Dominion Strategy Forum

Please login or register.

Login with username, password and session length

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.

Messages - DStu

Filter to certain boards:

Pages: 1 2 [3] 4 5 ... 98
51
PS the Dominion Online logs don't show how many piles are empty at the end, so the checking for the endgame is done in this order:
- did a player resign?
- was the string "gains Colony" detected enough times to empty the Colony pile?
- was the string "gains Province" detected enough times to empty the Province pile?
- it must have ended due to piles

edge-case: Ambassador.

52
I would conjecture that strong players and strong boards both push games towards 3-piles.

Also games with 1 strong and 1 weak player would 3-pile quite often.

It was already when you compare the 2013 to the 2010 numbers , 3 pile got more common.  I would agree that this trend continued.

53
I'm talking about ending on three piles, province depletion or colony depletion. Anyone got any numbers on it, or are able to check? I've looked at council room but can't see a way to check it on there.

I did this like 3 years ago, that has certainly changed, but I'll try to find it anyway.

:e Ok, this was a 1day statistics 2.25 years ago...  I'm quite sure i xposted it here, but it's easier to find something in my posts on bgg...

"Third column is percentage ignoring resigns to compare to the old data [That was some reply on some even older statistic, where resigns were not considered]."

Province game: (13224 games)

Province:   9058   68,5%   79,6%
3pile:      2323   17,6%   20,4%
resign:   1843   13,9%   ---


Colony game: (3042 games)

Colony:      1457   47,9%   57,7%
Province:   349   11,5%   13,8%
3pile:      718   23,6%   28,4%
resign:    518   17,0%   ---

54
General Discussion / Re: More Maths That I Could Use Help With
« on: May 13, 2015, 04:50:43 pm »
The pain in the the back of my brain tells me this must be vba, and afaik Google docs does not support vba. So you need office

55
Puzzles and Challenges / Re: Most points without a decision?
« on: May 12, 2015, 08:02:10 am »

Prince->TR->Golem->[[TR,TR],[TR,TR]]->[Golem->[[CR,CR (draw 2CR and 6Pla)],[TR,TR]->[CR,CR (draw some more Plat and 3 Bishops)]],[play 3 Bishops with the remaing 3 TRs and trash 6 Plat] for 30 points.


feel free to iterate for more

56
Game Reports / Re: Worth buying King's Court?
« on: April 28, 2015, 11:17:25 am »
It doesn't matter how much VP you have at the end. Winning is winning.

Who was it that said the object of Dominion is not to have the most points at the end of the game, but to end the game while you have the most points?
To be fair, the engine bot is still doing the first one. It is just not having the most points the most time before the game ends.

57
Game Reports / Re: Worth buying King's Court?
« on: April 27, 2015, 11:03:54 pm »
Interestingly, although @SheCantSayNo's bot beats either of my original bots 60-40, its average VP is lower at every turn. The games in which SheCantSayNo's bot is ahead must tend to end early.



You should look at the average VP score of each bot on turn 10, 11, etc. just in games that end on that turn.

If you compare a money bot to an engine bot that can end the game while it is ahead, the money bot is ahead throughout the game each time it wins. The engine bot is probably ahead only for the final turn for most of the games it wins. But the engine bot ends the game on the turn it gets ahead, which is worth way more than being ahead on turns 3-15 only to lose later on!

That's true. But the graph is drdndard dominiate output, and it's the first time I saw this effect there, that the winning bot has constantly lower average vp. Do nice demonstration of the principle.

58
Game Reports / Re: Worth buying King's Court?
« on: April 27, 2015, 02:28:16 pm »
Here's a very basic and rough engine script that utilizes Lookout and beats both scripts in the OP:

Code: [Select]
# "Simulators can't play engines!"
{
  name: 'KC-Engine'
  author: 'SheCantSayNo'
  requires: ['Lookout', 'Caravan', 'Bazaar', 'Festival', 'Embassy', "King's Court"]
  gainPriority: (state, my) -> [   
    "Festival" if my.countInDeck("King's Court") > 1 and my.countInDeck("Festival") == 0
    "King's Court" if my.countInDeck("King's Court") < 5
    "Colony"
    "Province" if state.gainsToEndGame() <= 3
    "Bazaar" if my.countInDeck("Bazaar") < 3   #5-2 opening
    "Embassy" if my.countInDeck("Embassy") == 0 
    "Caravan" if my.countInDeck("Caravan") < 1       #4-3 opening
    "Lookout" if my.countInDeck("Lookout") < 1       #4-3 opening
    "Festival" if my.countInDeck("Festival") < 2
    "Festival" if my.ai.drewDeck(state, my) == 1   
    "Duchy" if state.gainsToEndGame() <= 2
    "Bazaar"
    "Caravan" if state.gainsToEndGame() > 2
    "Silver" if state.gainsToEndGame() > 2
    "Estate" if state.gainsToEndGame() <= 2
  ]

drewDeck: (state, my) ->
    draw = my.cardsInDeck
    for card in my.hand
        draw--
    for card in my.inPlay
        draw--
    if draw == 0
        return 1
    else
        return 0

multipliedPriority: (state, my) -> [
      "King's Court"
      "Caravan" if my.ai.drewDeck(state, my) == 0
      "Bazaar"  if my.ai.drewDeck(state, my) == 0
      "Festival"     
      "Embassy" if my.ai.drewDeck(state, my) == 0
      "Bazaar"
      "Caravan"
      "Lookout" if state.gainsToEndGame() >= 6
]

playPriority: (state, my) -> [   
       "King's Court"
       "Lookout" if state.gainsToEndGame() >= 4
       "Caravan"
       "Bazaar"
       "Embassy"     
       "Silver"
       "Copper"
   ]
}

Try skipping Lookout to see how much poorer it does.

To even forgot to play festival of not multiplied. Or was this on intention?

59
Game Reports / Re: Worth buying King's Court?
« on: April 27, 2015, 08:05:09 am »
What else could you even need to make this a King's Court board?

Bridge  ;)

60
Game Reports / Re: Worth buying King's Court?
« on: April 26, 2015, 04:10:57 pm »
You can't really simulate King's Court strategies.

You can, it's just lots of work.

:e seriously.  You probably really want to specify your playPriority if you are building this kind of deck, but this can get really messy to decide when to play KC, and what to KC,...

61
Simulation / Re: Dominiate: a Dominion simulator that runs on the Web
« on: April 26, 2015, 03:04:32 pm »
Yeah, why and since when don't we call actionPriority anymor when it exists? It feels like playPriority is much more of a hassle to define costumely, as you also have to care for tje Treasures.  So why is this mixed?

62
Simulation / Re: Dominiate: a Dominion simulator that runs on the Web
« on: April 26, 2015, 02:15:30 pm »
Thanks, DStu. Is there a way I can patch it for myself if I'm running from a browser? Just dumping the fixed makeCard code into the "Player 1" window in my browser doesn't work.

I don't thin you can patch, but you can workaround:

Code: [Select]
{
  name: 'LookBug'
  requires: ['Lookout']
  gainPriority: (state, my) -> [
    "Colony" if my.countInDeck("Platinum") > 0
    "Province" if state.countInSupply("Colony") <= 6 \
               or state.countInSupply("Province") <= 6
    "Duchy" if 0 < state.gainsToEndGame() <= 5
    "Estate" if 0 < state.gainsToEndGame() <= 2
    "Platinum"
    "Gold" if my.countInDeck("Gold") < 2
    "Lookout"
    "Silver"
  ]

  actionPriority: (state, my) ->
    if state.gainsToEndGame() > 5 or ...
       "Lookout"
   
}

so just add an actionPriority function to your bot, which returns a list containing the priorities you want your cards played. Note that you have to add all the cards of you bot there, as this overwrites the call of ai_value from cards (IIRC).  You might want to look into old_actionPriority in basicAI.cofffee if you have some more cards in your bot to see how they used to be prioritized.
Maybe
Code: [Select]
  actionPriority: (state, my) ->
    old_actionPriority(state,my)
does also work, at least Lookout is implemented correctly there, but there might be some newer cards (like from Guilds and Dark Ages) that are not listed there anymore.

:e corrected name of old_actionPriority.  Maybe you also have to reference it over the ai, in this case it is
Code: [Select]
my.ai.old_actionPriority(state,my)
I'm not sure what exactly the scope is from which the code is called here.

edit: OK, actionPriority seems to be not called anymore, it's now playPriority if I understand this correctly. Problem is that this one is merged with the Treasures, so you must inculde all the treasures in your list, too, otherwise the bot does not play Treasures.

63
Simulation / Re: Dominiate: a Dominion simulator that runs on the Web
« on: April 26, 2015, 01:47:05 pm »
Found the bug.

My version is too far behind to patch fastly, but patch is in ai_value from Lookout
Code: [Select]
-- if state.gainsToEndGame >= 5 or state.cardInfo.Curse in my.draw
++ if state.gainsToEndGame() >= 5 or state.cardInfo.Curse in my.draw

gainsToEndGame is a function.

64
General Discussion / Re: Stop Shuffling Overhand
« on: April 25, 2015, 03:17:23 pm »
The operation that reverses pile shuffling is more pile shuffling.

If you pile-shuffle n card in k piles, and do this i(n,k) times, you get back to the original order.  That this is true is trivial, there are at most n! orderings, and as it is deterministic, it must repeat itself eventually*.  But i(n,k) is way smaller than n!.
Also, you could revert a pile shuffle by taking the "shuffled" deck, take k cards each, put them into a pile and than draw 1 card from each pile rotating. So just think on what you have to do to perform the pile shuffle backwards.

But shuffling is not commutative, so if you plan to get a bad deck by doing a reverse pile shuffle, after "pile shuffling then bad shuffling", so have P*B*P^(-1),you won't get B (just bad shuffle).


So why pile shuffling at all? In Dominion, it also has it's merit even if it does not add randomness. This is because you have many similar cards, and usual shuffling techniques will work better if these similar cards are distributed around the deck. See David desJardins explaining it to Donald

:e Forgot the *: Of course just by this it is not entierly clear that the cricle starts at the beginning, this is left as exercise to the reader.

66
Dominion General Discussion / Re: Interview with Donald X.
« on: April 25, 2015, 05:27:31 am »
What about recent/realtime sales (rather than cumulative)? Do people still buy more Intrigue than Seaside, or is it weighed more heavily toward the better reviewed sets? I would expect the current best selling ones would always be the base set and whatever the newest expansion is at the time (?).
let's stop talking about it.

amazon(.de) says (for the German edition):
Dominion: rank 778 in toys
Cornucopia + Alchemy: rank 2.017
Seaside: rank 4.226
Guilds: rank 10.436
Intrigue: rank 28.277
Hinterlands: rank 37.098
Dark Ages: rank 31.582
Prosperity: rank 40.540
Alchemy: rank 43.465

of course amazon might not produce unbiased samples. Especially, as some of these were not avaiable at amazon directly but at resellers, sometimes for more than 100€/expansion.

Honorable mention:
Dark Ages: rank 196.598 in kitchen & household



:e also, German edition seems to be 13+ for all of these the newer editions and 8+ for the old ones.

67
Goko Dominion Online / Re: Features Thread
« on: April 25, 2015, 05:07:28 am »
I haven't been following this thread too closely, but has Linux support been confirmed yet? I would have no way of playing otherwise.

At the very least you should be able to able to play in a browser, because we'll target Unity 5's WebGL build, which just needs any browser supporting WebGL and asm.js.

Incidentally, ignore anything you read about the Unity webplayer and Linux, because that's old tech. It's based on the Netscape plugin API which is being phased out in browsers these days due to the security holes it creates.

Sweet!  The post I read about you guys using webplayer was months ago, if you've decided to step up to WebGL that's great news for us Linux users!

Due to your guys' interest in this, we tried creating a native build for Linux for the first time today. It works.
theblankman, if you would like to beta-test (really alpha-test) the Linux build, please message me with your email.

If you need some more people to install a Linux version I would volunteer, but I don't have so much time to play and test, so if you want to limit your alpha-users, you better might take someone else.. ;)

68
Goko Dominion Online / Re: Features Thread
« on: April 25, 2015, 04:52:15 am »
How did you know I found the league? You are really paying attention  :)


69
Goko Dominion Online / Re: Features Thread
« on: April 24, 2015, 04:26:49 am »
2. Change the booting feature.... unless you are deliberately trying to lose customers? OK, lets say I am new and come to multiplayer for the first time, here's what happens. I go into a room, join a game, and BAM, "so-and-so has kicked you out." Try another table and, BAM same thing. Why is everybody kicking me out? Why does nobody like me? What have I done wrong? Why are there all these different rooms? What do they mean? Why can't I enter some of the rooms? Am I being banned? Do people not want me in this room? I guarantee you that you are losing thousands of potential players and customers right in this moment. The message should say "We are sorry your rating does not qualify to sit at this table, please visit our beginners table" AND there needs to be a special space for beginners.
That should be solved by a good auto-match system, where you just click "search for game", and than get paired with a player of approximately your skill, which they also can't reject anymore (except by resigning the game and therefore losing it)

70
Goko Dominion Online / Re: Features Thread
« on: April 19, 2015, 03:15:45 am »
I avoid the alchemy set because i have heard it doesn't mix well in full random.
it does

71
Dominion: Adventures Previews / Re: Events on the wiki!
« on: April 18, 2015, 11:04:35 am »
The bot should now also transform [[$event]] to {Event|$event} for each $event that I have not misspelled.

It also doesn't touch the Isotropic art gallery anymore

72
Goko Dominion Online / Re: Features Thread
« on: April 18, 2015, 09:42:34 am »
Thx Lord Humanton for the detailed answer on what exactly you are doing. Going to greenlight now.

Regarding log/deck tracker
If you keep the searchable full game log, just add a deck tracker that displays how many of each card are in everyones deck. Either that information should not be available at all, or it should be available without the hassle of going through the entire log.

This. Also, maybe fork (what have I done? should have known better...)

:e also, I think that should be decided ex cathedra by Donald.  He has stated often enough that these are all variants, and it is his to decide which variant we play online.

73
Goko Dominion Online / Re: Features Thread
« on: April 18, 2015, 05:55:55 am »
I'm not sure if it was stated here, but if we are talking about an always visible game log, we are talking about an always visible log of the current and last turn (I assume).  Otherwise you have people all the way scrolling back to turn 1 to count how many points they have, which is not what you want.

Especially not in Garden games etc., as this just takes a lot of time.

74
Dominion Articles / Re: Magpie
« on: April 17, 2015, 04:42:39 am »
Empty deck, YES but NO...

Because If your deck is empty, no treasure.
No treasure, no way to buy another card.
No way to buy, no victory card.
No victory card, no way to win...  ;D

Not completely trashed, completely drawn. Except for 2 Actions.

75
Dominion Articles / Re: Magpie
« on: April 17, 2015, 04:29:20 am »
Seems legit. Like flying rats...



:e Should, similar to rats, also work with an almost empty deck instead of Watchtower, or? (Details to the reader)

Pages: 1 2 [3] 4 5 ... 98

Page created in 0.072 seconds with 19 queries.