Enhancement: AI Earl should buy a chapel or other card destroyers

[GoogleCodeExporter]

The AI of Earl is about winning with almost no kingdom cards.

This can be enhanced by these four steps:

1) Buy a chapel if possible and destroy all copper and estates (destroy all 
copper till you have at least ten monetary units in the deck left).

2) If curses and card destroyers are in the game and a opponent buys a attack 
card buy a card destroyer (of course the chapel is preferred if existent).

3) Avoid buying any new copper. Do this only with a coppersmith or with a 
chancellor together with a pawnshop

4) Try to buy more Gold in the beginning instead of duchys (only till a 
specific amount of rounds which should be evaluated)

Earl is a strong AI but it will almost definitely loose against curses or other 
attacks. If I'm mistaken please just leave a comment why :-)

Original issue reported on code.google.com by A.R.Brid...@googlemail.com on 20 Sep 2012 at 4:04

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

I think the point of Earl is to be a "buy only treasure/victory cards", not 
necessarily an optimal minimum-action AI. That is, the "Earl" strategy is 
identical in every single Kingdom, and against a number of beginner strategies 
(such as the "buy the most expensive card I can currently afford, play every 
action I can in my hand, preferring the most expensive" that almost every 
beginner starts by trying) tends to be consistently stronger across numerous 
Kingdoms.

I tend to use it as a sort of baseline comparison for strategies - if my 
strategy can be beaten by Earl, then it's obviously suboptimal, because Earl 
uses the exact same strategy on every single Kingdom, and thus can *always* be 
implemented.

A new AI which implements more-optimal minimum-purchase strategies appears to 
have been checked in to the DA branch as "Patrick", which seems to be what 
you're looking for, so I would recommend against changing the Earl AI.

Original comment by Nicholas...@gmail.com on 15 Oct 2012 at 2:00

[GoogleCodeExporter]

Original comment by tkdenni...@gmail.com on 21 Oct 2012 at 5:52

• Added labels: Component-AI

[GoogleCodeExporter]

Earl is definitely not exclusively "Big Money" (a strategy that buys only money 
and victory cards). Earl will buy Witch, Library, Gardens, Moneylender, and I 
think Adventurer. On many boards this will be reasonably competitive (2-Witch 
and 2-3 Library especially so), though in general Earl is not a particularly 
strong AI -- a good player ought to beat Earl >90% of the time in 2p full 
random sets. See enhancement 127, 
http://code.google.com/p/androminion/issues/detail?id=127, for a simple AI that 
would probably beat most players most of the time.

As for the suggestions above:

1. Chapel, while extremely strong, is a fairly mediocre addition to an AI that 
buys almost entirely money and victory cards. The reason is that it first takes 
you a few turns to trash out the coppers and estates; on these turns you're not 
buying anything so your deck is actually getting worse. Then, once you have a 
bunch of golds and silvers, buying victory cards quickly dilutes your deck and 
makes it hard to hit $8 again. Plus, if you split the Provinces, you lose to 
the player who didn't trash all his Estates.

2. In general trashing cards are insufficient to keep up with Curse-giving 
cards. Chapel and Masquerade are the two exceptions that come to mind, and 
there are a few other more specific examples, like Lookout v. Sea Hag, but in 
general a deck with a couple of Cursers will beat a deck with a couple of 
trashers. Play a friend where one of you is allowed to buy a Steward or a 
Bishop, for example, and the other is allowed to buy any Curser, and see who 
wins a best of 5.

3. Not buying copper is a fine idea, but I'm pretty sure Earl already doesn't 
buy Copper (even in cases where he should, like if he has a bunch of Gardens 
and draws a hand of $1). There are only a few cases where I'd buy Copper, and 
Coppersmith by itself or Chancellor/Pawn by themselves aren't one of them. 

4. I think Earl already prefers Gold to Duchies until a set number of rounds 
has passed. A better strategy would be to prefer Gold to Duchies until a set 
number of Provinces have been bought (3-4 in a 2p game).

Original comment by wesc...@gmail.com on 5 Nov 2012 at 7:24