Dominion Strategy Forum

Please login or register.

Login with username, password and session length
Pages: 1 2 [3]  All

Author Topic: Dominion Academy, game #1  (Read 20951 times)

0 Members and 1 Guest are viewing this topic.

meandering mercury

  • Ambassador
  • ***
  • Offline Offline
  • Posts: 34
  • Respect: +36
    • View Profile
Re: Dominion Academy, game #1
« Reply #50 on: January 21, 2013, 01:29:57 pm »
0

I think the conventional wisdom is that the programmed engine will not be able to outplay a top-tier human engine builder like jonts. Although you can include some tricks like filtering over Estates with Courtyard, there will be other subtleties (reshuffle, etc) which are harder to code in.

On the other hand, I think it IS true that a programmed BM strategy will be able to simulate a reasonably good BM player (what the simulator lacks in decision making power, it can make up for by tuning the parameters a bit smarter). The fact that the programmed engine is able to beat the programmed BM is very suggestive to me. If you're talking engine versus engine-BM hybrid, well, that's a different story.

I still agree with Stef that for average players, even for top-tier players, BM is safer than engine just because engine is trickier to set up and requires much more skill to execute.

WW, in the two games you cited against jonts, one ended on piles and the other ended in 21 turns.
Logged

jonts26

  • Margrave
  • *****
  • Offline Offline
  • Posts: 2746
  • Shuffle iT Username: jonts
  • Respect: +3668
    • View Profile
Re: Dominion Academy, game #1
« Reply #51 on: January 21, 2013, 01:42:36 pm »
0

WW, in the two games you cited against jonts, one ended on piles and the other ended in 21 turns.

The two games he posted were ones where WW didn't go CY/BM. In a mirror, it's totally reasonable for piles to run. And in the second game, I think I tried a couple things which didn't work, or I got bad luck. Plus he was stealing my grand markets which really slow me down some. Anyway that was probably my slowest game of the like 7 we played. The uncontested engine can win in 18-19 turns usually no problem.

As for geronimoo's simulator, I'm wondering where you can make up an extra couple of turns. I found getting 4-5 schemes to be very helpful in ensuring consistency. Also, a cellar after 5 or so courtyards was helpful. So I feel the second Ironworks can be useful to pick up all of those extra components. Maybe you could give more details on the buy rules you employed and we can improve it some.
Logged

WanderingWinder

  • Adventurer
  • ******
  • Offline Offline
  • Posts: 5275
  • ...doesn't really matter to me
  • Respect: +4381
    • View Profile
    • WanderingWinder YouTube Page
Re: Dominion Academy, game #1
« Reply #52 on: January 21, 2013, 01:43:26 pm »
0


WW, in the two games you cited against jonts, one ended on piles and the other ended in 21 turns.
Right, but in both those games I was playing hybrids, and that slowed him down. A LOT.

Geronimoo

  • Saboteur
  • *****
  • Offline Offline
  • Posts: 1059
  • Respect: +868
    • View Profile
    • Geronimoo's Dominion Simulator
Re: Dominion Academy, game #1
« Reply #53 on: January 22, 2013, 04:43:08 am »
+3

OK, so I tweaked the tactical decisions of the sim and now I get the engine to beat the BM deck 83-15 on a $4/$3 opening and 93-5 on a $5/$2 opening. The average number of turns has gone down to 19.5, which still seems a little high, but the engine will sometimes have bad turns (Courtyards clumped together) which drags the average speed down.

What still surprises me looking over the logs is that the engine will often spend turns 7-10 hugely overpaying for Courtyards and Farming Villages and that this is the correct way to play it because buying a Gold in those spots will decrease the win rate.

I wasn't able to fit in more than 1 Ironworks or 1 Mine and have the win rate go up.
Logged

krs

  • Pawn
  • **
  • Offline Offline
  • Posts: 3
  • Respect: 0
    • View Profile
Re: Dominion Academy, game #1
« Reply #54 on: February 07, 2013, 11:32:15 am »
0

I have tried in the comments section but no reaction whatsoever. So here it goes:

I am curious if my findings are valid:

Shouldn’t you take into account the fact that you are going 1′st or second? Having 1 extra turn for engine is huge! (All the trial matches linked in forum could have gone either way. First player always won).

Secondly, starting hand is important! If you get 5-2 you have your perfect engine startup so you should go the engine way no matter what. (Or get a mine in “Big Money” and 1 Workers Village. First one helps you contest 1-2 Grand Markets and the other one helps you mine more and/or get some extra 2$ when badly needed => 18-19 turns 8 provinces).
Logged

philosophyguy

  • Minion
  • *****
  • Offline Offline
  • Posts: 575
  • Respect: +299
    • View Profile
Re: Dominion Academy, game #1
« Reply #55 on: February 07, 2013, 12:11:58 pm »
0

OK, so I tweaked the tactical decisions of the sim

When you say that, do you mean that you add rules to the XML that address special cases, or do you do stuff behind the scenes? If the latter, would it be possible for you to write up a post demonstrating how you do that? I'm sure lots of us would like to make those kinds of tweaks (for instance, to test whether the Mountebank/Chapel turn 3 collision should chapel or attack) but I have no idea how to do that.
Logged

Geronimoo

  • Saboteur
  • *****
  • Offline Offline
  • Posts: 1059
  • Respect: +868
    • View Profile
    • Geronimoo's Dominion Simulator
Re: Dominion Academy, game #1
« Reply #56 on: February 07, 2013, 01:15:25 pm »
0

You'll have to adjust the Java code to tweak tactical decisions. The biggest changes I made were adjustments of the "play priority number" and the "discard priority number" (can be found in the "DomCardName" enum):

https://github.com/Geronimoo/Geronimoo-s-Dominion-Simulator
Logged

dondon151

  • 2012 US Champion
  • *
  • Offline Offline
  • Posts: 2522
  • Respect: +1856
    • View Profile
Re: Dominion Academy, game #1
« Reply #57 on: February 07, 2013, 02:14:42 pm »
0

Shouldn’t you take into account the fact that you are going 1′st or second? Having 1 extra turn for engine is huge! (All the trial matches linked in forum could have gone either way. First player always won).

First player advantage is always present, maybe a little moreso for an engine, but by itself, I don't think it's an indicator for either strategy.
Logged

krs

  • Pawn
  • **
  • Offline Offline
  • Posts: 3
  • Respect: 0
    • View Profile
Re: Dominion Academy, game #1
« Reply #58 on: February 07, 2013, 04:33:28 pm »
0

What I want to say/ask is why being first /second is not a variable when making your strategy, if there are multiple close strategies like in this kingdom. To me (novice player) for this particular kingdom being first and starting with 5 Copper screams for engine.
Logged

jonts26

  • Margrave
  • *****
  • Offline Offline
  • Posts: 2746
  • Shuffle iT Username: jonts
  • Respect: +3668
    • View Profile
Re: Dominion Academy, game #1
« Reply #59 on: February 07, 2013, 05:03:58 pm »
0

What I want to say/ask is why being first /second is not a variable when making your strategy, if there are multiple close strategies like in this kingdom. To me (novice player) for this particular kingdom being first and starting with 5 Copper screams for engine.

I think very few kingdoms are going to change substantially depending on whether you get a 3/4 or 5/2 split, though it might alter tactical play fairly often. I see this kingdom and 3/4 also screams engine to me. But if you were convinced of a big money strategy being better, 5/2 still lets you open courtyard/silver so that doesn't even change the big money game.
Logged
Pages: 1 2 [3]  All
 

Page created in 0.114 seconds with 21 queries.