[Computer-go] computer Go as a social & educational tool?

[Jonathan Chetwynd]

David,

again, this does not look like an even game, please ensure you have  
given the correct handicap.
unfortunately, as already suggested, MC scoring goes awry once behind.

I am not clear as to the reason, but for the moment assume that white  
must be ahead or at least have some 'reasonable' chance of winning,
else you may safely assume the gradients are pretty useless.
There may be a particular issue related to one-eyed Fuego groups,  
raising with developers...

the current arrangement is intended for 25-12 kyu and those prepared  
to give sufficient handicap stones.

regards

Jonathan Chetwynd
http://www.peepo.com

On 28 Jan 2011, at 20:20, David Ongaro wrote:

> Hello Jonathan,
>
> Am 27.01.2011 um 15:50 schrieb Jonathan Chetwynd:
>
>> Terry, Petr & David, many thanks for taking the time to review  
>> peepo.com and the enthusiastic feedback,
>> html5 and the web offer fantastic opportunities to promote Go,
>> develop social networks and improve game tools and visualisations.
>
> Now that Batoo has finally failed in Korea, this might be a way to  
> win back the visual calibrated StarCraft generation for Baduk? Maybe  
> you should contact the KBA? ;-)
>
>> David,
>>
>> try  giving Fuego handicap (pass) stones assume 12k
>> see: http://peepo.com/pics/go.jpg
>> Fuego like most MC programs plays erratically once behind, in your  
>> example the loss is extreme,
>> see  thread: 4 'easy to avoid errors', your screengrab is a great  
>> representation of this issue.
>> also,
>
> You're right. By passing 4 times before first move I got a little  
> more interesting game, where the gradients made more sense. The last  
> move lead to a nice effekt, when the territory got darker when I got  
> the seki: http://img143.imageshack.us/img143/2839/ 
> peepocom20110127.png.
>
> The color of the upper middle dead group is not really  
> distinguishable from territories color though. And the seki also  
> could look more 'neutral'. Maybe it's just some fine tuning of the  
> gradient colors.
>
>> I have asked the Fuego authors about  enabling early resignation,  
>> there is currently no simple method.
>> also see thread:  Could a 'doubling dice'** encourage early  
>> resignation by programs?
>
> Yeah, earlier resignation would be good. But's really only playing  
> 12k?
>
> Regards
>
> David
>