Marking Related Cards for Smarter Scheduling

Hello,

So I got to asking some questions about Anki to chatGPT and it proposed I request a feature be added. See below on the details.

Feature Proposal: Marking Related Cards for Smarter Scheduling

Summary:
I propose a feature where users can easily mark cards as “related” directly within the Anki browser, with the goal of adjusting their scheduling so that they aren’t shown on the same day. This feature would make it easier to study related questions (e.g., asking the same concept in multiple ways) without repeating the same card too often on the same day, similar to how Cloze deletions work.

How it Would Work:

  • Mark Related Cards: Users could select two or more cards in the Anki browser, right-click, and select an option like “Mark as Related.”

  • Automatic Scheduling Adjustment: Once cards are marked as related, Anki would adjust the scheduling algorithm to prevent these related cards from being reviewed on the same day.

  • Use Case: This feature would be especially useful for users who want to ask the same question in multiple ways (e.g., different phrasings) or have similar concepts tied together in different formats (like multiple-choice questions or fact-based cards).

Why It’s Useful:

  • It would allow for more efficient study sessions by preventing the user from reviewing the same material too closely together.

  • It could help create more natural review patterns when learning concepts from multiple angles.

  • It would streamline the process of managing related questions without the need to create complex card templates or manually control card decks.

Example Use Case:

  • Card 1: “What is the capital of France?”

  • Card 2: “Which city is the capital of France?”
    Both cards should be treated as related so that Anki doesn’t show both on the same day. Instead, they could be spaced out based on each card’s individual schedule.

ChatGPT is generally an idiot when it comes to Anki-related things, so there’s no need for you to forward its suggestions to us. It sounds like you worked on this one together, so to the extent this is your proposal –

You’re missing a crucial fact. If those cards are siblings (made from the same note), this is already easy to handle with “Burying.” If those cards aren’t siblings, they are duplicates, and you shouldn’t be splitting your review history by studying both of them. This doesn’t even fit the mold of asking the question a different way to avoid memorizing the card, because it’s just the same question.

Folks have had ideas like this over the years – some have been implemented as add-ons (which you can look for). But by far the best way to handle this is within a single note.

A better way to do this is to have both questions in the same card and have them rotate, look at this note type for example: https://ankiweb.net/shared/info/171015247

For hierarchical systems, this has been done beautifully in other SRS systems. Look at GitHub - satchelspencer/hsrs: a hierarchical spaced-repetition system for learning structurally related materials. (webapp). Math Academy also has a really good way of doing that: Optimized, Individualized Spaced Repetition in Hierarchical Knowledge Structures - Justin Skycak

I haven’t used them myself (because I only use Anki), but they look very promising. Hopefully, someday we’ll be able to do that in Anki too.


These suggestions have come up before, what ChatGPT is doing (given it has been trained on this data) is simply regurgitate what it has already seen in its training data.

I support the original author’s proposal. This would be very helpful when studying languages with compound verbs, such as German or Polish.

German example: kommen (to come), ankommen (to arrive), abkommen (to deviate), zukommen (to approach), entkommen (to escape)
Polish example: pisać (to write), podpisać (to sign), przepisać (to copy out), wypisać (to remove from a list), zapisać (to note down)

Spacing out similar verbs would prevent me from reviewing multiple related forms in the same session.

I agree with what you’ve said about the siblings. I’ve used them before in the past when i was studying Japanese. However; I think we’re still talking about the same relative concept here with the exception of what anki manages automatically vs what some of us would like to do manually.

If you have a note like the one above here is how I would do it in Anki as of current:

Note:

Front I “What is the capital of France?”

Front II “Which city is the capital of France?”

Back “Paris”

Two cards generated. Siblings. I assume this is what you meant.

What I’m proposing is using two or more of any note, and linking them together to form siblings.

Basic Note I

Q: What is the capital of France?

A: Paris

Basic Note II

Q: What is so special about Paris?

It’s the capital of France

You probably don’t want to see those cards in the same day. Make them ‘siblings’.

Another somewhat good example to illustrate is managing Japanese text. (May not be the best example, but again, it’s just to illustrate the point)

If you look up some vocabulary in a dictionary (used wwwjdic), there may be several common ways of writing a word, while another word may have just one common way of writing it. So now you have two different choices:

A) Generate the extreme case amount of fields you will need on a note, and write conditionals to prevent generating cards you dont need. (My understanding is that note fields are fixed)

B) Make a different note type / card generation per scenerio.

Front

hai

back

yes

That’s one template

Now, suppose I want to add a card that has three different writing styles

warui → bad

ワルイ → also bad

悪い → also bad.

Create a note that supports 3 writing styles generating 3 (or more) cards .

Suppose I want to have 10 different writing styles (typically not the case, but to illustrate the point)

What I’m proposing is to make the notes more dynamic by adding as many notes types as you want and make them siblings with other cards.

And no, this reply was not a chatGPT generated.

Relevant: Can I link cards together? - Anki FAQs

It’s possible to improve scheduling (contrary to what this says) if we drop the “every card is independent from every other card, unless they are siblings” assumption. But Anki wasn’t designed for that, and at this point it’s too late to redesign it. It just wasn’t designed with modern machine learning algorithms in mind.

I understand what you’re asking for – you want a way to link cards from separate notes as something akin to “siblings.” The answer is – there’s no way to do that. That’s why I suggested you look for what has been implemented in add-ons.

Sometimes when it’s difficult to come up with a solid example, it’s because there’s no realistic use case that can’t already be handled by existing functionality.

You offer this as an unacceptable solution, but it’s not as complicated as you make it out to be. In the long run, you should want related information to stay together on a single note, with each piece of information in its own field. It makes it much easier to manage and “future-proof” your collection.

You don’t need to anticipate every need now, but when you have a good idea later, you can add another card type, and your note type can expand to accommodate it. As I write this, I’m looking at my own primary language-learning note type that has expanded over the years to 23 fields, and makes up to 7 cards (and I still have plans for more).

So, a note that “supports 3 writing styles generating 3 (or more) cards” seems not that challenging, and when you’re ready for 10 writing styles, you can do that too! :sweat_smile:

You use case also sounds like it could be served by sibling cards created from the same note. That might mean you want a compound-verb-centric note type, that you can set up in a certain way for how you want to learn these overlapping words, but separating them into separate notes seems to work against you.

Why is it impossible to add to notes some sort of third field similar to #GUID? For example, an ID that links related notes, so if I review one note today with that sibling/relative ID, the rest of the “family” would automatically be buried.

In my case, I struggle to memorize text like definitions in Anki, and I have a habit of breaking them into atomized chunks and making them into separate notes. I can’t just separate chunks within a single note, because the texts are inconsistent and vary in size.

If it were possible to simply mark notes with a third Family-UID field, memorizing texts would be much easier. I could make chunks as needed, tag them with the same Family-ID, and not worry about overlap.

This could also be useful in cases where people need to memorize poems, or for actors who need to memorize scripts.

Just use the same note with multiple card types that show you different fields.

I’ve been using Math Academy for about 4 months now. It’s works amazingly.

Good to hear that ^‿^