Creation of new landing page

Even you saw it.

Edit: Now my reply looks out of context.

@L.M.Sherlock the 43% is from your paper, right? I was just thinking that perhaps you should run simulations with FSRS-4.5 and compare cost (studying time) specifically with FSRS rather than with the model from the paper (which isn’t quite the same as FSRS).
Also, I think comparing spaced repetition with repetition at fixed intervals is more realistic. I would assume that most people review things at fixed intervals, like once per week, per month, or per semester (if they are college students), rather than completely randomly. But then we’d have to choose the interval length for the simulation.

You are right, and once a week is common here. But that still doesn’t mean fixed intervals if my interpretation of it is right.

Because think this, if you review every Sunday, you are reviewing everything from the past week. That probably includes’ Friday’s stuff. But also from Tuesday. So the intervals for the “first review” for anything is going to be varied. Also next Sunday, you might not cover everything that was taught in a week. Those will remain pending for a week. You also might not review already reviewed items from last Sunday. Just my two cents.

It’s pretty barebones at this point but it’s produced with the echarts library so it’s fully customisable - just need to learn how… Either way this demonstrates the rough idea. That echarts link should take you directly to the current code for the chart

1 Like

I think the final decision will be made by the official Anki Dae, so you might want to check his view before you start your development.
(Simply put, even if everyone in this thread agrees, if Dae doesn’t, it will not be implemented, so it’s not a majority vote.)
It should be beneficial to have lots of discussion and suggestions of ideas.

Does it mean each stuff only has one chance to review?

It’s hard to simulate the real case because it’s so random.

For fixed intervals, I have run an initial experiment:


No I didn’t mean that. At max, I will say you review something 5/6 times if you’re studying the course for a year. You also study for tests so including those will bump that number depending on how many tests you have.

My point was, because of time constraints you will never be able to review everything in fixed intervals. You’ll miss some of them. Those will be pending and you’ll only get to review them the next week/month.

One more thing to note, most students don’t really do that much “active recall” and I expect this to be true everywhere. I see a lot of people just passively reading over their notes thinking of it as “revising”/“reviewing” so in my opinion the time cost to learn something (in traditional method) will be much greater than you will find out.

It actually gets even more ambiguous, because many people skim over notes, not really studying everything, and they consider it revising but I don’t. Just adds so much burden without adding much of value.

I don’t have strong feelings on the graph. I would recommend focusing on an MVP though. Once we have a GitHub repo and CI building page updates, tweaks like the graph can then get bike-shedded in their own more-focused PRs.

1 Like

Sorry for the lack of activity recently, I’ve just had a lack of free time. The MVP for the website is ~90% implemented in SvelteKit now. I’m just finishing the downloads section and the navigation bar implementation. Unfortunately, I haven’t handled the i18n with fluent as I just don’t have the time to focus on that for now. I intend to have the MVP finished and publicly available in a GitHub repo quite soon so the code can be reviewed


I would add something like “State of the art spaced repetition algorithm” in Advantages, even if it’s without numbers