GoProblems.com
THE site for Tsume Go study. It offers more than 3000 problems.
Go problems can be selected by topic (fuseki, endgame, etc.), by difficulty (Go rank, or average time needed to solve a problem) and several other criteria.
The site is free, but requires a Java enabled browser.
juhtolv: If you do not want to or can not use Java for some reason, you can download all those problems as SGF-files. You'd better use goproblemscom2gquiz.sh for that purpose.
Reuven - I link to problems there using GP. It'd be nice if there was a standard of some kind.
(Adum "fixed" it - Now u get another "life" every certain amount of problems..)
Scartol: Am I the only one who finds the general difficulty rating system (Easy, Medium, Difficult) too vague? Easy seems to cover 30k-20k, and as a 9k, I can't even begin to follow most of the Difficult problems; which leads me to suspect that Medium covers 20k-5k or so. Quite a range! Maybe we could have a Low Medium and High Medium rating?
- jvt: Maybe it is because of the Chinese custom of having only three levels of difficulty: chuji (beginner), zhongji (intermediate), gaoji (advanced).
- MK: It has been recently switched into an estimate of kyu/dan level needed to solve the problem.
HolIgor: It seems to me that the problems are rated by the ratio of successful solutions there. A problem may change the level from hard to easy in a matter of hours as it is solved by more and more people. As the most of the solvers are of low kyu - dan level, all problems of 5 kyu level and below tend to become "easy" with time. I think one can ask adum to introduce new levels of difficulty. I think that now the grading is just for me (1k IGS). For me easy problems are easy, medium are medium and hard are hard. adum seems to be of the same level.
unkx80: Instead of using "easy", "medium", "hard" to see the difficulty of the problem, one can see the two numbers x/y instead. From http://www.goproblems.com/info.php3, we know that x is the percentage of the people who got the problem wrong (at the first attempt). Because "easy", "medium", "hard" is derived from x, it might be better to look at x instead. If I am not wrong, "easy" corresponds to x <= 60, "middle" corresponds to 60 < x <= 90, "hard" corresponds to 90 < x.
Benjamin Geiger: What's stopping goproblems.com from entering some sort of partnership with a Go server? That way, when you log into goproblems.com, it could present problems of appropriate difficulty. More importantly, it could combine the success ratio with your rank and calculate based on that? (For example, if most of the visitors are 5k-2d, the success ratios given aren't going to be very helpful for a 25k. However, if it could determine how successful other 28k-22k players have been...)
uxs: That would probably work, but I'm not sure if that is even necessary. Just having an account system on goproblems where information on the problems you solved is kept and used for balancing would probably work too.
Reuven -
I'd like to ask if it's a good idea to "systematically" copy problems from here to http://www.goproblems.com/.
Opinions are more than welcome..
Bill: I have a question. Do you think that doing so would discourage people from working on problems and commenting on them here on SL? Would it have the opposite effect?
(Any idea how to make a poll?)
It may seem funny that now that I have useful java (means I can try goproblems.com at all, bad design that it isn't even remotely usable otherwise!) I find it pretty useless. There's no way to keep a limitation to problem level when browsing, and no way to keep off from the problems that explore rules peculiarities or game theory (which I consider an uninteresting hobby.)
Of course one can say I should use the timed trials for keeping to the level of problems I want, but that's useless for training. There's no usable way to get to the next problem, even if I accept the fact that I did not solve this or that problem. Reloading the page really circumvents the idea, and of course is just the little extra that surpasses the annoyance limit.
This is not to say that goproblems.com might not be usable for some people, but at least setting the problem level restriction should be there -- and this applies regardless of the reader's level. As it is, it seems to remain a useless curiousity for the one who just wants to train him/herself.
Oh well, I bet some people tell me "Don't Use It, Then," and of course I won't. Yet there remains a remote possibility someone takes this as constructive criticism, so I took the effort. HAND.
-- ObOlli?
Well it does sound like criticism to me... Wherether it's constructive... Well... ;)
Thing is that problem levels aren't too accurate there. Since they're calculated by the % of people who failed to solve a certain problem. Which sounds preety reasonable... Until you think of the way you treat the problems - A lot of people don't take them seriously enough (most of the time at least), some click randomly, some just want to see the anwser... There's a way to divide them "easy"/"medium" and "hard". If you're interested in studying certain things, you can also access them by groups... Well if you really have a way to improve it, maybe you should contact adum or say something on the forum there... Reuven
ilan: I have problems with goproblems.com. First of all, I can't understand what their rating system means, i.e., "5k 30 seconds". Secondly, their full board problems come in such small diagrams that I can hardly see the stones. Finally, I don't undersand the time trial, does it ever end? It is kind of tiring to continually solve the problems without knowing when it stops. In general, I end up failing (at easy levels)because I can't see the full board problems, or because it takes me 30 seconds to read the statement of a problem (a real difficulty reading!), or the problem simply appears after only 15 seconds. Anyway, I haven't found it too satisfying so far...
Ectospheno: My main complaint regarding goproblems.com is that any user can add a problem that shows up in the time trials. So occasionally you get a problem that is just wrong or one that was intended as a joke. Hit a few of those and your time trial is over despite how seriously you may be approaching the problems. Then some authors don't realize their problem is stupid and argue forever when you point it out. I'd much rather use something like GoGrinder with my own problem sets.
Tirian: I won't deny that the site has its weaknesses, but it has been both entertaining and educational for me. The first step was to leave the time trials, create an account, and use the database engine to create untimed problem sets that were either "unsolved" or "untried" so that I wouldn't have to face the "fun" (sic) problems more than once. Perhaps I'm in a unique situation where random problems from the "easy" set go from things that I can see instantly to ones that I can usually get in a minute of cautious reading.
I asked about the curiosity of the ratings on their forum about a month ago, which seems doubly odd to me when I see an easy problem with a hard rating and then investigate the Attempt Paths and see that 90% of those users got the problem right. Didn't get a response to that, so lack of communication is probably the biggest shortcoming of the site that I can identify. Still, the vast majority of the problems are well-created and I learn a lot from playing there.
Truc: I was puzzled by this apparent discrepancy too. I'm lead to believe that when you do Navigate Solution it counts those clicks in Attempted Paths.
Ed?: If someone makes a wrong move/misclicks and then presses the back button and continues with a different move their record will show a fail but the path shown in Attempted Paths is the final one taken (the one that shows you either 'wrong' or 'solved'). Activating the navigate solution before completing the problem will also give a fail. A lot of people seem to make a mistake, realise that they have made a mistake before getting the 'wrong' and then use navigate solution to see the answer. I think that each time you refresh the page or open a new window for that problem it will count as another attempt if you click on the board.