Forgetting curve of first 24 hours, Steps and Graduating of New cards.

Helerzich's Avatar

Helerzich

23 Jul, 2015 06:59 PM

According to this graph: http://revunote.com/wp-content/uploads/2012/04/forgetting1.jpg - one should use additional 'steps' for 'new cards' before 'graduating' them. So the efficiency should become much more higher. Did anybody try to use additional 'steps' at the range of 45 - 600 minutes? The learning of 'new cards' should become more boring and time expensive, but 'graduated cards' should become more stable and there should be less mistakes on the next day (my 'graduating interval' is one day). Am I wrong? Or this graph is completely wrong?

And one more question. We all know the recommended 'retention rate' (about 90%) for 'mature cards' and we know how to use stats of 'mature cards' to calculate the 'interval modifier'. What is the recommended 'retencion rate' for the 'graduated card' which appears the first time after its 'graduating'? Can I get these stats? Can I use them to make correct 'steps' for 'new cards'?

  1. 1 Posted by Vit on 31 Jul, 2015 04:32 AM

    Vit's Avatar

    According to forgetting curve, one should use additional 'steps' for 'new cards' before 'graduating' them.

    "Should use" is not the right words ; "May need to use " is. You may need intra-day Steps only for Difficult material. How would you know if you need them ? For two days, track your retention for a Step YOU think you need. If it is >98% , increase the Step. Repeat the process until you get 95% ( let's say it was 30min step ). Now you add the second step of 60min and track the Retention ( now you do it for Both 30min and the second step ).

    Did anybody try to use additional 'steps' at the range of 45 - 600 minutes?

    This info would be useless for all but the original user. The length of Interval/Step will depend on ( Anki): material difficulty and how well you've learned it prior to making cards.

    SuperMemo tracks >10 difficulty levels. So if the material varies in difficulty, you have to keep the subjects in separate decks and obviously - have unique Option_Groups for each deck.

    The learning of 'new cards' should become more boring and time expensive...

    It was opposite for me when learning Japanese symbols; was challenging and fun -- it was "symbols' against me :-)
    As to the extra time - it is only 20 or so cards; it is negligible increase.

    ... but 'graduated cards' should become more stable and there should be less mistakes on the next day (my 'graduating interval' is one day). Am I wrong?

    No you are correct.
    SuperMemo: when a cards is lapsed, you have to re-learn it in 3 steps -- to compensate for the 'bad' insufficient initial learning.
    The Learning curve you are citing is just a curve to show that memory decays exponentially. Do not pay any attention to those numbers and retention - they are unique to that 60y old experiment.

    What is the recommended 'retention rate' for the 'graduated card' which appears the first time after its 'graduating' ?

    I could not find any .
    What i do is tracking MY retention for the First rep, separately for Active and passive vocab cards.

    Can I get these stats? Can I use them to make correct 'steps' for 'new cards'?

    You are the second person who is asking this important question.
    We have to wait for a programmer willing to write an Add-on ( easy task, IMO ).
    Meanwhile some new users keep using 1 10 min steps thinking that these are recommendations ! Students accept the 250% ease which will very quickly hurl the cards way past the exam date.

  2. 2 Posted by Helerzich on 08 Aug, 2015 06:55 AM

    Helerzich's Avatar

    Thank you, Vit! Your replies are always very useful.

    One more question: as you advise to me, I am going to create different decks. And relying on statistics of my retention rate, one of them should have about 80%, the second 125% and the third about 150% of interval modifier. As you see - they have very different difficulty.

    But when it comes to learinig - ANKI every day force me to learn the first deck, then the second and then the third. This seems to go wrong with principles of incremental learnig (mixing different facts is good), and it is very boring to learn a such way. I would like ANKI to mix them. Is there any solution or plugin?

  3. 3 Posted by A on 10 Aug, 2015 04:56 PM

    A's Avatar

    But when it comes to learinig - ANKI every day force me to learn the first deck, then the second and then the third. ......I would like ANKI to mix them. Is there any solution or plugin?

    If those are parent decks you are talking about then no, there's still a workaround though, you could turn them into child decks by creating a parent deck for all of them.
    If they are child decks, then you simply edit the options for the parent deck and change Order>Show new cards in random order.

    Try messing around with the parent settings and child settings to get what you like.

  4. 4 Posted by Vit on 11 Aug, 2015 05:17 PM

    Vit's Avatar

    I would like ANKI to mix cards ( from different sub-decks ). Is there any solution or plugin?

    Manual: "Learning cards are fetched from ALL decks at once and shown in the order they are due." That is the way Anki works, for now; for details, search Manual for 'Display order'.
    Soren: In the next release, we are considering adding ... randomization. "

    I did not try any special tools to achieve more randomeness then Anki normally does because i do not feel any inconvenience from not enough randomeness.


    What i do is choose a deck to review; if i 'get bored/tired' with deck-1, i go to another deck or/and take a short break to do body-weight exercises -- to make blood flowing.


    There are 80 Discussions, in Anki forums, with 'Random' in the Topic.

    I want to add these points:

    1. There is a recomendation to put ALL cards in one deck and using Tags for the Study Topics; i would never do that just to get a 'mythical' benefit from randomeness ( by mixing diff tipics ). Tags in the Browser do not provide a Nice list of Decks which shows the number of Cards by categories: New Due Unseen Buried. There are also other considerations against a Single deck.

    2. Using a Filtered deck looks good; i did try it for Two decks and it worked -- Anki jumped from one deck to another (not with every card of course ).
      The search string was:
      (deck:A or deck B) is:due You will be able to see, in the Browser, the Order the cards will be shown for Review and assess the degree of randomeness (hard to do because the Actual deck name is replaced with name 'Filyered); you may not like what you see.
      I did not include or is:new bec i study New cards separately ( Options_Group settings are different for them ).

    3. Make sure you learn New cards in their own decks otherwise some of them may never show up and remain in limbo forever; this will occur if:

    Manual: (notes inside parentheses are mine )
    "If you fall behind in your reviews ( on Review cards ), Anki will prioritize cards that have been waiting the longest. It does this by taking the 200 ( Review )cards that have been waiting the longest and showing them to you in a random order up until your ( Home Deck ) daily review limit (for Review cards). This ordering ensures that no ( Review )cards will be left waiting indefinitely, but it ( also ) means that if you introduce New cards, their reviews won’t appear until you’ve gotten through your ( Review cards ) backlog.

    God luck.

  5. 5 Posted by Helerzich on 25 Aug, 2015 03:16 AM

    Helerzich's Avatar

    Thank you Vit. I think my randomness problems are because I always make my dues (I have no limit for dues) everyday. In my case, randomess is very useful - because I have a deck which consists only of numbers. And another deck with clinical situations (which have long questions and short answers). Another deck is for language studying. So it's very useful for me to mix them. Thank you again. I tried cramming, it's nice but I have some difficulties with it. For not going off top I am going to make another thread.

  6. 6 Posted by Vit on 27 Aug, 2015 12:09 AM

    Vit's Avatar

    Hi, Helerzich

    I tried cramming, it's nice but I have some difficulties with it.

    Yes, the Learning cards are the problem; to empty the deck properly one has to wait till Learning cards are due again and 'earn' the Good rating.

    Let me hi-lite the details of my first very simple workaround ( to achieve Randomness ) i suggested before:
    1 Start your daily study in Deck-1;
    2 When you feel like going to Deck-2, click on the 'D' on keyboard --the list of decks will be displayed ( with the Current deck 'shadowed' so that it is easier for you to aim at Deck-2 .
    3 Click on Deck-2;
    4 Hit the Spacebar or Enter;
    You are reviewing a card from Deck-2 and can switch to any other deck at any time
    with only 2 clicks.

    There is also a good ( but not as good as the one i just delineated ) method to literally work around the Learning cards left in the Filter deck.

  7. System closed this discussion on 22 Feb, 2016 11:46 PM.

Comments are currently closed for this discussion. You can start a new one.

Keyboard shortcuts

Generic

? Show this help
ESC Blurs the current field

Comment Form

r Focus the comment reply box
^ + ↩ Submit the comment

You can use Command ⌘ instead of Control ^ on Mac