| Life In 19x19 http://www.lifein19x19.com/ |
|
| Crowdsourcing tsumego - Results and lessons learned? http://www.lifein19x19.com/viewtopic.php?f=18&t=1754 |
Page 1 of 1 |
| Author: | feather [ Tue Sep 28, 2010 6:13 am ] |
| Post subject: | Crowdsourcing tsumego - Results and lessons learned? |
There was a thread some time ago on godiscussions about Crowdsourcing Tsumego. One of the forum members (Jordus?) have made a web application for that and I happily participated in the experiment. Can anyone remember what were the results of it? Was it successful? Were there any issues with the approach? Do you think it would be beneficial to have such site available? I remember two issues that was limiting the usability: 1. A few hard problems produced hundreds of very easy ones. So most of the time the solver were presented with 30k problems. There were some ideas to avoid the issue by remembering the percentage of people that solved the problem correctly. That would allow the application to show only problems that are appropriate for the solver's level. 2. Sometimes there were some meaningless variations, produced by a player that just played randomly on the board or trying to escape in a e.g. Black to live problem. This could be avoided by adding one more button that says 'Not a good problem' or something like that. I think this was a very interesting approach that can be useful. I think both of the above problems can be solved. Were there any other problems you remember? For those who are not familiar with the concept: The system uses the fact that adding a stone to a problem usually reduces it to a simpler problem. If a problem is e.g. "White to kill", and someone answers it. the resulting position is then presented to another player as a (presumably simpler) "Black to live" problem. This continues until the problem becomes so simple that almost all players agree that "Black is dead" (if all white moves were correct) or "Black is alive" (if any white move was incorrect). In the process of doing this a tree is created with all possible attacks and refutations and the many board positions produce numerous simpler problems that allow the player to learn every aspect of the position in the original problem. |
|
| Author: | daal [ Tue Sep 28, 2010 9:39 am ] |
| Post subject: | Re: Crowdsourcing tsumego - Results and lessons learned? |
I liked this too. I believe it was ross who was working on it... |
|
| Page 1 of 1 | All times are UTC - 8 hours [ DST ] |
| Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group http://www.phpbb.com/ |
|