jts wrote:Yes, I signed up today and was confused by that too. From reading the description I thought that I would choose 5 patterns that I wanted to see variations sprouting from, so I picked 4-4, low approach as my first, 4-4, high approach as my second... and then it told me that it was full. Are you supposed to put in the entire joseki that you want to be tested on?
I was going to say, "Just put in a 4-4 stone; You should get examples of both," but I went and checked-- the low approach is nearly 50 times as frequent, so you probably would need to put in the high approach by itself (like you did) to get any/many examples of it.
My wording is probably not best and the system itself may not be final; at the moment it lets you add positions as long as there's five or fewer groups in your queue, but the thing that isn't obvious is that each position you add might end up creating more than one group (up to five if the variations are evenly distributed).
jts wrote:Other things that you could tweak:
*i* On the very last move, the board jumps directly to "difficulty" without giving you a chance to see what the last move was.
Others have noted this. I haven't yet thought of a good way to both ask "how hard was this" and show the final position. Suggestions are welcome though.
jts wrote:*ii* The program will jump over twenty moves in another part of the board, say "white to move in upper left", and then once you find that move, jump over another thirty moves, ask you to find a move "on the right side", which actually means in the center... I really like the idea of learning the big follow-ups to a joseki while you learn the joseki, but I'd like to see two changes: (i) presumably you can make it so that move "X" only gets shown as a follow-up if it appears as the next big move in some percentage of the games (to get rid of things like the random black stone that spills into the upper right from the lower left on move 70), and (ii) personally I would learn more if twenty intervening moves were shown as a quick animated sequence, rather suddenly popping onto the board.
Fair points. (i) is actually part of the system already, unfortunately those random moves end up matching between games sometimes. By default, the tests end when there are no longer at least 3 (IIRC) pro games that agree on the next move. However, if you adjust the test and make it longer, I don't have a system to filter that sort of thing out. (ii) is something I might consider doing in the future if enough people agree with you.
jts wrote:*ii* From the first five or six joseki I played through, it seems like the same joseki always shows up in the context of exactly the same game. From your description of how this works (i.e., the first joseki I see is the most commonly played, second is second most common), there should be hundreds of games where the joseki appeared, no? Would it be inordinately difficult for the program to call up a different board configuration each time it wants to test you on a given joseki?
Instead of answering directly, I think a little more explanation of what it's doing might be in order.
Let's say the distribution is AAAABBBCCDD, where each letter is an example of a variation. I add these to a group in your queue in the order ABCDABCDABA*. On the first two days, you'll only see the first thing in your queue, a single example of variation A. The next two days, you'll see the example of B, but not that example of A (unless it was very very difficult for you). Next C, and so on, until you finally get back to a separate example of variation A. In other words, the system wants you to "memorize" the particular examples of the variation(s) one at a time. After you've used it for a while, the variations will be all jumbled up (according to difficulty) and will appear to you in no particular order.
[*] That order is specifically so that you won't see the same thing over and over for days.
Now, the only way you should see the same thing over and over on the first day is:
* The system had to show you every move; if that happens, you'll get the same exact test again later in the session. If the system has to show you the answer a second time, you'll get the test a 3rd time and so on. It doesn't consider you to have really taken the test until you've gotten through it with minimal help. (You can tell if this happens-- it won't ask you how hard it was)
* You clicked "get another test from your queue" and there isn't a lot of variation in one of the groups in your queue, so all the tests are similar.
jts wrote:I hope this doesn't seem like whining! Just suggestions from a first-time user. I've always thought the graphic design on Daily Joseki is (mostly) amazing, btw - did you do it yourself?
I really appreciate the comments, thank you for trying it! If nothing else it tells me what sort of things I will need to explain better. My wife did the graphics for the website, any less than amazing parts are probably my own additions.