[Kde-games-devel] kigo

[Inge Wallin]

On Monday 26 October 2009 00:03:47 Arturo Silva wrote:

> ==========================
> Inge Wallin wrote
> ==========================
> 
> > Gnugo is around 10 kyu if I remember correctly.  This strength is reached
> > by a player after approximately a year if you play regularly.
> >
> > However, you can make gnugo weaker by using the --level option is I
> > remember correctly.  Perhaps kigo could make a nice interface for that so
> > the user can select "easy", "medium" and "dabbling" (there is no real
> > "hard" setting for gnugo :-) ).
> 
> lol!  ^^
> 
> I actually finally managed to beat the computer today,... after I
> selected the 9x9 grid and gave myself a 9 stone handicap,  (so sue me,
> I was desperate to see what happens when you win).   ^^

You can improve your game immensely if you just switch your goals in the game.  
By completely forgetting all about territory and just try to keep your stones 
connected, you will immediately be able to go down from 9 stones handicap to 
4, and you will win!  Promise!  The territory will come by itself.


> > Definitely lol :-D.  (and I say that as a former gnugo programmer) 
> > Remember that gnugo took 1st place in the computer olympiad at one time. 
> > Sadly enough, gnugo development has stalled much during the last few
> > years.
> 
> Wow!  Nice to be chatting with one of the ex-programmers of gnugo.  ^^

Yeah, those were the days. :-)

> > For kigo, a better "easy" level of difficulty might be to take the gnugo
> > moves and have a random chance of a: taking the best one, b: taking a
> > suggested move at random, c: taking a totally random move.
> 
> Ah, so I it WASN'T just me about KFourinLine.  And here I always
> thought I was just borderline moronic.  :P
> But yes, not sure how much can be done to the GnuGo AI itself, but I
> wonder if you can somehow force it to think with the above paradign at
> preset percentages of the time during the Weaker slider settings.
> After the slider is past the midpoint, THEN the default GnuGo
> difficulty can kick in and challenge experienced players. ^^

I strongly advice against using a random move.  A close game can be totally 
destroyed by just one bad move, and at least my enjoyment of the game would 
disappear if my opponent threw away moves like that.

> > If you choose a small board (say, 9x9) and 9 stones of handicap,
> > you'll very likely to win even against the highest difficulty setting
> > of GNU go engine. In such tight space 9 stones for free make a huge
> > difference.
> >
> > The preferred way for a total beginner to play is to set the engine
> > difficulty to the lowest (which is plenty enough for novice players),
> > choose a rather small board (it's easier to grasp for novice players)
> > and a very high handicap. Then, as the player builds skill he lowers
> > the handicap and enlarges the board. When you can win on standard
> > 19x19 board with no handicap, then you're ready to bump the engine
> > difficulty level, or start giving handicap to the engine.
> 
> hehe, that sounds familiar.   ;)
> 
> Admittedly, that does sound like a good, code-free proposition as long
> as it doesn't interfere with the "professional" feel of Go gaming.
> I'm not sure if defaulting with a handicap for the human player should
> be encouraged, but certainly defaulting to a smaller board wouldn't be
> such a bad thing to try.

Yes, this is the way to go.  

> Or better yet....
> 
> > What kigo should offer is a dialog to change difficulty settings,
> > possibly with a set of presets. 19x19, no handicap, default GNU go
> > engine difficulty should be called normal. 9x9, 9 handicap, lowest
> > GNU go engine difficulty can be called very easy.
> >
> > Such a dialog could also be used to configure komi value, byo yomi
> > time, etc.
> 
> Easy to understand presets seems more like the best compromise solution. 
>  ^^b

Yes.  Agreed 100%. 

> ==========================
> Ian Wadham wrote
> ==========================
> 
> > You have the right idea re the wood and the pieces.  Go is a game with
> > a long cultural tradition and I believe it is important to respect that. 
> > My own Go set, which I bought in Japan a long time ago but have now lost,
> > had elm wood boxes and pieces made of glassy material.  I think the black
> > were obsidian and the white milky quartz.  Although the black pieces had
> > a matte finish and the white pieces were glossy, the specularity had the
> > same look and feel for both.  I guess shiny black is somehow perceived as
> > "shinier" than shiny white.
> 
> Hi!  Wish I had read this earlier too. ^^'
> 
> Per Inge's advisement, I actually opted for polished slate black chips
> and shako white chips. The decision was a good once since I love the
> textures on both -- however, they don't have the level of gloss of
> obsidian or quartz.

> > I query whether it is appropriate to have a time-clock.  I have not
> > heard of or seen one being used in Go and would be generally
> > opposed to it, for the same reasons Parker advanced re Palapeli.
> > However I see clocks are used in formal competitions ...  Me, I just
> > like to enjoy a leisurely game at home.
> 
> As long as I've seen pictures online of Go players playing alongside
> such clocks, I'll have to assume that some people out there do use
> them for xyz reason.  But it all comes down to whether Sascha can
> actually make use of a clock function on the programming end -- since
> chances are he might not, I made sure not to spend too much time on
> that clock.  ;)

Clocks are always used in competitions.  For a casual game like this they are 
not that important.  And besides, gnugo can't make good use of the clock.  At 
least it couldn't when I left it.

	-Inge