[Computer-go] CrazyStone in the 5-dan footsteps of Zen

[Nick Wedd]

On 07/01/2012 19:30, Don Dailey wrote:
>
> On Sat, Jan 7, 2012 at 2:03 PM, Jeff Nowakowski <jeff at dilacero.org
> <mailto:jeff at dilacero.org>> wrote:
>
>     On 01/02/2012 05:04 PM, Rémi Coulom wrote:
>
>
>         Improving the KGS rank at 5d is very difficult. The current
>         experimental version of Crazy Stone wins 78% against Crazy Stone
>         2011. That's less than one stone difference on KGS. But I am sure
>         both Zen and Crazy Stone will reach 6d in 2012.
>
>
>     There are two problems with the KGS rankings:
>
>     One, they are weighed down a lot by past games from weaker versions.
>
>     Two, you have to worry about people trying to game the system. I
>     found strong evidence that one of the frequent CrazyStone players
>     was losing games intentionally against a weaker bot, while at the
>     same time winning frequently against CrazyStone.
>
>
> Why would someone do this?

As an admin, I waste enough time dealing with anti-social behaviour.  I 
have learned not to waste more by speculating on its motivations.

    Is it a vendetta against CrazyStone or

If I am asked to speculate - I think he is doing this just because he 
can.  He is strong enough to beat CrazyStone, but if that is all he ever 
does, his rating will rise until his wins are ineffective (or he is 
asked to give it enough handicap that he can't beat it).  So he reduces 
his rating by bogus losses against another program.

> could it be some kind of  pump and dump scheme?     What I mean is the
> scheme where you basically give a bunch of your rating points to another
> player for safe keeping,   recover them by playing normally against
> other players, and then take them back from the original player.

Over-fanciful, I think.  Users certainly do what you describe, but they
don't use bots, they use others of their own accounts

>   It's even possible that he operating the computer player he was using
> which would make this easy to do.     The admin can probably check to
> see if the login times frequently coincide or if this bot plays a lot of
> other players or mostly just him.

He uses several computer players, all operated by the same person, whom 
I know and trust to act honestly.

> And maybe it's just a coincidence?

If this is the user I think it is, there are too many losses for it to 
be coincidence.

> Don
>
>
>     I brought this up with an admin, and he said they were looking into
>     it, but several days have gone by and the player was not deranked
>     (which is the usual punishment for gaming the system, and would
>     remove past games from the rank calculation).

I am suspicious, but at 3k I am not strong enough to spot suspicious 
moves in his losses against the weaker bot.  I shall try to persuade a 
stronger admin to check on his losses.

Nick
-- 
Nick Wedd
nick at maproom.co.uk