901
Game Reports / Re: Familiar on a strong Remodel board
« on: June 25, 2011, 02:58:44 pm »
I think if you hit the turn 4 familiar, you are would have breezed to the win.
Show Posts
This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.
902
General Discussion / Re: Math Nerds...
« on: June 24, 2011, 06:56:40 pm »
I did a math minor during my undergrad. But as I read more about machine learning and statistics, I am beginning to think I am really a statistician who happens to have nice employment as a software engineer.
903
Dominion General Discussion / Re: Ironworks + Great Hall
« on: June 24, 2011, 04:49:09 pm »
guided, do you think you'll beat an ironworks, great hall, estate fast end game with standard play for most provs?
Looking at your own game data, it seems like you don't give this combo quite enough credit in your post. You do fairly often go for multiple ironworks for great halls.
It's a pretty even spread between you going for it, you ignoring it, you winning when you ignore the combo and your opponent goes for it, you losing when you ignore it but your opponent doesn't, etc. It's basically what you'd expect from a solid but not crazy dominating combo.
http://councilroom.com/search_result?p1_name=guided&p2_name=&kingdom=ironworks%2Cgreat+hall
Looking at your own game data, it seems like you don't give this combo quite enough credit in your post. You do fairly often go for multiple ironworks for great halls.
It's a pretty even spread between you going for it, you ignoring it, you winning when you ignore the combo and your opponent goes for it, you losing when you ignore it but your opponent doesn't, etc. It's basically what you'd expect from a solid but not crazy dominating combo.
http://councilroom.com/search_result?p1_name=guided&p2_name=&kingdom=ironworks%2Cgreat+hall
904
Dominion General Discussion / Re: Food For Thought: When to take the Curse for Torturer?
« on: June 24, 2011, 04:15:04 pm »
This was the absolute best "gems to gemonade" turn I've ever had playing Dominion. I turn something really bad (gaining 3 curses), into something really awesome (KCing a torturer).
I get hit by a KCed torturer, gain 3 curses, cellar off 6 cards from my deck, and turn the tables, hitting my own KCed torturer and finishing off by trading post away the 3 curses I picked up at the start of the turn.
http://councilroom.com/game?game_id=game-20110417-171217-038cf33a.html#mlo-show-turn-15
I get hit by a KCed torturer, gain 3 curses, cellar off 6 cards from my deck, and turn the tables, hitting my own KCed torturer and finishing off by trading post away the 3 curses I picked up at the start of the turn.
http://councilroom.com/game?game_id=game-20110417-171217-038cf33a.html#mlo-show-turn-15
905
Dominion Isotropic / Re: Isotropic Facts & Quirks
« on: June 24, 2011, 04:06:15 pm »
Software that requires you to read a manual tends to be pretty crappy software. All other things the same, you'd prefer not to have to do it.
Of course, there is (often) a trade off between simplicity of the software and fluidity of the user experience.
I love isotropic and it's super lean interface, but I think the way it handles moat (and pawn, remember my last choice and preselect it next time, please!) is annoying. If you have ever moated an attack, you don't need the option to moat again, regardless of funky interactions with secret chamber. If you have no more meaningful options, unblock the game and progress.
Of course, there is (often) a trade off between simplicity of the software and fluidity of the user experience.
I love isotropic and it's super lean interface, but I think the way it handles moat (and pawn, remember my last choice and preselect it next time, please!) is annoying. If you have ever moated an attack, you don't need the option to moat again, regardless of funky interactions with secret chamber. If you have no more meaningful options, unblock the game and progress.
906
Council Room Feedback / The CouncilRoom Simulation is Prediction challenge
« on: June 24, 2011, 03:45:08 pm »
rspeer has a dominion bot. I have been playing with doing winningness prediction. Geronimo has a dominion simulator. I have a bunch of so far not all that interesting game result pages.
Why not give users an easy way to see who is going to win the game and also answer the question of which system really understands Dominion the best?
For rspeer's system and my own system, we just need to feed the game state to our predictors and see what they think. For Gerinomoo's system (and other simulators), we need to be able give it a mid-game game state, and simulate the outcome many times (perhaps expose strategy choices as url params), and then use probability of winning as the predictor.
Other contestants are welcome to join as well.
To get this working really requires us agreeing on an encoding of the mid game states. I guess it also requires some work for Gerinomoo to support starting games from the middle and exposing a non GUI interface.
Why not give users an easy way to see who is going to win the game and also answer the question of which system really understands Dominion the best?
For rspeer's system and my own system, we just need to feed the game state to our predictors and see what they think. For Gerinomoo's system (and other simulators), we need to be able give it a mid-game game state, and simulate the outcome many times (perhaps expose strategy choices as url params), and then use probability of winning as the predictor.
Other contestants are welcome to join as well.
To get this working really requires us agreeing on an encoding of the mid game states. I guess it also requires some work for Gerinomoo to support starting games from the middle and exposing a non GUI interface.
907
Feedback / Re: Isotropic feedback?
« on: June 23, 2011, 10:46:40 pm »
I think he just means requiring a specific set of 10 real cards.
There is a card requirement box that just works.
There is a card requirement box that just works.
908
Game Reports / Re: Ambassador/Ambassador vs. Ambassador/Silver
« on: June 23, 2011, 05:25:50 pm »
+10 for posting a game report in which you didn't win 
I agree that the double ambassador buy turn 16 going overboard.
I don't know if you had the opportunity here, but a good way to beat a super ambassador player who is spamming you with estates is to pick up a salvager and start salvaging provinces for more provs. I am almost always the hyper amb player, but the turbo salvage provs for end game sometimes catches me. Usually the hyper amb player is going for a super long, 25+ turn game to make their investment pay off, and if you make the game not go super long, you take away the return on their investment.
I agree that the double ambassador buy turn 16 going overboard.
I don't know if you had the opportunity here, but a good way to beat a super ambassador player who is spamming you with estates is to pick up a salvager and start salvaging provinces for more provs. I am almost always the hyper amb player, but the turbo salvage provs for end game sometimes catches me. Usually the hyper amb player is going for a super long, 25+ turn game to make their investment pay off, and if you make the game not go super long, you take away the return on their investment.
909
Rules Questions / Re: Colonies/Platinum on Isotropic
« on: June 23, 2011, 03:11:04 pm »
Yes, the random proportion is the rule used by isotropic.
910
Dominion General Discussion / Re: >2 Treasure Maps?
« on: June 23, 2011, 02:37:36 pm »I still like the very simple play: Talisman, Royal Seal, Copper. Buy a Treasure Map...get another one...both go on the deck.
Indeed
911
Simulation / Re: Dominion Simulator available for download!
« on: June 23, 2011, 09:52:31 am »Well, I've never seen a genetic algorithm that was worth the time it takes to run, but I am using a different machine learning technique (stochastic gradient descent) in my code to determine what to buy. It's basically just fitting a hugely multivariate function whose input is your deck and an opponent's deck, and whose output is the probability that you win against that opponent.
It seems like I could adapt this kind of thing to the play of an individual hand, by making the input be your hand, number of actions, cards in play, etc. and the output be... money? Value of cards gained/bought? Improvement in your win probability as determined by the other function? I kind of like that last one.
What I would need is a lot of data about the intermediate stages one's hand goes through. I don't think dominionstats (the CouncilRoom code) currently does this; it only tracks the changes to decks, not hands. Is there existing code that can parse this from the game logs, or would I have to write it?
The iso logs don't contain enough information to get card play by card play game state info. You don't see what cards people draw when they play +card actions
.
912
Simulation / Re: Dominion Simulator available for download!
« on: June 23, 2011, 01:56:25 am »
It's near the end game, you want to find a Platinum to buy the final Colony, you play the Lookout to draw it near, and then the Pearl Diver to take it into your deck.
913
Dominion Articles / Re: Combo: University/Watchtower/X
« on: June 22, 2011, 10:05:29 pm »
Yeah, I like this a lot. I also like it with ironworks (even though it's not as good without the action fork).
914
Simulation / Re: Dominion Simulator available for download!
« on: June 22, 2011, 10:03:45 pm »
The perfect is the enemy of the good.
I'd much rather see a system where we leverage the incremental improvements of 10 or 20 skilled players gradually improving the play heuristics and getting better and better bots than us just giving up because getting it perfect is too hard.
I'd much rather see a system where we leverage the incremental improvements of 10 or 20 skilled players gradually improving the play heuristics and getting better and better bots than us just giving up because getting it perfect is too hard.
915
Dominion General Discussion / Re: 3-player-isotropic game ranking + draws
« on: June 22, 2011, 06:50:51 pm »
The clearest heuristic reasoning I can give is due to the saturation effects (like the 2 max vs 3 max) that I gave above.
Consider the limit case where a superstar wins 99% of games against average players in 2p games. Each average individual has a 1 in 100 chance of winning the lottery or hitting the hail mary to win. You might expect him to win 98% or something like that of his 3p games (he just needs to dodge 2 independent, rare events instead of 1), and so his 2p and 3p win rate would be something like 1.98 and 2.94 respectively.
The behavior stays at lesser skill differentials, but obviously to a lesser degree. I suggest you run/tweak this small program until you believe me, or can come up with a similarly small program that shows that I am wrong.
output:
[1.1701600000000001, 0.82984000000000002]
[1.3330200000000001, 0.83103000000000005, 0.83594999999999997]
[0.83675999999999995, 0.83099999999999996, 1.3322400000000001]
Consider the limit case where a superstar wins 99% of games against average players in 2p games. Each average individual has a 1 in 100 chance of winning the lottery or hitting the hail mary to win. You might expect him to win 98% or something like that of his 3p games (he just needs to dodge 2 independent, rare events instead of 1), and so his 2p and 3p win rate would be something like 1.98 and 2.94 respectively.
The behavior stays at lesser skill differentials, but obviously to a lesser degree. I suggest you run/tweak this small program until you believe me, or can come up with a similarly small program that shows that I am wrong.
Code: [Select]
#!/usr/bin/python
import random
GOOD_P_MAX = 120
AVG_P_MAX = 100
def sim(strength_list):
outcomes = [random.random() * s for s in strength_list]
winner = max(outcomes)
# ignore ties, will basically never happen with random floats anyway.
return outcomes.index(winner)
def sim_many(strength_list, N):
ret = [0 for i in strength_list]
for i in xrange(N):
ret[sim(strength_list)] += len(strength_list)
for ind, v in enumerate(ret):
ret[ind] = v / float(N)
return ret
print sim_many([GOOD_P_MAX, AVG_P_MAX], 100000)
print sim_many([GOOD_P_MAX, AVG_P_MAX, AVG_P_MAX], 100000)
# sanity check that order doesn't matter
print sim_many([AVG_P_MAX, AVG_P_MAX, GOOD_P_MAX], 100000)
output:
[1.1701600000000001, 0.82984000000000002]
[1.3330200000000001, 0.83103000000000005, 0.83594999999999997]
[0.83675999999999995, 0.83099999999999996, 1.3322400000000001]
916
Variants and Fan Cards / Re: What would you pay for a non-recursive KC?
« on: June 22, 2011, 05:27:43 pm »
Can you KC a Golem a KC?
mutual recursion for the win ;P
mutual recursion for the win ;P
917
Dominion General Discussion / Re: 3-player-isotropic game ranking + draws
« on: June 22, 2011, 05:22:43 pm »
I haven't thought much about trying to climb the isotropic ladder by playing 3p games rather than 2p games.
For skilled players, avg win points will tend to be higher for 3p than for 2p (the max for 3p is 3, max for 2p is 2), so that alone should not make you think you are better when playing 3p than 2p.
From a game design rather than statistics sense, I dislike encouraging players to take a guaranteed 2nd rather than a possible first. If I designed the isotropic rating system, I'd make 2nd == last for multiplayer games.
For skilled players, avg win points will tend to be higher for 3p than for 2p (the max for 3p is 3, max for 2p is 2), so that alone should not make you think you are better when playing 3p than 2p.
From a game design rather than statistics sense, I dislike encouraging players to take a guaranteed 2nd rather than a possible first. If I designed the isotropic rating system, I'd make 2nd == last for multiplayer games.
918
Dominion General Discussion / Re: 3-player-isotropic game ranking + draws
« on: June 22, 2011, 03:28:13 pm »
Win points == total winners / total players. I've forgotten how I account for partial wins on the record, but win points are a better measure than win/loss/ties anyway.
On isotropic, from reading dougz's trueskill code, the order matters for multiplayer games. It's better to finish 2nd than 4th. Because of the complicated nature of trueskill, finishing 2nd in a 4 player game is not the same as losing a 2p game against the winner, and winning two 2p games against the 3rd and 4th players.
If you tie with a lower rated player, you can lose points.
On isotropic, from reading dougz's trueskill code, the order matters for multiplayer games. It's better to finish 2nd than 4th. Because of the complicated nature of trueskill, finishing 2nd in a 4 player game is not the same as losing a 2p game against the winner, and winning two 2p games against the 3rd and 4th players.
If you tie with a lower rated player, you can lose points.
919
Dominion Articles / Re: Combo: Fortune Teller/Jester
« on: June 22, 2011, 01:00:12 pm »
Is fortune teller/jester another scout/wishing well? In theory it sounds good but in practice it's a lot of work for something that a single card does just as well.
Consider fortune teller/jester vs mountebank. FT -> Jester is a lot harder to pull off, and guarantees your opponent gets a curse. An unblocked mountebank gives your opponent a copper and a curse, which is a much stronger attack.
Consider fortune teller/jester vs mountebank. FT -> Jester is a lot harder to pull off, and guarantees your opponent gets a curse. An unblocked mountebank gives your opponent a copper and a curse, which is a much stronger attack.
920
Dominion Articles / Re: Combo: Hamlet / Watchtower or Library
« on: June 21, 2011, 04:47:21 pm »
The Hamlet and Watchtower/Library combo is totally solid. Cards that decrease hand size (and hopefully net actions) work really well with Library and Watchtower. Fishing Village, Upgrade, Shanty Town, even Vault, Remake, or Secret with some action forks do this well.
It's similar to the three piece combo of Vault/Library/+Action.
I noticed that Cornucopia has a lot of self hand size decreasing actions (hamlet, remake, young witch, horse traders), which I suspect are part of the design to make Menagerie combos work.
It's similar to the three piece combo of Vault/Library/+Action.
I noticed that Cornucopia has a lot of self hand size decreasing actions (hamlet, remake, young witch, horse traders), which I suspect are part of the design to make Menagerie combos work.
921
Variants and Fan Cards / Re: What would you like to see in Dominion?
« on: June 21, 2011, 10:09:56 am »
I think I could reasonably model player speed just from the existing game logs if I was so inclined.
First realize that many players play sequentially. A sample of 10 games with start and ends is probably not as good as a sample of 100 games with only ends. There are >1 million iso games.
Sequential plays tend to end say, in less than an 30 minutes after the last one.
Games between players will tend to take an average length of the sum of slowness of the participating players.
With a bit of hacking, I am pretty sure an expectation maximization type algorithm could reasonably rank players by speed.
First realize that many players play sequentially. A sample of 10 games with start and ends is probably not as good as a sample of 100 games with only ends. There are >1 million iso games.
Sequential plays tend to end say, in less than an 30 minutes after the last one.
Games between players will tend to take an average length of the sum of slowness of the participating players.
With a bit of hacking, I am pretty sure an expectation maximization type algorithm could reasonably rank players by speed.
922
Dominion General Discussion / Re: Best/Worst Openings discussion
« on: June 21, 2011, 09:20:14 am »
I am especially distrustful of big money simulations for heavy trashing type decks. Heavy trashing -> small deck -> easy combos.
923
Dominion General Discussion / Re: >2 Treasure Maps?
« on: June 20, 2011, 07:56:35 pm »I'd skip the golem and treasure map combination until you get throne-ironworks-watchtower-golem-market in hand.
What does the Golem do there? TR + IW, get two TMs on top of deck, play watchtower to draw them and cash them in. If you go TR + IW, put two TMs on top, and then play golem, you'll just trash them both (neither of them is in hand for the other to combo with).
924
Dominion General Discussion / Re: >2 Treasure Maps?
« on: June 20, 2011, 07:20:16 pm »
Buy a third TM if you hit 4 money and your TMs haven't collided yet.
Golem is actively harmful in a Treasure Map deck. Consider that,
1) the Golem cost you a potion, which could have been a TM. The opportunity cost is high.
2) if the Golem you purchased was instead a TM and it co-occured with a TM already in your hand, you would hit the jackpot anyway.
3) As you've said, if the Golem does not occur with a TM, you risk trashing your treasure maps if you play the Golem..
Golem is actively harmful in a Treasure Map deck. Consider that,
1) the Golem cost you a potion, which could have been a TM. The opportunity cost is high.
2) if the Golem you purchased was instead a TM and it co-occured with a TM already in your hand, you would hit the jackpot anyway.
3) As you've said, if the Golem does not occur with a TM, you risk trashing your treasure maps if you play the Golem..
925
Dominion Isotropic / Re: Decline of civility on isotropic?
« on: June 20, 2011, 04:10:38 pm »
I think tone is being lost in the text.
I was mostly pointing out that starting from Chess, where resigning is common place and as far as I can tell, games rarely end in real checkmates, and then applying social norms from the chess community to Dominion, where resigning is much less common, and some people think it's bad form to ever resign doesn't make a great analogy. In chess, where you expect the opponent to resign when they think the game is lost, the offer is certainly more arrogant in than in Dominion.
I think it's reasonable to believe that people prefer playing Dominion games which are yet undecided compared to playing games where they are almost certainly decided and the game will be a loss. Most people say thanks, gg, when I make the offer to accept resignation. I've never lost after I've made the offer. And occasionally, some "never resign" players go and endure the beating and that's fine. It's a little annoying if they both don't resign and expect me to play for a short end game, but hey, that's their pejorative (they aren't getting short end game until I can be certain that I win). But it's not really arrogant on my part, I just gave them an option. I don't do it to piss off my opponents, I just think the resignation is in our mutual interest, we both get to play the fun part of Dominion, where you don't actually know the outcome of a particular game yet.
I was mostly pointing out that starting from Chess, where resigning is common place and as far as I can tell, games rarely end in real checkmates, and then applying social norms from the chess community to Dominion, where resigning is much less common, and some people think it's bad form to ever resign doesn't make a great analogy. In chess, where you expect the opponent to resign when they think the game is lost, the offer is certainly more arrogant in than in Dominion.
I think it's reasonable to believe that people prefer playing Dominion games which are yet undecided compared to playing games where they are almost certainly decided and the game will be a loss. Most people say thanks, gg, when I make the offer to accept resignation. I've never lost after I've made the offer. And occasionally, some "never resign" players go and endure the beating and that's fine. It's a little annoying if they both don't resign and expect me to play for a short end game, but hey, that's their pejorative (they aren't getting short end game until I can be certain that I win). But it's not really arrogant on my part, I just gave them an option. I don't do it to piss off my opponents, I just think the resignation is in our mutual interest, we both get to play the fun part of Dominion, where you don't actually know the outcome of a particular game yet.