Question about Daily Joseki method

[ethanb]

I like that one - used to play it sometimes. I've heard it called a "Kitani variation" of the taisha, though I have no idea if Kitani actually invented it, or just liked to play it (or even whether he actually played it at all, come to think of it...)

It's simple and solid for both sides. It feels a little submissive though (note white will eventually need an extra move to catch the two black stones) so I don't play it much anymore.

Another request, Joaz - not giving out the SGF files makes sense, but could you tell us the names of the players and the year? For some problems with multiple "correct" sequences following it would help figure out which variation we're looking for (check out 4-4, low approach, two-space low pincer... once I guess whether black's next move is the 3-3, a low change of direction, or a high change of direction, it basically tells me which decade the game is from) but also I keep seeing one game (from a 3-4, two-space low approach variation) and I REALLY WANT TO KNOW WHAT IT IS!! The whole-board picture at the end of the puzzle is so interesting, I just want to see the whole game.

[daniel_the_smith]

I like that one - used to play it sometimes. I've heard it called a "Kitani variation" of the taisha, though I have no idea if Kitani actually invented it, or just liked to play it (or even whether he actually played it at all, come to think of it...)

It's simple and solid for both sides. It feels a little submissive though (note white will eventually need an extra move to catch the two black stones) so I don't play it much anymore.

Another request, Joaz - not giving out the SGF files makes sense, but could you tell us the names of the players and the year? For some problems with multiple "correct" sequences following it would help figure out which variation we're looking for (check out 4-4, low approach, two-space low pincer... once I guess whether black's next move is the 3-3, a low change of direction, or a high change of direction, it basically tells me which decade the game is from) but also I keep seeing one game (from a 3-4, two-space low approach variation) and I REALLY WANT TO KNOW WHAT IT IS!! The whole-board picture at the end of the puzzle is so interesting, I just want to see the whole game.

Haha, yeah I'm allowed to display that (I've asked previously), I just currently don't. I should add it to the side with the hints.

I've even thought about allowing a spot in settings for you to type in the location of GoGoD on your computer (if you own it and have copied it somewhere), and providing a link to the file on your own hard-drive, just for your convenience.

OK, now it's time for me to be a hermit for a bit and actually do some of this stuff. I've got some of it done, but am not ready to publish it yet.

Oh, and on a side note: Last time I checked, I'm not Joaz.

[ethanb]

Oh, and on a side note: Last time I checked, I'm not Joaz.

Ack! Sorry, I'd just responded to him in another thread and his name came easily to the fingers. Sorry!

(soon as I read your post I had a mental image of "and don't call me Shirley!")

[daniel_the_smith]

Alright, since this thread is the popular one nowadays, I'll put this here:

Site updated.

I think the message is a lot better now, I made it both more vague and more specific. Also, graphic stones. One wasn't enough to register subconsciously for me (tenuki still surprised me!) so I put a few. Plus some behind the scenes stuff. I may be able to do some more stuff tonight but other than that my week is pretty booked, so don't expect another update this week unless I broke something.

@ethanb: hehe

How does it determine which test to give you next? ... How is the EF adjusted each time? Does it take into account how long it's been since it's shown you that position?

Short answer: I read lots of stuff while making the study algorithm, but it follows the below reasonably closely:

* http://www.supermemo.com/english/ol/sm2.htm
* http://ichi2.net/anki/wiki/FrequentlyAs ... Ankiuse.3F

That should give you the gist of things. The last question is a definite "yes", the other two would have very long answers and I have to go to work.

[Abydos1]

How does it determine which test to give you next? ... How is the EF adjusted each time? Does it take into account how long it's been since it's shown you that position?

Short answer: I read lots of stuff while making the study algorithm, but it follows the below reasonably closely:

* http://www.supermemo.com/english/ol/sm2.htm
* http://ichi2.net/anki/wiki/FrequentlyAs ... Ankiuse.3F

That should give you the gist of things. The last question is a definite "yes", the other two would have very long answers and I have to go to work.

Ok thanks, no need to go in to lots of detail.

A couple ideas here, in your queue the last and next test date seem a little too specific it'd probably be fine if you just simplified it to something like "today, yesterday, 5 days, etc..."; also what's the -0000 supposed to be, is that time zone offset? It'd be nice to be able to sort the queue too; for example, to sort by EF to identify josekis you've mastered. Being able to click on the joseki to study or view it would be nice. Being able to see a history of tests would be nice too; although that's probably a lot of work for something that isn't necessary (not trying to bury you in work here )

Are the upcoming study items added in at a rate of X per day based on your settings or how does that work? Is there a limit to your queue size?

[daniel_the_smith]

A couple ideas here, in your queue the last and next test date seem a little too specific it'd probably be fine if you just simplified it to something like "today, yesterday, 5 days, etc..."; also what's the -0000 supposed to be, is that time zone offset?

I agree, and actually "Improve timezone handling" is already on my list. -0000 means it's in UTC.

It'd be nice to be able to sort the queue too; for example, to sort by EF to identify josekis you've mastered. Being able to click on the joseki to study or view it would be nice. Being able to see a history of tests would be nice too; although that's probably a lot of work for something that isn't necessary (not trying to bury you in work here )

Clicking on the thumbnail should definitely do something, I agree. The other stuff will be... lower on my list.

Are the upcoming study items added in at a rate of X per day based on your settings or how does that work? Is there a limit to your queue size?

Initially, it sets it up to dole out one from each group every other day. If you get behind, it will dole them out no faster than one per group per day. The setting I just added should allow you to limit that to something a little more reasonable. It defaults to 7, so unless you change it, it won't give you more than 7 new tests per day.

Oh, as we speak, my home computer is building a version of my db which includes sgf properties. So, for the next update, I expect to add a "hint 3" which will state the players & date.

In other news:

As of sometime yesterday, dailyjoseki.com has administered more than 1,000 tests! Many thanks to everyone who has registered and tried it out, and especially to those of you who've upgraded!

[daniel_the_smith]

Site updated.

... could you tell us the names of the players and the year? For some problems with multiple "correct" sequences following it would help figure out which variation we're looking for (check out 4-4, low approach, two-space low pincer... once I guess whether black's next move is the 3-3, a low change of direction, or a high change of direction, it basically tells me which decade the game is from) but also I keep seeing one game (from a 3-4, two-space low approach variation) and I REALLY WANT TO KNOW WHAT IT IS!! The whole-board picture at the end of the puzzle is so interesting, I just want to see the whole game.

Wish granted ("Hint 3"). Also, Hint 1 actually is useful now. And you can click a second time to hide a hint.

...
2) The tutor could reject moves it wasn't looking for, without penalizing you. It could effectively say "that's a fine move, but not what I'm looking for."
...

I plan to do pretty much this, but I'm going to leave it how it is for the moment because it will be much easier once I write some behind the scenes stuff.

[flOvermind]

I just had a very strange test:
The joseki + followup was played out until a ko popped up. Then every ko threat was of course treated as a non-local move, so ko threat and response were automatically played, and I just had to re-capture the ko again and again, in total 17 times

My username is flOvermind, and it's currently the top item on my queue.
Next test date: May 13, 2011 23:47:14 -0000 (which in itself is strange)
Last test date: never (??)
Attempts: 0 (I'm sure I had this position at least three times.)
Group: 4

The position:

In the browser, you can only browse it until move 25, which would imho be a reasonable cutoff point for the test as well. The test goes on a lot longer, including the ko sequence...

I'm not sure myself what should be done about that. Perhaps a ko detector, to cut off the problem after the first ko capture? Or perhaps skip the ko until it is won or lost?

EDIT: In case it matters: The test turned up after I selected moves 1-3, because I wanted to study applications of and alternatives to the attach-and-extend joseki.

[daniel_the_smith]

...I just had to re-capture the ko again and again, in total 17 times

I suppose that is a logical consequence of what I told it to do . I'll have to make it detect that situation and do something that makes more sense. For now, I recommend either shortening the test (with the adjust button) or removing it. Thanks for bringing that to my attention.

...

Also-- my server experienced an... unexpected... shutdown last night. It may have lost some small amount of progress for people that used it immediately before it shut down. I've since fixed the problem and restarted it. Sorry, everyone.

[perceval]

Another small problem:
sometimes some move that are obviously part of the joseki are auto played.
typically this is when a move is not in direct contact with the joseki (but sometime it is close ie a knight move from last stone and on the median line of the board
no specific example sorry i'll try to find one if needed

[CnP]

Also-- my server experienced an... unexpected... shutdown last night.

Glad to be of service finding bugs - honestly I'm surprised I haven't crashed KGS yet.

[daniel_the_smith]

Another small problem:
sometimes some move that are obviously part of the joseki are auto played.
typically this is when a move is not in direct contact with the joseki (but sometime it is close ie a knight move from last stone and on the median line of the board
no specific example sorry i'll try to find one if needed

It's ok, I know what you're talking about. I believe this will be fixed when I get the second version of my data collection tool working. I'm furiously working on it every chance I get!

[daniel_the_smith]

Also-- my server experienced an... unexpected... shutdown last night.

Glad to be of service finding bugs - honestly I'm surprised I haven't crashed KGS yet.

Yeah, I do appreciate the help, I'm just glad it was easy to find and fix.

For others: CnP was culprit behi--, er, I mean the victim of the shutdown.

Achievement get: Infinite memory gobbling loop!

[Abydos1]

When browsing if I play -tenuki- and then hit the back button it brings it back to white to play skipping the tenuki.

[daniel_the_smith]

When browsing if I play -tenuki- and then hit the back button it brings it back to white to play skipping the tenuki.

Hm, I thought I fixed that a long time ago. I guess not. I'll put it on my list.