Joined: Nov 2011
Posts: 27
Thanks:
0
sorry if that is an odd subject. what i am trying to ask is this: is there any kind of estimate for how many reviews a day you will be looking at (once you're into the swing of things) based on how many new cards you add a day?
it's very easy to add a bunch of new cards and learn them that day but i want to be careful so that i don't end up overwhelming myself with tons of due reviews daily.
Joined: Feb 2013
Posts: 92
Thanks:
0
good question. often new people here get very enthusiastic and add tons of words every day only to discover that they can't handle all the repetitions that come with it. then they get fed up with anki, burned out, or stressed and hurt their progress by being overly ambitious.
the answer to your question is strongly dependent on your anki review settings such as 'interval modifier', 'easy bonus', 'graduating interval'. it's definitely worth having a look at what they actually do and adjust your settings. and of course it also depends on your actual performance.
let me give you an overly simple model of how you could tackle the question:
let x[n] be the number of cards you add on day n.
let y[n] be the number of cards that you have to review on day n.
and let's assume that for a new card, you will see it on day 0 (the day it was added) and then review it on days 1 2 4 16 32 etc. E.g. the interval always gets doubled, which is what anki actually does on default settings if I remember it correctly (interval modifier = 100%).
then we get:
y[n] = x[n] + x[n-1] + x[n-2] + x[n-4] + x[n-8]...
or if you prefer to use the 'right shift operator R', which when acting upon x[n] spits out x[n-1] and hence Rx[n]=x[n-1], you can write it as:
y[n] = (1 + R + R^2 + R^4 + R^8 + ...)x[n]
The above formula basically shows you - as you have probably guessed by now - that reviews will start to pile up quite mercilessly. However, assuming that you add cards eternally is of course an unrealistic assumption, so you can cut off the series at some point. Also, you could assume that you have mastered a card after a certain number of reviews (as is the hope behind the whole srs idea). So you can truncate the series.
So in steady state, the number of daily reviews will be a multiple of the number of cards that you add every day. It really depends on your settings, but I'd say a good estimate would be a factor of around 10 (plus minus 50%).
This quick and dirty estimate aligns quite well with my experiences. Adding 30 cards/day quickly turned into 300 reviews/day. Of course it must be noted that this model is a huge oversimplification, but it should give you a rough idea about the scaling.
TLDR: Number of daily reviews ~ 10 * number of cards added daily.
Edit: For better estimates based on your actual anki deck you can also open the statistics window (ctrl-s) and have a look at the forecast of reviews which is shown at the very top.
Edited: 2014-06-12, 10:44 am
Joined: Aug 2009
Posts: 63
Thanks:
0
Assume that your reviews will average out on about 7 times your typical daily dose if you have a failure rate of about 0.1 to 0.2.
If you review regularly, you will get a review spike on your 3rd day, 12th day, 1st month, 3rd month and 8th month. By the 10th month if you aren't finished yet, you should have a fairly stable review count (the next bump period is 2.5 years, enough to get through 25k facts at 30/day).
Joined: Nov 2013
Posts: 105
Thanks:
0
Well, in my experience, it's more like 20x. I was doing 50 cards a day, and the last few weeks of Core 10k were consistently 1000+. The more cards you add, the higher your reviews become, but it's not a linear increase. Also, the more reviews you have, the longer it takes, becoming easier to get fatigued, leading to more mistakes, which results in more reviews, etc.
Joined: Jun 2012
Posts: 427
Thanks:
0
I did 25 a day, and it never increased beyond 200, even before I changed the interval modifier to 200%. I think there are other factors here as well, for instance how you managed failed cards (back to square 1 or a certain percentage of previous intervals), how many cards you generally fail (with pre-made decks, this is probably generally higher) etc.
Joined: Jun 2013
Posts: 854
Thanks:
31
When you say "reviews per day", do you count the initial learning reps? Because if you don't, note that Anki counts learning and relearning together at the bottom - you have to estimate from the Review Count graph instead.
I'm at 25321 reviews (~7700 of which were learning) for 2067 cards; 12.3 or 8.5 reviews per card added per day.
Edited: 2014-06-12, 9:37 am
Joined: Jan 2014
Posts: 199
Thanks:
5
I've been doing 100 a day for the last month or so and I'm getting 550-600, and if I keep it up I imagine it will be 600-700.
new words * 7 seems like a good generic formula to use.
Edited: 2014-06-12, 7:02 pm
Joined: Aug 2009
Posts: 63
Thanks:
0
@hyvel
The formula for Anki's review cycle is:
f(x) = initial_interval * ease^period
With default ease (2.5) and initial (1), the review cycles grow like this: 1,3,7,16,40,98,245,611,1526 (in days). Unless you plan on adding at a constant rate for over 4 years, you'll never get more reviews from the 7th cycle before you are done with the deck. You are not even counting failures which count as an extra card per day with default settings (failure multiplier=0).
Assuming he is doing the core 10k, he only has enough cards for 100 days at this rate, so it's more like 1 additional increment of about 100 reviews.
