hehe, did I miss an announcement or does the Skritter team like to let users discover new features of the system like it's surprise?

In any case, I really dig the breakdown for Writing, Reading, etc... ^_^

Sweeeeeeet!

I love it - especially the retention rate info for different aspects of study. In my case, my lowest aspect is definitions, so I know that I need to focus more on recalling the definitions that challenge me rather than clicking quickly past them to get to the next item.

It feels artificially inflated to me. I would much rather my progress cover what words I know (reading writing and tones) than add them all up as separate items.

When I first saw the new numbers, I thought something had gone completely wacko. It only took a couple seconds to sort it out, though. It's quite interesting to see the different numbers, and consider implications for future practice. A bit of a shocker for me, too, how my definition figures are so low and the worst retention, too. In one way it's not all that surprising since I've just added definition practice for my whole list.

One thing that makes definitions difficult is having a lot of individual characters in your study list, since the majority of characters seem to have three or four definitions, and they're not always related. This is definitely a downside of studying from Heisig or Tuttle's Learning Chinese Characters.

@董雅各 - As long as you don't focus on the 'everything' tab, you'll get a very good breakdown without inflated stats. Your total numbers will add up to a much bigger total now than they would if you just looked at the old characters learned results, but that's part of the reason it feels inflated - you're used to much smaller numbers. If there's a deficit in one aspect of your study, the new progress page is a good way to pinpoint it so you can start adjusting your study methods to your needs.

Personally, I'm quite happy with the new interface - my characters-learned bar has been growing quite slowly lately, but I've learned about 100 new words in the last week, so I'm glad to see the pretty line move up in response to that - the old version was much less rewarding in this situation.

Yeah, this is great! Thanks again Skritter Gods.

Any chance of getting the twitter notifications to include "words learned"?

James

This is pretty nice, especially the 'words learned' as my 'characters learned' has tailed off a lot recently.

Can't help but think that the 'all items learned' is a bit meaningless though. Not a big problem to have it except it appears on the front page where I would prefer to see either 'characters' or 'words'.

It would be nice to be able to set the range to "day". I know that stat isn't actually very useful (since a character I learn today is also the result of yesterday's studying), but I think it will have great motivational effects. I think keeping "week" to be the default is good, but since there's already a drop down menu, I think it would be nice.

It seems that I am also in a "club" of unpleasantly surprised with a low retention rate for definitions.
Interesting...

For "all items learned" - i found it particularly interested only in a meaning of "today". If it shows decline, or great decline today - it's not my day. As for the total number - it's really same useful as arithmetic mean average body temperature of all patients in hospital :)

Huh, for the year my definitions also have the lowest retention rate (88.0%), but not much less than writing (88.2%). My readings and tones retention rates are (somewhat surprisingly) much higher (94.8% and 95.5%).

Are people seeing much bigger differences?

Also, I stopped studying tones a while ago, as I felt it was redundant to study readings and tones. I noticed that the "tones" graph is still trending upwards; are tones still tracked independently of readings? If so that sounds like it might definitely inflate the "all" number.

James

Nope we hadn't explicitly announced it. I was actually kind of embarrassed to mention it because it took so long to arrive after definitions and readings were introduced and fully integrated (we gave it to an intern to do as his first project for the spring semester, but he never got it or anything else done. Our summer intern managed to get it done lickity split though).

Looks like it made a big splash anyway!

The artificially inflated numbers do pose a conundrum though. We want to to graph data that encompasses everything, so that it goes up and down no matter what subset of parts you're studying (some people study all parts, some only study a single part), but it's true the actual total numbers themselves don't have as much meaning as the numbers for the individual parts.

One minor change that could be made is to make it possible to choose which graph you want to see, both on the home page and on the all tab in the progress page. Then users could choose which 'part' is most important to them.

Could consider replacing the absolute numbers with relative ones instead, that is make them bar graphs that show you how many you learned (or unlearned) that day, rather than your overall total for each day. Then these graphs could include all parts together.

Or we could look at the progress page holistically and try to come up with other ways to organize the data, rather than having it being just a variation on the old way. That would take more time to do, though.

@scott - for the "all" chart, is there any possibility of generating a stacked line chart like this one?

http://flowingdata.com/wp-content/uploads/2008/12/23-545x370.png

Yeah that's something like I had in mind, in terms of changing from absolute to relative values. That would nicely encompass a lot of data.

OK, help me out here. I don't understand my totals at all.

Under "writing" the "total added words" is 1282. Under "reading" the "total added words" is 780, a difference of 502 items. Under "definitions" the "total added words" is 899, a difference of 383.

When I saw those differences I thought "OK, there are items that I'm studying that never got readings or definitions added", so I exported everything and added it back through the queue. That added a few new items; my queue was around 75 when I started and it's at 150 now, so that's about 75 items. That's nowhere near the numbers above, so something is screwy somewhere...

James

Have you deleted words before? Deleted words do not get removed from your progress totals. Perhaps you added some writings before readings and definitions were added, then deleted them. And deleted words do not appear in your exported data.

Oh yeah, I nuked everything back in December or January. However, I was back at 0 written items after the nuke. I've deleted a few other items manually as well.

If deleted items show up in progress totals, that sounds like a major bug, no?

I did an export recently (on the 12th) and it has 2,126 lines. My current "words added" and "characters added" add up to 2,814. I doubt I've manually deleted over 700 items.

James

P.S. "over 700" should be "almost 700"

Oh there's one other reason! You're studying both simplified and traditional. And for traditional characters with no variants, only the writing is counted, since everything else is the same. In fact, if variants have any parts that are the same as the simplified character, those parts are not counted separately. So you're liable to have fairly different numbers there for that reason at least.

Counting deleted words as part of your stats is less of a bug and more of an inaccuracy problem. That is, you don't forget those words if you remove them so it wouldn't be accurate for the page to have a sudden drop. Also there's no efficient, accurate and understandable way for the system to update those values gradually (since you aren't studying them so the system can't know if you've forgotten them). At the moment, I don't really see a better solution than leaving it the way it is.

@scott That explains why the readings and definitions don't match, but why does the export have almost 700 fewer items than my writings total?

James

@scott Also, when we delete items, why can't that adjust the total? It's not that Skritter is trying to track all the words that we know, it's trying to track all the words that Skritter knows that we know. It seems much less surprising to have it delete the word from our "added" and "known" totals (if it's actually known, that is) than to have deleted words included, particularly when you consider that we might have seen the word once and immediately deleted it. I do this for personal names in textbook lists, for example.

The fewer exported items is probably because you're not actively studying some of the characters you know. When you add a word, the characters are added too, but don't show up on their own unless you have "Also Add Characters When Adding Words" enabled. So it keeps track of how many characters you know how to write, whether you've studied them on an individual basis or just part of a word.

As for adjusting the progress data, it seems a little more inaccurate to me to do what you suggest. I could go either way really, but at this point this is how the system works and it would be a major endeavor liable to cause confusion and not really make a difference for most people (since we discourage the removal of words from studies anyway). So for now we'll keep it the way it is at least.

@scott Wow, that's very confusing. Any possibility of summarizing "how these numbers are calculated" and putting that info somewhere accessible from the stats page?

Also, if deleting items is going to bogus up the stats, it would be nice to have some kind of warning about that somewhere.

I'll put on my to do list to make a document that describes how progress stats are tracked and what exactly they mean.

I don't think it boguses up the stats, but I can include that deleted words/items are not removed from the progress stats in that document.