Its not working for me, I guess.

I´m using chrome.

Working perfectly with Firefox
When you mouse over the line it would be great if you get the day along with the number of cards:  500,135  for instance.
I am interested in the number of cards I will have till a certain date and even I get a general idea it is difficult to say.

No, its shows the success multiplier is far too low.
100 new cards per day = 1100 reviews per day in the long term
100 new cards per day = 520 reviews per day in the long term if you change the success multiplier to 5.5.
My experience w/ Anki says if you pick "very easy" (which you won't for all cards, but hey), you get ~7 days the first time and ~1.3 months the second time.  The multiplier is ~5.5.

If you backsolve this (I think 520 reviews/day is too many), if you add 19 cards/day you end up w/ 100 reviews per day in the long term.

I think his "young rate" is way too high - I set it at 0.5 and got 720 long term reviews for 100 cards/day.

Last edited by jcdietz03 (2010 May 20, 10:47 pm)

jcdietz03 wrote:

No, its shows the success multiplier is far too low.
100 new cards per day = 1100 reviews per day in the long term
100 new cards per day = 520 reviews per day in the long term if you change the success multiplier to 5.5.
My experience w/ Anki says if you pick "very easy" (which you won't for all cards, but hey), you get ~7 days the first time and ~1.3 months the second time.  The multiplier is ~5.5.

If you backsolve this (I think 520 reviews/day is too many), if you add 19 cards/day you end up w/ 100 reviews per day in the long term.

I think his "young rate" is way too high - I set it at 0.5 and got 720 long term reviews for 100 cards/day.

His rates were actually lower than mine,  and his success multiplier just a little lower than mine, so i guess that's really personal

Last edited by BlueFinger (2010 May 21, 3:00 pm)

Blahah wrote:

This doesn't work in chrome or safari - is this a problem with a framework you're using?

It should work fine on both now.

Daichi wrote:

Can you explain what variables you stick in with a bit more detail?

The model works like this: each card has an interval, measured in days, and a due date. When the due date equals the current date, the card is tested. The test is probabilistic: if the interval is less than or equal to 21 days, the card has a young rate probabilty of passing; if the interval is greater than 21 days, the card has a mature rate probability of passing. When a card passes the test, its interval is multiplied by the success multiplier and it's scheduled to tested again in interval days. When it fails, its interval is multiplied by the failure multiplier and it's scheduled to be tested again in the following day. After all due cards are tested, up to new cards per day are created, and are scheduled to be tested in the following day. You can define the initial interval of the new cards.

This model differs from SRS'ing with Anki in two ways: first, there are only two buttons: pass or fail. Second, the success multiplier is fixed; Anki changes it based on how you grade the cards.

This model differs from reality in at least one significant way: the test is probabilistic and the rates are fixed. To see why this is a problem, just imagine a deck with a fixed number of cards. No matter how hard the cards are, it's almost certain you'll master them in a finite amount of time. In other words, it's almost certain that your rates will raise until they reach 1. As the simulator uses fixed rates, this will never happen.

But anyway, I hope it's close enough for a rough estimate. If you can think of a better model, please do suggest it, and I'll try to implement it.

Last edited by iSoron (2010 May 27, 1:00 pm)