I am looking for your opinion and ideas on how to improve the UI usability (which I think would have a tremendous impact in the memory strength (thus improving performance and possibly affecting the capacity of retention overall) as it has a direct effect on the user's mood and disposition) of the SRS concept used by the SRS application of your choice.

My only complaint with Anki is that the UI isn't very OSX-like. It breaks some keyboard shortcuts like apple+m (minimize) by remapping them (in this case, to mark).

I'd also like a reliable official appstore Anki client (iPod/iPhone) instead of the current html based one (which lacks many features and isn't overly reliable).

The perfect SRS already exists. It's called Anki.

The only thing missing is a decent Android port.

I make rather simple decks (no sound, no color, graphics only when a description won't suffice, such as with plant names), so it hasn't been a problem. One thing I'd like is some sort of dashboard when you open anki that lists all of the decks that you are using (and want shown) showing what is due in each, maybe as graphs. Resolve has said that it is one of his goals for 1.0 iirc.

Can you schedule Anki to automatically open at specific times and jump right into reviews on your latest deck? If not, that'd be a nice feature for the procrastinators amongst us. I find myself opening it, seeing 70 or 80 reviews are due, closing it, forgetting about it before I go to sleep and waking up to 200+ reviews...

I am not a Mac user so I don't realise when standard shortcuts are being clobbered. And if you don't report the problems, they will never get fixed. :-) Please file a report on the issue tracker with any conflicting shortcuts and I will give alternatives on OSX.

I wish Anki would show you the interval graph while you're reviewing, sort of like how jMemorize does it, it's motivational

Otherwise Anki is perfect.

- More flexibility, when decks start to grow and due cards accumulate, you need to have powerful tools to choose what to review and in what order. There aren't enough option for that. For example, defining virtual sets of cards based on some search criteria would be awesome.

- When working with multiple decks, it would be really cool to also have a screen that can give you the status of all your decks.

- Since this topic is about imagination, here's one that would really hard to get right. The ability to work on the word or character level would be a significant step. When you fail a card, you could blame the failure on a particular word or character and instead of repeating the same card after a few days, the software would show you a different card using the same word.

Codexus Wrote:- Since this topic is about imagination, here's one that would really hard to get right. The ability to work on the word or character level would be a significant step. When you fail a card, you could blame the failure on a particular word or character and instead of repeating the same card after a few days, the software would show you a different card using the same word.

Now, that would be an AWESOME feature !!

Codexus Wrote:- Since this topic is about imagination, here's one that would really hard to get right. The ability to work on the word or character level would be a significant step. When you fail a card, you could blame the failure on a particular word or character and instead of repeating the same card after a few days, the software would show you a different card using the same word.

I have some ideas on how to actually implement something like that which I've been playing around with.

Stay tuned...

jorgebucaran Wrote:3) Do you think graded spaced repetition (like SuperMemo's implementation) is significantly better than the Leitner flascard algorithm (like the one used by RevTK)?

What is the difference?

I use supermemo and I tried to move to anki. I found the move from the windows explorer type organisation of the cards very difficult to move to. Plus I have 7500 items in and I cannot be bothered to move them across. I didn't understand how the cards were made or reversed either. So that also made me stick with SM.

I have two Anki complaints, both about using it with Japanese.
1) Why doesn't it know the readings for the kanji I just typed in? I just typed it, the computer should know this.
2) When I'm reviewing recognition, when I flip the card it should show the reading as furigana instead of as a separate field down below.

Now, I'm a computer programmer by trade and I've dealt with IMEs. So I know very well that (1) probably can't be done. Even if it could, it would break with copy/paste and import use cases. BUT I think there are some ways you could mitigate the autofill: when there are alternative choices for a reading, the text field could be a little smart about it. So instead of having to use ~30 keystrokes to edit the text to get the right reading, it could instead let me choose reading #1 or #2 with a single keystroke and then jump to the next one.

(2) is... nontrivial. I know, I tried to write an anki extension to do it. I ended up needing to hook into the reading generator (forgot the name) in order to get enough data to disambiguate various cases, but I got sick of it and gave up at that point. I don't know if such a thing *can* be done automatically. But maybe, just maybe, if it could remember the choices I made when setting up the reading field...

I don't understand why you'd complain about 1 if you already know it's impossible. And as is mentioned in the docs, double click the reading you want and the others will be removed.

As for two, webkit does not support the ruby html extensions. You could hack it with tables, but I really don't see the point. You're not reading a text with arbitrary characters from a foreign source, you're reading a text in which you're already supposed to know the reading.

I think Anki is quite good as it is. A lot of people want something to manage multiple decks, but I use only one deck for everything, so I'm perfectly happy. I also think the UI is just fine as it is, no complaints here.

I would like a feature that allows your mouse to hover over kanji (in sentence-mode), returning info from a separate database (i.e. key word, possibly the story, possibly with the feature to instant-edit the story).

alyks Wrote:I think Anki is quite good as it is. A lot of people want something to manage multiple decks, but I use only one deck for everything, so I'm perfectly happy. I also think the UI is just fine as it is, no complaints here.

Kanji and sentences too? I think having more than two would be a pain, but mixed kanji and sentences... I dunno... just seems like it would be odd. Do you not mind it?

I seriously think that the focus should be on mobile devices. Either improve web Anki (audio and picture support, etc.) or improve the clients for both Android and iPhone. Anki itself has pretty much everything a decent flash card application needs.

I'd love to see a dev stop for the desktop Anki with 1.0 (except bugfixing) and total focus on mobile devices and the web client.

Is it really that troublesome to go File >> Open Recent? It takes about two seconds.

Do you know that anki is open source, don't you? I see a few developers here. Why not stop cumplaining and write some code?

jorgebucaran Wrote:This is very good input, thanks. Now if you please address this question:

3) Do you think graded spaced repetition (like SuperMemo's implementation) is significantly better than the Leitner flascard algorithm (like the one used by RevTK)?

If SRSing was the sole source of your input, then a different algorithm could make a huge difference. But you're supposed to be listening to a lot of Japanese and reading tons of text, which give you exposure to already mined words/phrases/grammar points/whatever. I venture a guess that the two algorithms you mentioned make no noticeable difference in real life when it comes to learning a foreign language.

I also guess that this is one of the reasons why SuperMemo needs a more sophisticated algorithm; it's meant to be used for virtually anything while Anki and other simpler free SRS are mainly used by language learners.

sethg Wrote:

alyks Wrote:I think Anki is quite good as it is. A lot of people want something to manage multiple decks, but I use only one deck for everything, so I'm perfectly happy. I also think the UI is just fine as it is, no complaints here.

Kanji and sentences too? I think having more than two would be a pain, but mixed kanji and sentences... I dunno... just seems like it would be odd. Do you not mind it?

Never been a problem for me.

I do prefer to keep different languages in different decks, however.

HerrPetersen Wrote:I would like a feature that allows your mouse to hover over kanji (in sentence-mode), returning info from a separate database (i.e. key word, possibly the story, possibly with the feature to instant-edit the story).

Yeah, i would really like something like Rikaichan directly in Anki. Its a little bit tedious to copy words/Kanji in the browser first to look them up.

harhol Wrote:Is it really that troublesome to go File >> Open Recent? It takes about two seconds.

And even more than that, just press Alt+1, Alt+2, Alt+3, etc. I think having a little selection at startup would be cool, but it's not incredibly necessary... I like keyboard shortcuts

How about embedding video into the main interface as opposed to popping up in a separate mplayer window? (Or can you already do this somehow?)

cb4960 Wrote:How about embedding video into the main interface as opposed to popping up in a separate mplayer window? (Or can you already do this somehow?)

This would be sweet. Reviewing cards would be more fun that way.