How did you guys learn how to set the intervals right in Anki? It seems I do a lot more reviewing than I need. But i also don't wanna realise after a month that there are 200 cards I already forgot because I set the interval too large...

I don't usually mess much with the intervals. What kind of deck are you doing? RTK? Vocabulary? Grammar?

If you have too many unnecessary reviews then it shouldn't take long to get rid of them: You see a word, recognize it immediately, mark it as easy - Less than one second. Even if you are getting 50-60 unnecessary reviews reviews it doesn't make much longer than 1-2 minutes to go through them! It just happened to me today when I thought "Oh no, the # of due cards is way too high, I can never finish this today". Then I realized most of the cards were mature and within less than 5 minutes the # was down to half.

Just be honest with yourself when you mark a card as hard/good/easy. Don't be afraid to click on "easy". In the long run it won't make much of a difference if you modify the intervals slightly.

I don't think it matters much, but the only thing I changed was the initial hard interval to ~18-24 hours, so my first review for new/restudied cards doesn't come up for a bit. I think it was something at Supermemo that made me want to test it, and I liked it better, so.

I do wish there was a way to set intervals in overall settings, rather than having to do it for every deck.

nest0r, can't you adjust that within the "model's" settings, so that every deck using that model will have the same interval? It's not quite overall, but it would help if most of your decks used the "basic" model format, or whatever...

I haven't bothered messing with the intervals, but what made a huge difference for me was when somebody here kindly showed me mathematically how insane it was to mark everything hard. Once I gave myself permission to mark easy questions as easy, and medium questions as medium, with hard reserved for only those questions that were really difficult to remember, it made my reviews much more manageable, and I still get 90-95% on the mature cards. Trust the algorithm in the SRS, it won't let you down.

So far those changes require I do so deck by deck, regardless of whether they share models.

The only time I mark cards hard is when I restudy or first study them, and if they're, well, hard. I very rarely use the easiest grade, mostly just 1/2/3.

nest0r Wrote:The only time I mark cards hard is when I restudy or first study them, and if they're, well, hard. I very rarely use the easiest grade, mostly just 1/2/3.

Sorry, that second bit about marking them hard was for the OP, I understood what you meant in your earlier reply about the first study...

That is a shame it is something you have to do deck-by-deck, since if you found something that works for you, you probably want to use it on all/most decks. An option to set the intervals globally, and then re-adjust them locally as desired would be cool.

That's what I was thinking. And yes, I was just adding my thoughts on grading, as I forgot to mention how 'initial hard interval' connected to 'restudy/new' anyway.

Also kendo99, since you're into this stuff, I tried to pass the torch on this topic to you, since I've rambled enough on the forum ;p - http://forum.koohii.com/showthread.php?tid=5431

lol, I took the bait, although, I really should be doing something productive...

kendo99 Wrote:...but what made a huge difference for me was when somebody here kindly showed me mathematically how insane it was to mark everything hard.

Do you have the link to that post by any chance?

I agree that using the "hard" grading sparingly helps keep the daily review load down. It took me some time to realize that a conservative grading approach lead to a lot of "review churn" that left no time for learning new cards. I learned that the key is to keep "batting" those cards as far out into the future as possible (within reason of course), thus spreading them out, which means grading Good the vast majority of cards that I get correctly.

I'm still trying to optimize my pacing with Anki. I think that the method Anki has for setting the pace (namely, a setting for the number of new cards per day) can be very misleading, especially if one is starting out, because it gives no clue about the long-term load that any new-cards/day setting will entail. If one is new to SRS, one is likely at first to underestimate one's eventual SRS workload. (At any rate, that's what happened to me!)

It would be nice if instead one could tell Anki the maximum number of cards (both review and new) one wants to deal with each day, and have it figure out, on a day-by-day basis (depending on the "foreseeable daily workloads") the number of new cards to present (if any). Of course, this would require a lot of educated guessing and estimation on Anki's part, but as long as the actual daily workloads are roughly in the desired ballpark (and, in particular, they don't greatly overshoot my desired daily workload) it would be fine with me.

I did some crummy computer simulations of Anki usage and found that a good approach was to pick some desired daily workload (DDW), and every day add as many new cards as the difference between this DDW and the maximum foreseeable workload (MFW) (as long as this difference is positive, of course). I don't remember how my simulation code estimated this MFW, but one can use Anki's Due graph to achieve a similar effect. For example, if my DDW is 100 cards/day and my MFW (i.e. the highest peak in my Anki Due graph as of today) is at 90 cards, this means that I can only add 10 cards per day until I get over that MFW peak, assuming that in the meantime this highest peak does not grow, and that no higher peak sprouts elsewhere in the graph. (In practice, the highest peak remains pretty constant until one or two days before reaching it; then it grows somewhat, mostly due to failed cards and newly introduced cards.)

I started implementing this approach a couple of weeks ago, and it's still too soon to tell how well it works. The MFW on my Due graph has been well above my DDW (meaning: "no new cards") during all this time. Not adding new cards sucks, of course, but at least my actual daily workload is slowly coming down to more manageable levels. As I see it, these last two weeks I have been basically "digesting" (like the snake that ate the elephant) some uncontrolled new-card binges of 2-3 weeks ago. At my current rate of "digestion", my MFW will go below my DDW in about one week. After that my MFW will be about 5 cards below my DDW for another two weeks or so (again, barring the emergence of a higher peak in my Due graph, or growth in the current maximum peak). After that, if I play my cards right (literally) I should be able to add between 15-20 new cards per day and still stay within my DDW. (Of course, YMMV. I work full time, so on weekdays I can devote only a couple of hours each day to this, and coming up with stories takes me a long time. Hence, my SRS DDW is much lower than it is for many other RevTK-sters.)

@aovora
I encourage you to start logging your daily workloads. My approach for this is a bit laborious (but it has been helpful): I use my Mac's screen capture function to take snapshots of my Due and Interval graphs everyday after I finish doing my SRS work. (Facilitating, or even automating, this sort of daily logging is an area where Anki could easily improve, and it would not be too difficult to implement.) Examining these graphs over the last couple of weeks has given me a better handle on my SRS workload. (I.e. I get fewer surprises these days.)

gfb:
http://forum.koohii.com/showthread.php?p...1#pid95621

kendo99 Wrote:gfb:
http://forum.koohii.com/showthread.php?p...1#pid95621

thx!