"Balanced means that it is no easy way to define winning strategy on map." I cant see why you think that?

Balanced means "in good proportions" ie that no area is significantly better than another. In your example, I cant tell you what 3 picks are the best, but theyre likely to be in the '3 income' bonuses, and greenland, aus, and the '4 income 5 territory' bonuses. I can tell you which ones suck: canada, east russia, north africa, west US, East US. Yes obviously sometimes theyre pickable.

The map is imbalanced, north america is weak, africa is strong when you have its choke points, weak when you dont, asia gives very strong expansion options, the south is very safe etc. These imbalances make the game interesting. That is what a template needs, and the current template does not have it. It is balanced around the africa axis. Aus on one side, south america on the other, and then a vast expanse of neutrals, to be totally ignored until you reach a stalemate in north africa.

Edited 5/13/2014 15:23:00

i'm happy that the challenge is over soon, after 2 weeks without updating my bot, i can't understand what i wrote anymore, it is such a mess. if i take parrt in the next challenge, i'll have to start from scratch.
Jim: do you have an idea of how long the next challenge will take, from opening to finals?

i'm happy that the challenge is over soon, after 2 weeks without updating my bot, i can't understand what i wrote anymore, it is such a mess.

it can only get better :) you need to learn how to refactor your code for readability. it only comes with practice. i still largely suck at it for example. also, commenting what does what usually helps. :)

Once I get my membership, I'm going to be getting a starter bot set up for the Warlight API to distribute so we can start playing games on the fly with each other. I'll see if I can get a game server queued up so your bots can just join at your leisure. It should be a pretty kick-ass project :)

I've got the repository all set up (https://bitbucket.org/trogatog/warlightapistarterbot), but I can't use the API until I get the membership to start parsing out the json objects and sending/receiving test data.

when it is over, i'll make my project public, you'll see what i mean by "it is a mess" lol.I do have a lot of comments to keep it readable, but i hardcoded so much stuff that it is hard to keep track of everything. I have 3016 lines in the main class file, i wonder how many lines of code do you guys have? I bought a 867 pages book about JAVA progamming but i didn't have the chance nor the courage to take a look at it yet, maybe i'll start to read it for the next challenge. I also saw that there is some JAVA courses on youtube when i was looking into the A* pathfinding you guys talked about, it might be more entertaining than a book.

http://theaigames.com/competitions/warlight-ai-challenge/games/53727c424b5ab26e84c80128

Dalek's last move was basically Norman's way of saying,

Yeah, I could've... but I didn't.

...but I could've

Made me laugh at least :)

A thrilling game of back and forth.... neither bot manages to wrest control of the situation completely at any point, every turn is on the edge of deciding the entire game yet every time the opponent manages to pull things around. Breathtaking, exciting and... ah no, wrong game, sorry.

http://theaigames.com/competitions/warlight-ai-challenge/games/537273bc4b5ab26e84c8008e

By the way thanks for all your compliments Trogatrog, and thanks everyone for being such a great community. Lets keep some of this going after the finals!

Alright, I don't know which bot to laugh at more:

http://theaigames.com/competitions/warlight-ai-challenge/games/5372cacd4b5ab26e84c80705

125ch209: the Beta of poker will be online this friday, then after the finals of Warlight it will go fully live. Then it'll probably take around 2-3 months before we start the finals of the poker game.

damn i'll probably have to play this challenge as well, if there is a game i love more than warlight, it is poker...

We've just had an idea to create 'random' maps. Generating completely random maps bring too many complications with it, but we could create pieces of a large map and also different configurations to connect those pieces. Then before the game starts it randomly picks a number of those pieces and then randomly picks one of the ways to connect them. This way you never know what map you're playing on because it can be any of, let's say, 20-40 configurations.

125ch209: Happy to hear you like it :)

Edited 5/14/2014 07:26:24

We've just had an idea to create 'random' maps.

Great idea!

Once I get my membership, I'm going to be getting a starter bot set up for the Warlight API to distribute so we can start playing games on the fly with each other. I'll see if I can get a game server queued up so your bots can just join at your leisure. It should be a pretty kick-ass project :)

I've got the repository all set up (https://bitbucket.org/trogatog/warlightapistarterbot), but I can't use the API until I get the membership to start parsing out the json objects and sending/receiving test data.

Maybe i'm not understanding you right, but afaik the warlight api only lets you create games, not submit moves. I think i asked Fizzer for bot support at some point (my idea was to be able to let your bot play for you when you get booted) but it wasn't very high on priority list.

Maybe i'm not understanding you right, but afaik the warlight api only lets you create games, not submit moves.

Yes its true. Bot can't play entire game for human.

But I think it we should somehow integrate our bots into warlight. Not now, but maby in future. Because top bots already play much better then default warlight AI.

The bot just need to pass some acceptance tests that prove its stable and adequate play on maps with different settings and on different maps. This is not easy task..

The map is imbalanced, north america is weak, africa is strong when you have its choke points, weak when you dont, asia gives very strong expansion options, the south is very safe etc. These imbalances make the game interesting.

So advantage of one regions (good for expand) compensated by advantages of another regions (safety). This is how I understand 'balance'.

+1 for randomly generated maps

Yes its true. Bot can't play entire game for human.

Then I guess it's time to set up my own server just for bots :P I'll at least be able to use map data from the api

http://theaigames.com/competitions/warlight-ai-challenge/games/53779fca4b5ab24d42614ec3

so funny to watch games of this hardcoded strategies, like Herz here ;)

Right... and pedrito moving this big stack in Asia instead of defending Africa was completely dynamic?

Looks like Herz v170 follows the strategy I proposed. Nice.... although not completely accurate. Actually I'm cool with Herz having this thing hard coded because he takes out other hardcoded fake AI bots especially those from certain german authors. Gadzbot also kind of plays that way, (though not to the extreme) that's why he keeps his top ranked spot.

mweb keep running into Herz's stack:
http://theaigames.com/competitions/warlight-ai-challenge/games/5377da3a4b5ab23b41683dad

With Herz v169 you see him not following my strategy by moving his North Africa stack as he loses sight to the opponent:
http://theaigames.com/competitions/warlight-ai-challenge/games/5377ce784b5ab23b41683c98

looking at this wayy too late, but it seems fun. Only sucks that there is no starter example in C or C++, because while I know basic coding, I've never written an AI before :P