Arimaa Forum - Print Page


    
      
        Arimaa Forum
        (http://arimaa.com/arimaa/forum/cgi/YaBB.cgi)
      

        Arimaa >> Game Analysis >> Game 139489 (172731) - knarl v bot_marwin
        
(Message started by: knarl on Mar 26^th, 2010, 2:07pm)

Title: Game 139489 (172731) - knarl v bot_marwin
Post by knarl on Mar 26^th, 2010, 2:07pm

Comments for Game 139489 (172731) - knarl v bot_marwin

http://arimaa.com/arimaa/gameroom/comments.cgi?gid=139489

I'd like some opinions on this game. I'm a weak player, but I could have sworn I had the upper hand on marwin at one stage in this game.

Cheers,
knarl.

Title: Re: Game 139489 (172731) - knarl v bot_marwin
Post by Fritzlein on Mar 26^th, 2010, 4:52pm

Thanks for participating in the screening, knarl. Every game makes the final results more accurate and increases humanity's chance of defending. In particular it is good to see someone race out of the opening because bots have shown surprising weakness in races in the past.

According to tize's own description of marwin's time management, marwin will try to gain reserve when it is either winning or improving. Ergo the long think on 18g means that marwin thought it was losing. I'm not sure how to understand that, though, given marwin's massive material advantage. It could have lost the camel and still been winning materially, so it could only think it was losing based on forced goal, yet when marwin sees forced goal it will terminate search and move. I must be missing something.

I'm horrible at goal attack, but the first move I would examine would be 18s R4vvvEv. Gold could then stop goal only by occupying b1 with the camel or the f1 rabbit. If the camel, then 19s EM>> appears to win. If the rabbit then 19s ERc^R7> and I don't see how Gold can get a piece on c1. So it seems like goal in three, but don't trust me on that.

Maybe the answer to the time management conundrum is that marwin had enough time to see that its favorite move gave up a forced goal, so it was going through all the other moves one by one trying to find one that saved the day, and hit its maximum time allotment before finding a good one.

Title: Re: Game 139489 (172731) - knarl v bot_marwin
Post by knarl on Mar 26^th, 2010, 5:28pm

Thanks Fitzlein, Jdb also pointed out that move. I knew I had him! I should have taken a bit longer to make that move. I even tried out that move at the time, but didn't see the forced goal.

Cheers,
knarl.

PS. My money is on clueless to win the screening, he punished my reckless play in comparison.

Title: Re: Game 139489 (172731) - knarl v bot_marwin
Post by tize on Mar 27^th, 2010, 1:41am

on 03/26/10 at 16:52:44, Fritzlein wrote:

According to tize's own description of marwin's time management, marwin will try to gain reserve when it is either winning or improving. Ergo the long think on 18g means that marwin thought it was losing. I'm not sure how to understand that, though, given marwin's massive material advantage. It could have lost the camel and still been winning materially, so it could only think it was losing based on forced goal, yet when marwin sees forced goal it will terminate search and move. I must be missing something.

Yes you are missing something, but just because I missed it first.

There is a little detail that I forgot to mention in the chatroom about marwin's time management. And that is that marwin isn't allowed to see a position as winning (as far as time management goes) if the position is beginning to fall apart, that is the opponent is improving a lot. And at 18g he think he has a massive lead due to material, but that A-file rabbit is bad news.

But there is no way he's going to see the forced loss in three, he still struggles with some loss in two.

And thanks for the game knarl you gave plenty of positions to work with.

Title: Re: Game 139489 (172731) - knarl v bot_marwin
Post by knarl on Mar 27^th, 2010, 2:57am

on 03/27/10 at 01:41:42, tize wrote:

And thanks for the game knarl you gave plenty of positions to work with.

No worries =)