So after using the new interface almost exclusively for the past week-ish, there are a few places it could be improved:

First - in what I assume is just a straight bug, the "show character" is a toggle switch but the character disappears after the first stroke. Could that be changed so that the character outline stays visible (preferable), or the show button can be made to not toggle?

Related to that, the "hint" function seems to be somewhat too punitive. It'll go to red after three strokes regardless of how many are in the character, and if you've already drawn three strokes it automatically goes red. Is there some way to make it a little friendlier so I don't feel like a cheater for setting myself to '2'? :p

Also, I think I remember something about pinyin/definition practice, so if that's so, then ignore this. But could there be a function added to turn off either the pinyin or the definition and have that shown at the end?

When we first started, we had the phantom work the way you describe: it would stay visible as you wrote. We quickly discovered that tracing characters was surprisingly useless for learning them. It was actually pretty amazing how much tracing you could do and immediately afterward be unable to remember any of it.

So we solved it by having it fade out, so that you actually have to put the character in memory and recall it, even if only for a few seconds. Several people have asked for it to stay on, but we're pretty convinced that it's counterproductive.

The hint scoring could be changed, sure--it's too simple right now, as you describe. What do others feel about the hint scoring? How many strokes should one see before getting a 2, and then a 1?

We might be able to do something where the pinyin or definition are hidden until you finish a prompt or click to reveal them if there are ambiguities. Or maybe they could just reveal on hover. There is full pinyin and definition practice on the way, though, starting shortly after we get the new interface solid.

It doesn't sound like you would change it anyway, but just to add my voice/vote - I like the way the phantom works now - even though I get furious with myself when I can't remember a character long enough to get it drawn even right after seeing it.

I am completely in agreement that tracing does exactly nothing for retention.

Along those lines, I may have mentioned before that there's something I've started doing for learning brand-new characters which has helped me a great deal. When I run into a character that I've never seen before, I click "show" and study it a little, then try to draw it. Typically I get it wrong and have to click "show" again. I do this repeatedly until I get it completely right. Then I click "erase" and wait a couple of seconds and try it again. I repeat this process, gradually increasing the time I wait between repetitions until I get to 10 or 15 seconds' delay. Then I click "Next" and go on studying.

I don't do it for characters that are basically simple, or for characters that are just new combinations of characters I already know.

This basically repeats, in a very compressed timeframe, the spaced repetition idea. Before I started doing this, I had a great deal of trouble with new characters, and would repeat them over and over and over in very frustrating fashion. Now I have a much higher rate of getting the new characters right the next time I see them (i.e. the first time it comes back after the very first time I see them).

James

And another thought on the new interface - I know you were trying to remove clutter, but I miss not being able to see "Where in the world did *this* character come from?" What's the easiest way to figure that out now?

@James: Ooh! I've essentially never used the "show" and "erase" features before, but that sounds like a pretty worthwhile thing to try. I like the idea of forcing myself to be able to draw any one character completely once, with no hints, before moving on to another one.
- Mike

I had one other though on the new interface, (hopefully you aren't tired of hearing them yet!), there was something visceral and satisfying about watching the list bars count down in the old format (I liked to keep them open the whole time). You could literally see the progress you were making. I found that to be a good small goal to help keep me motivated in the face of so many characters to learn.

I think the new interface is a definite improvement and much cleaner, but I do miss that short term satisfaction. I don't think you have to have the list bars to capture that, some kind of message letting you know when you finish a section could probably capture that same thing. Something to think about.

@James: The last time you mentioned that technique, it struck me as a really good idea, so I started doing it, and it really worked! Then I got lazy, and slowly stopped, (it seems so easy to just push that little green arrow....) and forgot about it. Till you reminded me! Hopefully I stay on the wagon longer this time...

Right now we have the Flash give a message when a word is added, but in the near future we're going to try having the Flash instead show the progress bar and show it growing, so you see which list and section the word was added from, and how far along you are. Might be cool to have some sort of special notification when a section or list is complete.

We're adding the list and section info to the lookup box (the one you see when you click the magnifying glass). It'll be there by the middle of next week.

I erase and rewrite new characters too. Since it's such an effective technique, maybe we can make the system encourage you to do that.

@scott - That sounds like a great idea. What do you have in mind? Maybe a "learn this character" mode?

James

I'm thinking something on the flash that would encourage the user to write it again, possibly a model dialogue. It would only come up when the character was new, and it wouldn't be too pushy. That's just the vague idea in my head though; will have to wait until Nick and George are back to see what they think and toss around ideas. Vague ideas almost never turn out to be very close to what happens, so tough to say if it will happen, and what it would be like!

I think it's worth thinking about. Learning completely-new things seems to be fundamentally much harder than reinforcing stuff that you already knew at one point and forgot.

A very simple thing you could do without changing the Flash would be to simply reduce the amount of time until brand-new characters are shown again. For example, if you were to show a new character, let's say 点, and the user were to get it wrong, then you could show just one other character and then immediately repeat 点. After a little while, you could treat 点 just like any other character for scheduling purposes.

James

I'd second James' point. For some reason that first time is just so much harder. I think it could be quite effective to start with an interval of like 15 seconds for new characters. I wonder if it might be too repetitive/boring for new learners though?

I also definitely like the idea of a special notification when a list or section is complete. Sometimes it is the littlest of things that keep you going...

On that note, it might be fun to someday come up with a more graphic way of showing progress (like in two years when you have spare time...). I read someone's account (maybe on Khatzumoto?) of how they hand shaded all the characters they had learned on a giant poster, watching it slowly fill in. You could probably come up with a similar graphic for the most common 3500 characters or something like that. Just an idea!

I like the idea of the poster, which has come up before. They already provide the "export" feature, but that shows characters that you're learning, not necessarily characters you've already learned.

If the developers can provide an export option for "characters learned" (whatever that means internally) I will happily contribute a program to display them in PDF format.

Presumably people will want to specify which 3500 characters they are working on? For example, if you're working on HSK, maybe you'll want to see your progress through the HSK list.

James

Scott, what's the minimum new character scheduling interval currently, and are these gentlemen at it?

The poster thing is a good idea, but it's too far from the core for us to work on it soon. We might be able to provide some learning data with the export, such that you could use it for those purposes. Most of the characters would just show that you learned them, though, since you only have a small number of "learning" characters at any time.

You could just assume everything added is learned and assemble your target corpus independently of Skritter for the purpose of generating a PDF or physical poster. If we do it dynamically online, it'll be sweet, but you won't be able to hang it on your wall.

For words that are gotten wrong the first time, James has his time intervals initiated at the lowest possible time difference, 60 seconds, though it's gone up and down. jcardenio's is getting higher, around half an hour at this point. The default for new users is 10 minutes. Basically the more you get wrong on the second try, the lower that value goes, depending on your target retention rate.

I'll lower the minimum to 30 seconds, and Nick will set up the Flash to give some sort of encouragement to users who often get it wrong the second time, probably by highlighting the eraser button and pointing to it inside the flash. How does that sound?

Sounds good to me.

James

Just to follow up, I can already feel a very positive difference in the scheduler change to a minimum of 30 seconds. Now I don't erase and redraw new characters so much, because I know they'll be repeated pretty quickly. Thanks!

James