Even if I win against you I'll probably just turn into a super smooth smoothie after that because apparently I wasn't blended enough the first time. I checked his record versus SupremeDalek. It's 2/15. So that seems to be highly in favor of Blender losing, then playing the winner of our match. So this bracket might be even worse for me than the first one.

People that have managed so far to beat their higher ranked opponent:
mweb (15) beat psWarlighter (11) and next faces AdsRiskbot (3)
Webmartino (16) beat RaptorBot (12) and next faces Pedrito (4)
SupremeLeader (22) beat HouplessBot (18) and next faces Muli (20)

Which means 20 or 22 will go to finals (18 & 24 were knocked out of their bracket). Also 17 or 19 will go to finals (21 & 23 were knocked out of their bracket).

The match SupremeDalek <--> Trogabot was the highest ranked pairup and it didn't disappoint due to the fact that Trogabot was acutally playing with me instead of some hardcoded stuff.

Next the match SupremeDalek <--> Blender will be the highest ranked pairup. I don't think these games will be that interesting however since Blender isn't really playing the game but has his hardcoded strategy implemented. If SupremeDalek gets Blender in a messy situation it should be a clear victory of SupremeDalek and if Blender manages to pull his hardcoded strategy off he has good chances winning.

Darkness will come over this land. No matter how many high ranked bots you throw in my way, I will exterminate them all.

SupremeDalek <--> Trogabot game 1 (Trogabot win):
One of the rare situations where my bot looses in Australia and instead wins South America (I give Australia some extra points to make him win there). Turn 6 stuff gets interesting. As my bot looses sight of the opponent he immediately goes strong in his direction. He has no intel that the opponent has Australia due to the fact that he only uses the opponent deployment to calculate which bonuses he has. As Trogabot deploys visible 7 on his border he finds out turn 8 that Trogabot has Australia. What SupremeDalek now does is stacking high enough against Trogabot until a stalemate occrus and then he starts expanding. However the stalemate detection has no luck buffer and doesen't count repeated attacks. So in the end SupremeDalek looses this already won game. Him making this expansion step in North America is a bug since I probably forgot to prevent him from expanding when I have more income and lesser armies on board.

SupremeDalek <--> Trogabot game 2 (SupremeDalek win):
SupemeDalek faces Trogabot both in Africa and South America from the start. Trogabot tries to fight both for the unimportant Africa spot and the important South America spot. This leads to a crushing victory for SupremeDalek.

SupremeDalek <--> Trogabot game 3 (SupremeDalek win):
Round 5 stuff gets interesting. Instead of deploying to Brazil SupremeDalek deploys to his stack in the background and moves the bigger stack first order to Brazil. This prevents him from getting two bad fights when he doesen't get first order. However he doesen't get first order and in the next couple turns things get more and more bad for SupremeDalek since for some reason Trogabot defends the unimportant spots that SupremeDalek attacks leading SupremeDalek to loose more armies. What Trogabot should have done from Round 6 on is keeping his big stack together and hiting me full force each turn on that spot where he can kill the most of my armies. Attacking those 1 army spots isn't helping. SupremeDalek also doesen't play very smart since he had to build one huge stack. Trogabot however plays worse since he doesen't even alwys uses all his armies for attack. Turn 21 SupremeDalek and Trogabot have about the same armiies on the board but Trogabot has two stacks. In such a situation SupremeDalek plays less stupid than the other bots (but still pretty bad) so he wins.

SupremeDalek <--> Trogabot game 5 (SupremeDalek win):
A classic example showing the importance of North Africa. Trogabot wins in South America and Surpeme Dalek wins in Australia. SupremeDalek however has North Africa and is able to sit there until he reaches a stalemate and he can take Africa. As he has the income advantage he uses it to stack against Trogabot until he has enough armies to attack him. He isn't using his income advantage to take another bonus as some other bots would have done.

Edited 6/3/2014 17:58:26

pwned by north america counter strategies. bah. hope i can still make it through the losers bracket.

there will be 3 finalists on the loosers bracket and only 2 are supposed to pass, how is that gonna work?

each bot will play against the other 2, and the bot with 2 losses loses...and if each bot goes 1-1, they bot that needed the more rounds to get there loses

I should note that it doesn't bother me that I'm getting knocked out quickly from the running. My bot isn't that good. It's totally blind to enemy move and deploy orders and can't even take Australia properly. What bothers me is that two 17th+ ranked people are making it into the finals without having had to beat anyone above them in rank (or at most had to beat someone 4 ranks above for SupremeLeader I suppose which is something, but still that person wasn't even top 16 let alone top 8) while people with significantly better bots are going to be knocked out and won't make it to the finals. And the anti-climactic way they just spew out all the games prevents it from happening in a more engaging and interesting fashion. I hope they display the finals a bit differently and in a fashion that is more like the games being played live to build a little suspense for those able to sit and watch it as it's happening.

Edited 6/3/2014 22:01:28

Grats Doushibag on your unexpected victory against Trogabot :O

http://theaigames.com/competitions/warlight-ai-challenge/semi-finals/round/2
Well that was the closest matchup of all the games in the tournament so far!
You even had a turn 100 win! Was pretty stupid of my bot to attack that second time, I think I screwed up and failed to put the proper army check in that order as is present in most of my attacks. http://theaigames.com/competitions/warlight-ai-challenge/games/538c72e64b5ab23cee6dbcc8
11 games total we played. That's the most of any matchups I think. The next most games played is 9 by TheeYo and RaptorBot and then there are a several 8s (including atleast 1 in the first round).

Edited 6/3/2014 22:13:51

Dbagbot(13) beats Trogabot(5) 6/4/1
Assimilator(21) beats Alleirbag(23) 5/2/1
psWarlighter(11) beats Herz(7) 5/1/0
TheeYo(8) beats RaptorBot(12) 5/3/1
HouplessBot(18) beats ToIdea(24) 5/2/0
KewaKareban(14) beats WisdomsPower(6) 5/3/0

#5, #6, and #7 all go down!

@Doushibag - Don't judge that hard. My SupremeLeader was top 10-15 long time (with peak top 5). It's also not truth that some bots are significantly better. You could say that in situation when every bot is all-around. However most top 15 bots are hardcoded so I can't wait to kick their asses in Warlight AI challenge on pseudo-random maps. You will see then the real force of the dark side (php mafia [due to Pedrito claims that he got all-aroud bot too]).

Edited 6/3/2014 22:29:10

Im shocked, so many good bots out..

Sorry if that sounded insulting. It's not that I'm saying you shouldn't be in the finals and only the top 8 should be there or something like that. I was merely saying it would make more sense to me if you at least had to go up against someone in the top 12 first to get there. I think it would be more fair if everyone that made it to the finals had at least one opponent in the top 12. The altered setup I suggested had a top 6 person in each of the 6 brackets so you either had to beat a top 6 person (or someone who did) or claw your way through the loser's bracket to get there. It was a more even distribution of the ranks as they were at the end of the qualifiers. And yes I know people shift around a lot. I hit higher ranks than 13th (10th was best I ever saw myself hit) and I even slipped all the way back to 36th at one point after a bad losing streak. I was just thinking if I was in the position of a few of these higher ranked bots I'd be a little annoyed that they got to go to the finals and I didn't. I'm not suggesting my bot should be there (and it will likely get knocked out in its next matchup).

The truth is, only reasonable distribution of semi-finals should be random one.

My bot (#22) destroyed #18, but I can't get to the final because of muli (#20) that my record against is 2:18.

What I mean is, every bot has it's counterbot. The fact that I am #22 doesn't mean I am worse than 21 bots. I am better than #18 but worse than #20 etc. etc.

Good bots knocked out are example of unluckiness. They prolly been fighting against their counters.

Edited 6/3/2014 22:59:59

Yeah I would have been happier with random opponents. But they seemed set on controlling it and then essentially 'fixing' it on some level which I didn't care for. They didn't use a simple or standard method for matching people, but controlled it in a custom fashion to 'make it interesting' or something as they say. I'd have preferred random and just letting the chips fall as they may. As you say some bots counter others so rank is by no means straight forward. Agreeing with you there.

Hello

but I can't get to the final because of muli (#20)
I have no idea what this Muli bot is doing among the top 24. Seriously, this is just a nonsense bot with probably a few hundred lines of code. The only thing that he does right is going strong into Africa from Australia.

Muli at his worst vs SupremeDalek:
http://theaigames.com/competitions/warlight-ai-challenge/games/5384e0984b5ab2088e1bb8fc

I disagree that the bots are at approximately the same strength. Of course with these semi final pairups it will appear that the bots are close since a couple bad / medium bots will get a free way to the finals. Also the 100% luck setting makes it look like the playing field is closer together than it actually is.

Edited 6/4/2014 03:59:45

Muli has beaten my bot a few times, but my bot can be pretty retarded as I didn't have enough time to clean up some of its behaviors and added a lot in the last day or so (also had to comment out a large portion that I didn't have time to find the problem with).

Just watched this game against Troga:
http://theaigames.com/competitions/warlight-ai-challenge/games/538c72db4b5ab23cee6dbcc7

That may be me at my best. Doing well in a standoff and taking a bonus to then win with. Of course if my bot was better I might have been able to kick him out of South America with all those advantageous hits I took. Still was nice to see something I added real late working well enough to get me a win.

Checked my record against Muli, he clocked a few wins against me. I'm 10/3 against him if I counted right. Didn't look to see if it was the Australia to Africa thing, but my bot could fall prey to that. I did manage to add in some guarding behavior for South America, but it also makes my bot a bit retarded sometimes in Africa costing me some games.

Lol Muli can't play himself, So it's hard to expect, that his bot will be good :P

condolences to trogabot, you deserved to be in the finals imho!

i was thinking herz would also kick my ass, my back games log against him didn't look too favorable for me. lucky to go through. hope adsriskbot loses against mweb. i think i have better chance against him. mweb's north america counter strategy seems quite defendable but somehow still managed to beat me. :(

feels abit wrong to play against same bot twice on the semifinals, shouldnt we be playing against the loser from one of the other brackets?

ps: so you've finally noticed the North America strategy :P?

I also feel unlucky in this bracket. I think I have only roughly 25% chance to go to the finals at this point, where the main probability comes from being in the 3-way pre-final, but expecting to lose to Gadzbot (bad record), as well as 8-9 unfavorable odds to even get there against you, ps. Lastly being slow as we discussed in the stream will not help considering number of rounds will be used as a tie-breaker.