Please excuse any poor English, I’m still learning!
I’ve been using the regular Anki scheduling with minor tweaks for about 5 years now, and I’ve come to the conclusion it’s not how I learn. Cards show up far too infrequently for me to really concretize the information.
Using default deck options / custom scheduling, is there a way to set Anki to use fixed intervals instead of multiplicative ones? For example, instead of seeing this when reviewing:
Default
1st review
Answer Good - 10 days
2nd review (10 days have passed since 1st review)
Answer Good - 25 days
3rd review (25 days have passed since 2nd review)
Answer Good - 2 months
I’d like it to add a set number of days to the interval, say +5 days each time I answer “Good”.
Fixed Interval
1st review
Answer Good - 10 days
2nd review (10 days have passed since 1st review)
Answer Good - 15 days
3rd review (15 days have passed since 2nd review)
Answer Good - 20 days
I essentially want “Easy” to add 15 days to the pre-existing interval, “Good” to add 5 days, and “Hard” to keep the same interval. Is this possible using default settings or the Custom Scheduling?
Thanks everyone who takes the time to look at this and/or answer it!
While there might be a way to achieve this using custom scheduling, I wouldn’t advise it. It’s not efficient. I would recommend you to take a look at FSRS, it allows you to adjust intervals based on desired retention aka how much you want to remember. You can set high desired retention to make all cards appear frequently.
Thanks for the fast response @Expertium, I appreciate it! I’m looking into now, but FSRS retention prediction also seems a bit overkill for my learning goals. Wouldn’t FSRS also attempt to modulate / grow those intervals?
If you set w8 to 0, w9 to 0.8, and w10 to 0.01, you can (almost) achieve fixed-interval scheduling. OP, in case you’re reading this, I don’t actually recommend it. I just think it’s interesting that theoretically, in some fringe case, FSRS could give the user fixed interval lengths.
If you’re having trouble remembering cards, I would recommend doing more with them when you are seeing them. Are you using images, mnemonics, practice, etc. (ehhh – it’s hard to envision what might work for you without knowing what you’re studying)?
With the default SM-2 algorithm, there are a LOT of things you can do to get more predictable scheduling – but I join others in wondering if that’s the best path for you. [The only reason I’m not recommending FSRS to you right off the bat is that it seems like you’re looking for predictability. While FSRS is the superior algorithm in pretty much every way, it’s not going to make things more predictable for you. But if you are willing to sacrifice predictability for something that will help you increase your retention and make studying easier – FSRS is your huckleberry.]
Based on your example, you do want your intervals to grow – just not by 2.5x with each success. But adding 5d to a 10d interval is 1.5x – to a 15d interval is 1.33x – to a 20d interval is 1.25x – with that system, you’d be having your intervals grow less and less the longer you successfully study the card. That seems backwards, doesn’t it? How about just adjusting your intervals to have slower growth?
Thank you all for your responses, and I very much appreciate the help to try and solve this!
@Danika_Dakika and @Expertium I certainly understand what you all mean, but with my current card load and my dozen or so decks, I’ve noticed my learning process doesn’t co-align with the default multiplicative Anki system. I think very many beginner Anki-users get turned off by the sometimes absurd length between cards, and I think I unfortunately still align with that camp. I want my intervals to grow, but I want that growth to be flat, not exponential.
I am using images and audio, and my retention is ok. I probably want this feature, because I use it more for learning new topics (languages) as opposed to creating a deck for strict review.
I think a fairly large, non-vocal group of users, do not want to use Anki to maximally save time and efficiency, but just as a simple review/learning software that is decently customizable (if that makes sense). I definitely fall in with that camp, although I wish my brain worked like the hardcore users.
I highly, and I mean highly doubt that fixed intervals are actually optimal for you. It’s not impossible, but extremely unlikely. I still advise you to give FSRS a try (see this post of mine, especially link 3 from that post) and choose a high value of desired retention. Also, out of curiosity, once you optimize parameters, please copy-paste them here so I can tell you what optimal interval growth looks like in your case.
You can, though I don’t advise it. @L.M.Sherlock maybe we should add another check to the field with parameters, and either automatically clamp the parameters, or give the user a warning about trying to set the parameters outside of their allowed ranges?
Completely unrelated, but I’m kinda frustrated by the whole “weights” thing. It was fixed ages ago, yet people still call parameters “weights”. I was not expecting that this bad terminology will spread so quickly and stick so much.
I hope that doesn’t happen. A regular user will not change the settings manually. And those who do it either figure it out or go where they don’t need to, that’s their problem.
I agree with you. The person has not even tried it, but already writes that it does not suit him.
Although I agree with him that many people use ANKI to periodically view the material.
This is all about the default weights, but over time the optimizer will be able to select such intervals that will correspond to your memory.
Thank you @Keks and @Expertium, I’ll try out the FSRS and tinker with it… The learning curve is a bit steep for someone not experienced in this field, but I’ll get it!
You are both probably right that whatever I am doing is sub-optimal. I don’t disagree. At this point, it’s likely more due to my note design and card load than anything related to the learning algorithm. But yes Keks, that is essentially what I use it for. Much more as an initial learning tool. FSRS seems like a great alternative according though to what many on the forums have said, so hopefully I find more luck with this. I’ll provide updates!