hacker.org Forum Index
RegisterSearchFAQMemberlistUsergroupsLog in
Feature request: alternate red/blue positions in series

 
Reply to topic    hacker.org Forum Index » BitBath View previous topic
View next topic
Feature request: alternate red/blue positions in series
Author Message
Bedevere



Joined: 26 Aug 2007
Posts: 2

Post Feature request: alternate red/blue positions in series Reply with quote
I always keep several old versions of my bot around and test changes that I make to my main bot against other versions to see what sort of a percentage increase in wins those changes generate. I don't know why I never noticed this before, but I just forked my code again and ran some series and have noticed that running identical bots against each other, the red one in the top position wins at about a 2-1 margin. I have switched them up to make sure that there wasn't an accidental code change and been running sims since last night and am consistently getting the 2-1 ratio.

It probably depends a lot on the bot, though. Run two Ghengis' against each other and I would expect it would be near 50-50 because there is so much luck involved with such a dumb bot, but I guess mine is good at quickly getting the cities it can get and then goes into very long and protracted wars against itself in which the bots order apparently makes a difference.

Anyway, at the least, this is a heads up for anyone who is doing back testing like this. Run you sim one way for 2 hours, then flip it for another two hours if you want an accurate comparison.

(It could also be some other bug, like always awarding the win to the first bot in the case of a stalemate since my bot stalemates against itself a huge percentage of the time.)
Tue May 27, 2008 6:07 pm View user's profile Send private message
adum



Joined: 19 Apr 2007
Posts: 392

Post Reply with quote
hmmn, i ran my bot overnight and got this:
com.adum.bitbath.botA
524
com.adum.bitbath.botAcopy
519

so no significant difference. now, my bot isn't as deadly as yours, but it's not Ghengis. are you totally sure about your finding? i can't think of any real reason in terms of the game design that it would make a difference going first. ie, in one turn both teams will move, fire, etc. two bots can kill eachother in the same turn; it's not like the first team kills the second one because they move first.

adum
Wed May 28, 2008 4:45 pm View user's profile Send private message Visit poster's website
Bedevere



Joined: 26 Aug 2007
Posts: 2

Post Reply with quote
I did about 200 simulations over the period of about 20 hours, 100 one way and 100 the other and both were very close to a 2-1 ratio. I would do more trials, but unfortunately, my bot gets bogged down against itself and thats about as many as I can get in a day of testing.

This sounds like it's a problem that only I'm having, so I guess it's not a big deal. I will just have to split all my testing.
Wed May 28, 2008 8:09 pm View user's profile Send private message
Kender



Joined: 25 Jun 2009
Posts: 18

Post Reply with quote
You probably misread this line from the readme: "Other teams have a faction value of greater than 0."

If you're checking for objFaction == 1, then that would explain the results you're seeing.
Try objFaction > 0 for enemy objects.
Sat Aug 01, 2009 8:57 pm View user's profile Send private message Visit poster's website
Nick-Aotmzgin



Joined: 14 Jun 2009
Posts: 64
Location: Microsoft Labs

Post yep Reply with quote
Bots game play auto adjustment. the best fastest game solutions.

_________________
(i) Hacerks Group -> Greek Hackers Group (House of Hackers Offical Community)
Tue Sep 15, 2009 12:52 pm View user's profile Send private message
Display posts from previous:    
Reply to topic    hacker.org Forum Index » BitBath All times are GMT
Page 1 of 1

 
Jump to: 
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum


Powered by phpBB © 2001, 2005 phpBB Group
Design by Freestyle XL / Flowers Online.