Arimaa Forum - Print Page


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

        Arimaa >> Game Analysis >> 5HT vs bot_Marwin2013CC g#280091 forced win gold?
        
(Message started by: dht on Nov 2^nd, 2013, 4:23pm)

Title: 5HT vs bot_Marwin2013CC g#280091 forced win gold?
Post by dht on Nov 2^nd, 2013, 4:23pm

http://arimaa.com/arimaa/gameroom/comments.cgi?gid=280091

[edit - did you know shift-return posts your post?]

I'm pretty sure there's a forced win for gold past the timeout of this game

Continuation?

41s ed3n Mc3e Md3e ed4s
42g Ec4w cc5s dd7n Hc7e
42s ed3n cc4n ed4w hb8s
43g Dd5e cc5e dd6e rc6x Hd7s

?

Title: Re: 5HT vs bot_Marwin2013CC g#280091 forced win go
Post by kzb52 on Nov 2^nd, 2013, 10:43pm

I think gold wins, but not immediately. For example, after the end of your variation silver has moves like
43s: ec4n Rb5w ec5w hb7w
pinning the rabbit to the side of the board. If silver plays like this, gold wins material and eventually wins the game. However, I don't see anything like a forced win that can be calculated over the board in a live game. As a disclaimer I note simply that I have no idea what I'm doing... ???

Title: Re: 5HT vs bot_Marwin2013CC g#280091 forced win go
Post by dht on Nov 3^rd, 2013, 2:52am

on 11/02/13 at 22:43:20, kzb52 wrote:

Hey I'm the resident kicktoy for every player and bot rated over 1500
;D

We can get better by analysing play our betters, amrite? ::) 8)

[Editing note: later dht went to look at ongoing postal matches and found out kzb's rating is 1973 :o]

I dunno if it can be calculated over the board in a live game, but maaaybe it could be intuited, but I do reckon 5HT underrates some of his best moves and is a bit pessimistic.

Like he seemed to think moving the camel onto the trap was a bad move from what I could tell, but sometimes moving a piece onto a trap fortifies it and makes it harder for the opponent to take advantage of, possibly?

He had a forced win, as long as he avoided major blunders from his position?

Was there a stronger move than moving the camel onto the trap on the move before marwin conceded by timeout?

Finally...

43s: ec4n Rb5w ec5w hb7w

Could silver do better than that move?

Are there any better variations earlier in the post game playout sequence for silver?

Title: Re: 5HT vs bot_Marwin2013CC g#280091 forced win go
Post by kzb52 on Nov 3^rd, 2013, 1:06pm

First of all, ignore my rating, it's wrong.

Second of all, yes analysis of other people's games is a great way to get better. That's why I posted in the first place.

Third, I think I need to clarify my definition of the word "forced" a bit. Because there is no chance, the position is completely known, and draws are impossible in arimaa, everything is in a very real sense forced. ANY position one examines is a forced win for gold or a forced win for silver...eventually. The only reason the game is worthwhile is because calculating such things is impossible unless the end is very near. So since 5HT is up materially and positionally, I argue he has a forced win (if he avoids blunders). However, I also argue that it is far beyond our ability to calculate it precisely and prove my intuition correct.

My proposed 43s is a terrible move. If silver plays it, he unnecessarily gives up any hope whatsoever of winning. However, it still must be examined if you're trying to find a forced goal because such a defense makes immediate goal very difficult even as it concedes the long-term win. A better move on 43s may be capturing the b2 cat and trying to start a goal race. Such a move might allow goal faster, but it would probably be better because it doesn't so easily concede a long term win and keeps things complex and unclear.

In order find the forced win for gold, you must examine EVERY possible defense and show that they all fail. That means both of my suggested strategies for silver - the 'give up hope and defend as long as possible' and the 'gamble it all on a race' ideas - must be examined thoroughly in all their subtle variations. What if the last step is different in my first 43s? What if silver tries to activate his camel? What if silver races on the eastern side instead of the western side? My opinion is that because silver has a lot of moves which look decent this is a tedious exercise. The win is there, but it's a reasonable number of moves in the future and I can't say exactly how far. Of course I may have missed something, but unless gold has moves for which we can say "silver must play A, B, or C or else he loses immediately" then I don't see how we can prove the forced win.

Also I can't read 5HT's mind but I'm guessing he didn't like his 41g because of the 41s you proposed. Silver can flip the gold camel away from the silver horse, and that's the opposite of what the gold camel wants to do. I'm betting that tactic is why he didn't like it (it's why I don't like it). With this in mind I think moving the elephant to c4 and spending a fourth step on something like Rg3 is fine for 41g, although I haven't analyzed too deep.

Hopefully this clarified my first comment and answered your questions without rambling too much.

Title: Re: 5HT vs bot_Marwin2013CC g#280091 forced win go
Post by Hippo on Nov 3^rd, 2013, 4:20pm

What am I missing ... what is wrong with flipping camel back on 42s in your variant?

Title: Re: 5HT vs bot_Marwin2013CC g#280091 forced win go
Post by kzb52 on Nov 4^th, 2013, 6:58pm

Ha! Well spotted. I don't see anything wrong with it, I just missed it completely as I played out the moves.

Title: Re: 5HT vs bot_Marwin2013CC g#280091 forced win go
Post by dht on Nov 5^th, 2013, 5:09pm

http://arimaa.com/arimaa/gameroom/comments.cgi?gid=280548

The thought that marwin timed out because it kept looking for a way out of a forced win came from watching this game

http://arimaa.com/arimaa/gameroom/comments.cgi?gid=280908

Given this game and marwin timing out over deciding on a camel horse exchange, maybe it's just neurotic

I maintain 5ht is much stronger than they give themselves credit for