+- kanji koohii FORUM (http://forum.koohii.com)
+-- Forum: Learning Japanese (http://forum.koohii.com/forum-4.html)
+--- Forum: Learning resources (http://forum.koohii.com/forum-9.html)
+--- Thread: "Review early" bad? (/thread-8899.html)
"Review early" bad? - frony0 - 2012-01-12
Hi guys,
I remember reading somewhere that reviewing early in anki should really be avoided at all costs, since it breaks the rhythm that the supermemo-based algorithm tries to get the memories into or something like that, but is that really true?
I understand that the algorithm works by reminding you of the fact at the optimum time (something like just before or just after you forget it), but then surely reviewing early would just reinforce the memory this round, and allow for a further interval for the next?
In the context of my kanji deck I come across a few kanji that I've learnt randomly in the wild every now and then and identify them, and I guess that would count as a kind of early reviewing, no?
"Review early" bad? - Betelgeuzah - 2012-01-12
I guess it's not bad per se.....just a waste of time.
I don't know much about the intricacies of Anki though!
"Review early" bad? - wccrawford - 2012-01-12
Betelgeuzah Wrote:I guess it's not bad per se.....just a waste of time.It's inefficient, that's all. Otherwise, trying to use your knowledge would be bad, and we all know that's not true.
I don't know much about the intricacies of Anki though!
"Review early" bad? - frony0 - 2012-01-13
Hmm, okay, thanks!
"Review early" bad? - Jarvik7 - 2012-01-13
It is potentially bad if you over review something you only remember short term and end up pushing it's interval too far ahead, by which time you forget it.
Simply "re-exposing" yourself to something too-soon is not a concern though.
"Review early" bad? - netsplitter - 2012-01-13
frony0 Wrote:I understand that the algorithm works by reminding you of the fact at the optimum time (something like just before or just after you forget it),No, it's not optimum. There's no way anki can read your mind to figure out the precise moment you will forget a fact. All intervals are approximations, and even intentionally skewed to avoid review spiking. Your mind is also capable of remembering a fact for much longer than its original due date.
Using the "review early" feature can be helpful if you know you have time to review today but not tomorrow (and on some days you're really into reviewing, so take advantage of the flow). Although you should probably avoid using it for relatively young cards, since you're kind of defeating the purpose of spaced repetition, and you're kind of breaking the algorithm as well by reviewing cards with 1-2 day spacing a day earlier. To avoid getting young cards, in Study Options, set the Display Order of reviews to "Review cards from largest interval". I normally prune the first 10 to 20 this way and stop.
"Review early" bad? - nadiatims - 2012-01-13
If I had some list of information that I know I'm going to be tested on tomorrow, and the consequences of the test matter, then I'd review the information a bunch of times to to ensure I remember it for tomorrow. With language learning however, how well you still remember some new fact tomorrow is way less important than how rapidly your stock of usable knowledge is growing and usable knowledge means knowledge that you're able to retrieve even without having reviewed it recently. You have to get facts to that point where you just know it and that comes from re-exposure of long times, having given your brain time to actually change it's structure. Think of some reasonably infrequent word that you hear maybe once a month if you use Japanese everyday. That means outside of srs, you'll get re-exposure to the word once a month: 1m, 1m, 1m, 1m, 1m, 1m....keep this up for 2 years, and you've been exposed to that word 24 times over the space of two years. You're gonna end up knowing it even if you forgot it the first 2 or 3 time. Now add in the srs and you get something like:
1min, 1hour, 1day, 1week, 1m, 1m/2m, 1m, 1m, 1m, 1m/4m, 1m, 1m, 1m, 1m, 1m, 1m, 1m, 1m/8m, 1m, 1m, 1m, 1m, 1m, 1m, 1m, 1m, 1m, 1m, 1m, 1m,1m, 1m, 1m, 1m/16m, 1m, 1m, 1m, 1m, 1m, 1m, 1m, 1m, 1m....
Basically the time you spent in the srs in the long run amounts to little more than a some background noise layered over a much stronger signal. In short it doesn't matter.
Now some may say sure, but not everyone uses Japanese daily. In that case the natural frequency drops, but then so does the usefulness of anything you're learning in the srs anyway. Basically the usefulness of the srs in learning is inversely proportional to the frequency with which you actually use japanese. Usefulness of the srs as a learning tool is inversely proportional to the actual usefulness of what you're using it to learn...
"Review early" bad? - frony0 - 2012-01-14
Wow, thanks for the explanations guys! I'll probably do as netsplitter suggested and do 10 or 20 if I'm "in the zone" but else I won't, for the sake of preserving the exposure.