) 0.0000 MarkH (2) -0.1111 William (5) -0.1176 BradSmith (5) -0.1429 Marty (2) -0.1923 Steve (8) -0.3333 Fina (3) -0.3448 MarkY (8) -0.4444 Dan (4) -0.5000 RussD (3) -0.6667 PaulM (3) -0.7143 BradS (3) -0.7143 Clayton (2) -0.7500 Whendy (1) |
New Win ratings: 0.6000 JimG (3) 0.5000 Ben (4) 0.4286 JeffF (2) 0.3571 Carly (4) 0.3529 JP (6) 0.2105 Daniel (6) 0.1765 BradSmith (5) 0.0588 RussW (5) 0.0000 MarkH (2) -0.0714 Zack (5) -0.0769 Steve (8) -0.1379 MarkY (8) -0.1667 Fina (3) -0.1667 PaulM (3) -0.2222 Dan (4) -0.2222 William (5) -0.2500 Whendy (1) -0.2857 Clay (2) -0.2857 Clayton (2) -0.2857 Marty (2) -0.3636 PJ (4) -0.3750 RussD (3) -0.4286 BradS (3) |
I spent a frustrating few hours experimenting with the planned revision to the ratings system, but sadly it's not ready for primetime yet. Grr. One side effect of this was reminding me again why I burned out on programming, but the main issues are the math formulas, not the implementation. One of the design goals is that playing lots of games shouldn't affect your rating if you perform "as expected". The existing system just adds your devil points from each games, and this meets the goal easily since the mean is clearly 0. Trying to map this into a multiplicative system (which I think would better capture the notion that winning a 2-player game 7 times is way less likely than winning a single 8 player game even though you get the same rating from either feat in the current system) is proving challenging. My notion is to consider, for each game, what was the probability that you'd do as well as you did (e.g. winning a 2 player game has p=0.5; losing a 2 player game has p=1.0; winning an 8 player game has p=0.125; etc.), and then multiply your probabilities together (so the smaller your score, the better you did at "beating the odds"). The catch is then attempting to normalize that score so that you can't improve your rating simply by playing lots of games even if you lose most of them. The idea I had was to multiply each probability by a normalizing constant (which depends solely on the number of players in the game). This should then have the nice property that if you win a 2-player game then lose a 2-player game, the net effect would be multiplying by 1 (instead of by 0.5). The obvious constant for 2-player games is square-root-of-2. The problem is that this assumes the only possible game devil point results are +1 and -1, and forgets about ties. One could derive the appropriate constant with the assumption that +1, 0, and -1 are all possible results... but for most games, it is a bogus assumption that they are equally likely. I'm starting to think this framework can't work, and I need some cleverer way to normalize the results. Details left as an exercise for the reader.
Progress Quest, the noninteractive CRPG for busy gamers.
Photos of tugboat having a bad day.
Ken Thompson's Turing Award speech proves why you cannot trust any software you didn't write yourself.
Cthulhu Pagan Whately Nativity Scene Art is a cool hand-made sculpture that was auctioned on ebay. If that link dies, try here.
Get lost!
Check out my webcomics.