Hi,

I just took a look all the way thru my vocab with the viewer. I discovered that... I have 3513 items or elements. And no life. Also, the viewer kind of sucks. The only key feature to improve is that it should sort both ascending and descending on the fields. But paging thru 3513 items 20 items at a time is no fun. Yes, see above, re my life.

Now, for those still with me... I have a handful of items that aren't scheduled to come up for more than 5 years, some as long as nine years. Admittedly, these are mostly characters that appear in other words, but I kind of think that the SRS algo is pushing these repetitions out pretty far, pretty fast... I mean, I've only been using Skritter for about 5 months, and it has decided that I can likely remember something for 9 years!? I think the algo should be tweaked so that the interval grows more slowly once it gets past, say, 2 years, and it should have some sort of asymptotic limit out around 5 years. I mean if you want to keep vocabulary knowledge forever, reviewing it once every 5 years doesn't seem like to much to ask.

Of course, tweaking the algo for how it handles my study schedule in 2014 may not be at the top of Nick's to do list...

If these are characters that appear in other words which show up more frequently, I don't see the problem, really. Unless you really want to see all characters individually. I certainly don't, not any more at any rate.

Byzanti,

I partly agree. If it's only because they are appearing in other words that their delay is getting so large, than it's not really a problem. I just think that when the delay gets out past a couple of years, if there isn't another way the information is getting reviewed, it will likely get lost. And, reviewing an item once every 3 years, at 10 seconds a review, 10,000 items, only takes...

um...

10,000 X 10 = 100,000 seconds divided by about 1000 days in 3 years, 100 seconds / day, or about 1.5 minutes a day. 50,000 items - only maks it about 9 minutes a day.

Not that I have that many words, I have noticed some characters that their review time got pushed a long way a way when they shouldn't have.

If Skritter is the only place you expect to see and use Chinese characters, you might as well give up now. In five years of studying via Skritter, you should not only know simple and common words like 以后 by heart and pretty much permanently, but you should be reading them in other sources, and probably using them in your own writing as well.

I agree on the viewer, though - it's pretty limited as-is, definitely needs a makeover when the team can get to it.

The viewer is limited because of performance reasons. Every sort order we add to it makes saving vocab reviews (which is way more important) slower. And sorting by descending "next" dates will only reveal nearly meaningless nine-year intervals. Those intervals are usually the result of very common characters appearing in lots of words and aren't ever reviewed on their own anyway.

Would you actually remember them in nine years if you stopped all Chinese exposure right now? We don't know, but the question comes up every now and then. The more interesting thing is what happens to stuff that gets over a year and doesn't occur in the wild very much. As we get more data on those really long reviews, we'll have more information about answering those questions, but we haven't seen any grossly miscalibrated scheduling on the long intervals that have been recorded.

In a sense, with the way Skritter combines reviews of component characters with reviews of their containing words, individual easy character learned statistics don't mean very much, because you can't see them on your own unless you forget them as parts of other words more frequently than the other words themselves.

Anyway, the max delay is ten years (which you see as nine usually). There can be some legitimately scheduled prompts that aren't components that could get out to five years, and since Skritter is going to be around a long time, that is good. If it's around long enough for those intervals to surpass ten years, we'll have to raise the cap.

Fair enough. Enhancing the viewer wouldn't be high on priority list either.

My main issue with the viewer is primarily its slow paging system - going 20 items at a time makes it basically non-functional in my view. I suppose it is what it is, and I don't know that I'd use it much even if it were more powerful (if anything, it would probably just serve as a distraction to actually studying), but that's my take on it.

If functions are added to the viewer, I'd like to see searching and more tweaking ability in it. This would be especially useful for mnemonics and custom definitions, but there are also certain words (发展, for example), that SRS has been a bit too generous with, so I would like the ability to increase the urgency on words like that.