I think all of us agreed on something like this one?
1 Like
Hmmm, I dont think it is important. In my experience, users are not interested in Anki even if I explain logically how Spaced Repetition works.
FSRS is effective, it is a latest model and high performance algorithm, and users like new things.
v3 scheduler has virtually nothing to do with scheduling efficiency (despite the name), and including it will make things confusing since v3 scheduler works with both SM-2 and FSRS, so if you plan to add a comparison of algorithms, just leave SM-2 and FSRS, though even that is a bit excessive IMO. I would leave only FSRS.
I think any comparison should be on a secondary page, if needed
Yeah.
What’s important is to make people understand what spaced repetition is, not why this spaced repetition software is better than the others, in my opinion.
I disagree on this. Both are important.
2 Likes
Sure, they’re both important, but for the main page? Also, I think just stating that Anki is free, open-source and that it uses the latest scheduling research should be enough. What I mean is that playing its own merits without comparing it with other options should be sufficient for a landing page. People who know of other SRS will specifically look for comparisons wherever the are. But I’m guessing most people that get to Anki’s website don’t even know there are other apps. But I might be wrong.
3 Likes
V3 scheduler is not an algorithm, but I think it is related to scheduling. It is related to the number of review cards and new cards, card order, Bury, Filtered decks, number of buttons, etc. Simply put, it is like a manual scheduler.
If these are used properly, many problems can be solved without add-ons, so in my opinion the importance of V3 is extremely high.
These were not intended to be comparisons, but to present the features related to Spaced Repetition that users can use in Anki.
I know you love FSRS but it is too early to do that, because SM2 is still the default.
The display of the description of Spaced Repetition (SM2, FSRS, V3) was just an idea, so it may not be necessary, and the current official Anki page does not seem to have a description of Spaced Repetition.
There are many important points of Anki besides the algorithm, such as synchronization and cross platform, etc.
1 Like
You’re completely correct, yes. 
1 Like
Yes, it’s the scheduler – not a scheduling algorithm. It’s not something you can compare to SM-2 or FSRS – each of them exists with v3 [choose your metaphor - on top of, underlying, surrounding, throughout, …]. Just like SM-2 previously existed on top of the v2 or v1 scheduler. The scheduler is very important, but not for any of the same reasons that the algorithm is.
2 Likes
I agree, V3 scheduler is not the same as or comparable to the algorithm.
In my opinion V3 scheduler is for enhancing the algorithm.
SM-2 and FSRS are open source and are used by other flash card apps. So emphasizing the FSRS algorithm does not differentiate Anki from other companies. In fact RemNote already has FSRS built in, so if comparing by algorithm, Anki and RemNote have the same performance.
But Anki’s SM-2 and FSRS are enhanced by Anki’s own V3 scheduler, so Anki’s Spaced Repetition should be more efficient than other companies (at least SM-2 should be enhanced).
So if you want to differentiate Anki from other apps by algorithm, the V3 scheduler is important, even if FSRS becomes more popular and common in the future, it should be possible to enhance Anki’s Spaced Repetition advantage by enhancing it to V4, V5, etc.
That’s all well and good – but all of this goes back you you putting the 3 side-by-side in a way that doesn’t make any sense (and under a heading that doesn’t make much sense either given what you’re trying to say about the algorithms and scheduler) –
4 Likes
Well, that’s just a crude idea. As already explained, the current official Anki page does not mention Spaced Repetition, so the Spaced Repetition section may be redundant.
If put these three side by side (just a idea), I think enclosing them in squares like this would solve the problem.(and fix the text)
1 Like
I understand your points, but I still think that this image would be very confusing no matter whar you wrote in the rectangles. Also, as Danika Dakika said, V3 scheduler is not something you can compare to SM-2 and FSRS. And no, having “Algorithm” and “Enhancement” is not better, sorry.
We need something that satisfies the following criteria:
- Numbers, not just words. Like “Ten times less work for the same retention” or something like that.
- It has to be easy to understand, so no log loss, RMSE, or anything of that kind.
Comparisons with other apps can be helpful to make Anki look like the best option, but I’m not 100% sure that they are necessary. I’m also not 100% sure that they are unnecessary. So I would leave that to someone who really knows what he’s doing.
1 Like
Hmmm, I feel fine but if so it would be better to remove it.
I think it would be good to be specific about the number of add-ons. Many of them are not available, but there are about 1400+ and growing, so I think there are about 1000+ available.
The number of shared decks would also be good, but the number is unknown.
- Fully Extensible
- There are a large number of add-ons available.
→ There are 1,000+ add-ons available.
1 Like
Comparisons are only necessary when there are other well-known alternatives. Shampoo commercials would need them. But iPhone would not. Pidilite is a company in my country that never compares it’s product with alternatives, they also never use generic product names. Most still know one of their products as M-Seal and not as Epoxy Compound. In cases where a brand of product becomes the product, you can completly ignore the alternatives. This also means that any reference to SRS shouldn’t reduce Anki to just another SRS software. So Anki employs spaced repetition in it’s algorithm but it’s not really a SRS software.
So Anki employs spaced repetition in it’s algorithm but it’s not really a SRS software.
I don’t understand what you’re trying to say here.
I think the current description in apps.ankiweb.net employs the same principle I’m talking of. When I said “Anki employs SRS but it’s not really a SRS software” it simply meant that the language used for Anki shouldn’t allow the readers identify Anki with a generic name. To give another Pidilite example, their product Fevikwik is simply a gum but they never say the word. If I’m not wrong, Apple also doesn’t call iPhone smartphones even though it is one. So as Shigeyuki said we need to mention Spaced Repetition, I’m trying to point out the sort of language that should be used. The perception should be that “Anki itself is the product” and that perception already exists I think. (remember that in marketing, the word product wouldn’t mean a particular brand of shampoo but shampoo itself, that’s not always the case)
My other view is that since Anki is free, there is no need to promote it in the first place. In short, it is a $0 product so there is no need to try hard to sell it.
So if you want to promote it, I think you need to emphasize the selling points such as SRS and FSRS, but if not, I think you can remove all the promotion and focus on functionality.
Okay so I think we need to agree on the metric we will use for the graph. The comparison of Anki’s scheduling efficiency with randomly reviewing cards is likely the best choice here as I would take a wild guess that the majority of people stumbling across Anki would be unaware of other spaced repetition apps. Either way, comparing the efficiency of Anki’s scheduling with randomly reviewing places more emphasis on Anki itself, and less of a focus on how it compares to ‘competitors’ in the space. If it’s agreed that this is likely the best metric FOR USE IN A GRAPH* then I believe we should agree on a suitable visualisation.
I would also like some clarification. When stating Anki has a 43% cost reduction in comparison to random reviews, is this effectively saying you spend 43% less time studying in order to achieve the same retention?
If the above is true, maybe a horizontal bar chart showing “time spent studying” with two bars, random (100%) and Anki (43%) is suitable?
Open to suggestions though
2 Likes
@L.M.Sherlock Is 43% cost reduction = 43% less time?
100% - 43% = 57% should be used probably. That shows 43% reduction.
One more thing to remember, it’s at least 43% cost reduction as most people irl would review things past 50% retrievability.
Yeah. The cost is time cost.
2 Likes
Where did you get the 43% figure from? From Jarret’s paper?