format change

Post Reply
kaf

Post by kaf »

Sorry for re-posting this one, - it's been widely ignored in the "coolness" session, and I 'd like to hear some response.

I believe, that the real source of frustration is when people can't understand why the particular move is wrong. Of course, one can build an extensive tree fully explaining the problem, but here go two obstacles:
1) It is technically difficult finishing up all possible wrong moves, and
2) Because of possibility to "roll back", anybody would be able to solve the problem applying no effort. This would downgrade a 6d problem to 30k.

The solution of the first problem is to modify graphical editor so it could copy branches of the problem tree.

The solution of the second problem is to introduce "stop" mark into a code (as an opposite to "solved" mark). When the player hits this mark, - game is over, but then, he can follow the variation to see why he was wrong.

I think, the second suggestion is more simple to implement.

{Posted by kaf}
admin

Post by admin »

one thing i've been meaning to do is to automatically have the applet link isomorphic positions. in other words, if you have a position A, and code responses B and C, then if you reach the same position A thru a different move sequence it would automatically know to answer with B and C. this is better than copying and pasting because it's automatic, introduces less clutter, and is easier to maintain.

i'm not really sure what the stop mark would do. you can already continue to explore after you solve a problem.

adum


{Posted by admin}
kaf

Post by kaf »

Hi Adum.

I didn't know about loops in a graph. Thanks. I should have read manual. :)

About the stop mark. What I meant is this, - if you make very explicit variations to please beginners, then, because of "back one" button, any person could easily solve the problem. The stop mark would help here.
I've been asked several times to explain in the variation, why some particular move is wrong.
The stop mark should set "wrong" message not at the end of wrong variation (as it is now) but in the middle of it.

Just to prove my point, I can take any 6d problem and complete all wrong variations. Then, sit back and watch this problem becoming 30k. :)


{Posted by kaf}
santa c

Post by santa c »

sounds like a great idea... only how would the author compare them? going throught the new sgf? maybe it could have a markup for changes in the applet?

{Posted by santa c}
kaf

Post by kaf »

I am sorry, Santa, I didn't get what "to compare"? All I am suggesting is a simple mark, (maybe WRONG in the SGF comment?), by which program declares the attempt to solve as "forfeited".

{Posted by kaf}
admin

Post by admin »

ah, i see what you mean.

actually, internally the applet marks you as having failed as soon as you step foot down a path that has no solution. you just don't see it.

so i guess you could just write "Wrong!" in the comment box if you wanted, and let people explore from that point.

adum

{Posted by admin}
kaf

Post by kaf »

thanks, adum.
That helps a lot! I had no idea...
At least now I understand, why my rating is so low :)

{Posted by kaf}
kaf

Post by kaf »

[quote]
one thing i've been meaning to do is to automatically have the applet link isomorphic positions. in other words, if you have a position A, and code responses B and C, then if you reach the same position A thru a different move sequence it would automatically know to answer with B and C. this is better than copying and pasting because it's automatic, introduces less clutter, and is easier to maintain.

adum

[/quote]

I couldn't get this working. Am I correct, assuming, that the identical positions share a common evolution? That didn't happen in my test.

Wouldn't it be easier to give the author possibility linking nodes? There are many cases, when positions are somewhat different, but the variation is the same. In this case, instead of copying variation, one can just link the node to the existing variation.

kaf

{Posted by kaf}
admin

Post by admin »

well, this isn't implemented yet... someday in the not too distant future, hopefully.

having editors manually link areas could work, but i'd like to see how the automatic linking worked first.

adum

{Posted by admin}
Post Reply