TL;DR: You've been using Anki right for ages; why don't you start using it wrong?
Using Anki for tasks it's suboptimal for has synergistic benefits for someone already using Anki quite a lot.[1] There are many great introductions to Anki, so I offer undiluted hot takes, and I add whimsicality to teach by example.[2]
Using Anki right
Language learning
Don't. It works, but it's much more inefficient than learning languages the right way: using Input Hypothesis learning allows you not only to have much more fun, but also to use time more efficiently by learning other stuff at the same time.
Human memory is proof God friendzones us and wants us to despair about his mixed signals
For the rest of your life, you will never, ever, forget that "a funny face guy inherited land from his four grandparents some time in the 1300s or maybe the 1600s, and also for some reason this makes me think of that white dude's biting face quote from that funny what-was-it-called movie".
For the rest of your life, if anyone asks you to say the name of that king, you won't remember.
You will never be able to remember the things you want to remember in the same natural, effortless way in which you remember the things God wants you to remember.
God created whimsical memories; all else has to be the work of man.
Scaling up the the time or compute you use to try to infer the answers to your Anki questions won't substantially change how well you remember them. Deep learning with Anki is only possible on the basis of Kronecker's finitist understanding of human memory, as an intuitionistic approach to Anki that rejects all cards except those that can be constructed from intuitive whimsical memories. One finitist,Anki-enlightenedajahn puts it so:
The art of Anki is finding the cards that specifically prompt you to autocomplete correctly; brute-force memorization and regurgitation is much more of a slog than casting spotlights on specific features of your mental structures.
This is the whole Anki; all the rest is commentary.
ε-machining your Anki cards
The perfect Anki card should be maximally short but contain all the information from your past life which can be useful for the retrieval of a certain word/number in the future. "Maximally short" means that it should be objectively ambiguous/incomprehensible for others (otherwise you aren't being concise enough); in the next section I advocate for using up to 5 words. Containing all useful information for future retrieval means that, even if you might forget the answer, you should never forget the meaning of your question. If you don't immediately know what you're being asked when reading a question, the question is wrong. It might happen sometimes that you read a question, don't get it, and after some seconds (or after reading the answer) you remember the original God-given intuition and why you wrote your card. This is a good sign! It means the intuition is still present in your mind, but your maximally short question now is a bit too short. Add a couple of words to make it obvious to future you what they're being asked.
Rarely it might happen that you have completely forgotten what the original connection was and you might have to change the question to a boring rote memorization on. This happens almost never to me, so maybe the key is rather to get an authentic feeling of when your brain has formed a monkey-brain connection, and avoid lying yourself into thinking two things are monkey-brain connected just because you wish them to be.
Epitomize Laconically in 5
Your question shouldn't have more than 5 words, and the answer should be one single word or number. Some strategies:
Avoid using grammar and syntax
Develop a system of symbols and abbreviations. I use ⌛ for "year in which the following event happened" or sometimes "year I have chosen to associate with a vague phenomenon that doesn't have an objective start of end",⌛⌛ for "year in which the following event started and year in which it ended",[3] 🗭 for "author of the quote, or author of the concept, or "creator" in a possibly very vague sense of creating" ➡️ for "what happened after this, possibly in a very vague sense of happening".
Use images instead of words
Related: Save yourself some brain cents by always writing Charles 5 and 1500s, never Charles V or the 16th century.
The specifics of a card that is perfect for you will depend on your idiosyncrasies, and you'll need to test to get a feeling of them, but here's a good starting point:
Another advantage of images, colorful Unicode symbols, and ELI5 in general, is that short, whimsical cards are fun, especially when you answer 50 in a row and get 45 right. On the other hand, you won't memorize anything if you get bored of Anki. So never ask yourself if you really need an image or a symbol, but instead try to always have them unless reading the question you think about the meta level of how weird the question is (even asking yourself if it sounds weird is bad; change the question) instead of staying at the object level of thinking "oh yeah, I remember that! What was his name?".
For some people, sexual imagery or racial slurs or blasphemy or whatever your taboo is might become associated in your brain with things you are trying to remember: always use them (and possibly never share your deck with anyone)[4], under pain of being pruder than medieval priests.
Add any pun that comes to mind, or write "grandparentZ" if it makes you chuckle to imagine Charles 5 dressed like a rapper and dancing all around the globe in a never ending night party.
Regarding feelings, Anki enlightenment is more demanding than bodhi, which only demands that we feel indifference towards our emotions, like the Mahabharata king who would experience nothing if one of his arms were cut while the other was perfumed. Anki enlightenment demands from us that we transmute every emotion into undifferentiated joy, be it the pride of getting an answer right, the titillation of a bad pun, or the flow state reached when quickly answering many ELI5 questions right. But, furthermore, it also demands that we become attached to having fun while using Anki.
Principle of Excluded Boredom (PEB)
A less advanced Anki user might ask whether things like "grandparentZ" don't contradict the well known maxim that every peculiarity in a card makes it less likely that you'll remember the relevant information when real life prompts you in a non-descript manner.
But that reader obviously isn't using incremental reading to memorize the previous section about intuitionism[5], or else they would remember that contradiction is not the way to prove theorems in Anki science, in which, way above the law of noncontradiction, we find the following Principle of Excluded Boredom:
Having an Anki burnout will be so bad for your memorization that not getting bored trumps every other consideration.
A student asked, "Does the dog card belong in the "Buddha nature" deck, or in the "non Buddha nature" deck? Ganto picked up an axe and deleted the dog card.
You should ruthlessly delete every card you find both boring and difficult to memorize.
You probably should delete boring cards even if you find them easy, but never spend more than one second thinking about this. If in doubt, delete them.
You should think about whether you are really interested in the things you keep forgetting
You should send cards which awake in you a lower-than-maximal interest to your OASR deck.
Similarly to the idea of not using idiosincratic questions being a true but ultimately non absolute statement overcome by the PEB, the hidden Cesàro cost is a concept for which the aspiring Anki-ajahn must learn both its truth and its insufficiency. We start with its truth:
Over a lifetime using Anki, every question will have come up a certain number of times, the higher the bigger the number of times that the answer was wrong. The summation of all the seconds spent with one specific question is its hidden Cesàro cost.
Anki-ajahns who meditate on the HCC of a card sometimes develop detachment from it, and delete it if the conventionally defined threshold of is crossed.
The Yes Free Lunch Theorem (YFLT)
The HCC is not only buddhavacana, but is in fact a very advanced teaching that every Anki-ajahn should internalize. And yet, the HCC is not true in the absolute sense.
The HCC wrongly presupposes that mental energy is limited.[6] In fact, having a (low) percentage of your deck be about trivia from the latest Netflix series on Brouwer's turbulent love life, mysticism and fist fights with Hilbert is a very efficient way of learning serious stuff.
Memo-in-Sup consists in using one card for more than one fact, at the price of exponentially longer (and thus less fun) questions, and a certain memorization error. For example:
Answering these questions right won't make you remember all 10 facts, nor produce the whole quote out of context. But it will help produce some part of both. A different point along fun vs error trade-off is found in the dual card where we ask about the whole thing, but accept not completely right answer. See the second OASR example for more.
Etymological aside
The L in the LJ lemma comes from the Lindenstrauss family, a prime[9] example of non-atomic data (a nuclear family), where each member had published papers listed in Mathematical Reviews. The surnames of four mathematicians could thus be memorized with the use of a single Anki card. The eponymous Johnson, Samuel, wrote:[10]
Knowledge is of two kinds. We know a subject ourselves, or we know where we can find information upon it.
Latitudinarian uses of Anki
Having a deck with exclusively short, fun questions that you feel intrinsically curious about is not only a lot of fun, but also extremely effective. Prune the deck to keep it fun, and be strict with what "getting an answer right" means to keep it effective.
But what should you do if, as if often happens, you also half-extrinsically would like to memorize some stuff but like, don't really care super much about it and forcing you to ELI5 them would be boring? We already saw one example of this in the Memo-in-Sup section, where we traded off some fun and some error tolerance to gain some memorization. The generalization of that trade-off is called "half-assing". When considered as a problem in bounded rationality, half-assing is also called "optimally assing".
Since optimally assed questions are less fun than the ELI5 questions described in the previous section, at this point (but not before), it becomes necessary to divide our cards into two decks: the maximally joyful one, and the other one.
It is recommended to name the decks "1" and "2"[11]. While decluttering the names of your decks adds very little fun per se, an aspiring Anki-ajahn should be suspicious of rationalizations of why one needs names longer than one character. They are symptomatic of you not having the right attitude, which will cause a decrease in fun in other situations.
Decluttering the number of decks is even more important. Most of the thoughts the come to mind about the creation of further decks arise from attachment, and the true Anki-ajahn sees those thoughts arise with indifference, neither repressing them nor acting on them. Some exceptional legitimate uses of new decks will be discussed here and here.
An attentive reader might notice the contradiction between claiming that deck ELI5* should only contain maximally fun questions, and our having accepted a trade-off in fun in the Memo-in-Sup section. The attentive reader has indeed noticed a point of Anki doctrine where exact science is not possible, and only heuristics exist. For more information, see YFLT's converse.
Seeing a question, not getting it right, and clicking on "Easy" creates very vague memories. Very vague memories are better than no memories.
OASR shares some properties with Memo-in-Sup, and indeed is its dual: Both are trade-offs in memorization error / fun space, and while Memo-in-Sup preserves strictness about what "getting it right" means, and within that constrain maximize the funness for a non-atomic fact, OASR sacrifices the "getting it right" premise, and because of that is compatible with lower values of fun.
OASR has another advantage: As mentioned in the Samuel Johnson quote, it is often helpful to know where some knowledge is found. Having a vague recollection is often enough to find the card in your Anki collection[12].
If you only remember "some funny fact about grandparents", typing "grandparents" in Anki's search function will provide a bounded list of candidates to inspect, whereas Claude won't be able to be of any help. For a generalization of the search function, see LINK D.
Scheduling principles of OASR
Set very long spacing intervals
"Easy" means "I don't particularly want to be reminded of this any time soon"
Be very generous with what getting it right means ("It started with C" or "Charlie 4" are right answer unless you feel like thinking again about Charles V in the near future)
Failed questions don't get asked the next day, but after 5 or 10 days. It is really annoying to have a backlog of 30 questions you got wrong yesterday, and to fail 25 of them, and then find a backlog of 40 the day after that, 25 of which you know you probably won't learn. Anki-ajahns might want might want to apply this to the ELI5* deck as well.
Ignore the daily duties and let the questions well up. Only "study" this deck when you are in a situation in which you would otherwise be scrolling (bus, toilet, etc.). Think of it as "I would be wasting time otherwise, and this is not so boring, so I get to have some fun and feel proud of my discipline at the same time".
Kinds of cards fit for OASR
ELI5 cards for which you don't have enough enthusiasm
Cards about connected atomic facts
Cards about a long, non-divisible piece of information, like a quote.
The second and third kind are sometimes known as "long-form OASR". Here is one example of each:
-Burgundian Netherlands (from his father Philip the Handsome, 1506),
-Spain plus its American possessions (from his mother Joanna and grandfather Ferdinand, 1516)
-Austrian Habsburg lands plus the imperial claim (from Maximilian I, 1519).
-Elected Emperor in 1519, beating Francis I of France, financed by loans from the Fugger banking house.
-His mother Joanna ("la Loca") remained nominal co-queen of Castile, confined at Tordesillas, until her death in 1555.
Q: Apocryphal quote attributed to Charles V
"I speak Spanish to God,
Italian to women,
French to men,
and German to my horse"
Three reflections: These questions are long and thus boring. They should only be read in full if your first thought when starting them is "oh yeah, that was a nice quote / bunch of fact, how was it?".
This implies being very tolerant with the answers, to the point where "Easy" means "am I particularly interested in getting this question again, to possibly read it in full then?". If an answer is provided at all, basically anything should be accepted. Getting 1 fact right and 2 kind-of-right should automatically implies pressing the "Easy" button.
For some questions[14], marking the central word of the quote with a different color[15] is recommended, so answers can be evaluated (if at all) based on how close they are to that one word. "Something about horses" thus counts as right. (Saving time and keeping it fun still have non-zero importance in the OASR deck).
More Anki theorems
Rebirth in a new deck
The joy & strictness division between the ELI5* and OASR decks is essential. The doctrine of impermanence teaches that the division must be constantly updated; one should always be on the lookout for cards that now find themselves in the wrong deck.
If one gets an ELI5* answer wrong, and one notices that the level of enthusiasm is low (asking oneself automatically means it's low), but high enough that one doesn't want to delete it, the card actually has an OASR nature.
If one gets an OASR card right and gets a smile at the question/topic/whatever, the card has an ELI5* nature despite the appearances. One will need to refactor the question as an fun question, which might involve fragmenting it into several cards, using less grammar, etc. If the card is about a question, one should use a cloze question and hide one single word, no exceptions. The word should be uncommon enough that one never answer with a synonym that forces one to waste time thinking about whether the answer counts as right or wrong (if this happens with a ELI5* question, either the question was ill posed or the answer is not fit for ELI5*, and one has to rethink how to refactor the card).
The untrained mind often sees a transmigration from ELI5* to OASR as a demotion to a lower kind of life, but one should become unattached to those feelings. Also, when moving in that direction, one should not reset the card; from OASR to ELI5* one may do so.
YFLT as tertiary choice
Having gained the tools from the OASR section, we now see that the choice is not binary but tertiary: a card can be kept in ELI5*, sent to OASR, or be deleted.
The analysis also becomes more accurate with the added variable of Google-independence, which measures how difficult it is to find a certain information with a given memory of it. For example, the fact that the Significs group was founded by van Eeden is something than can be googled in five seconds; it has zero Google-independence. We mentioned the example of trying to find Charles 5 from "some funny fact about grandparents"; this has infinite Google-independence.
In ELI5*, the lower the Google-independence, the higher the threshold of enthusiasm to delete a card, so that e.g. the question "Founder of Significs group" should be deleted as soon as it stops being maximally fun. The questions about Charles, however, should be sent to OASR while there is some wish of having a vague memory of them, even if they aren't maximally fun.
In OASR there is no place for cards with low Google-independence; if the user isn't particularly interested in memorizing them, but they can be easily googled, the user can simply google them if they ever need the answer.
YFLT's converse
The Yes Free Lunch Theorem proved that maximally fun cards should be kept in ELI5* and not deleted. In the ELI5 context, we skipped asking the converse question, and in the Joy Division section it was announced that the question doesn't have an objective answer. Now we have the tools to dive deeper into the question:
Should cards in ELI5* be maximally fun?
A scientifically correct answer would specify the levels of boredom and Google-independence that require a question to be sent to OASR or be deleted. Such answer, however, doesn't exist, and each Anki-ajahn is forced to find heuristics of what works for them. An ajahn able to skim Memo-in-Sup questions quickly, and answer them without reading them whole, will be able to keep them as relatively-fun ELI5* cards, whereas an ajahn with a tendency to read cards whole will be forced to restrict the ELI5* deck to only proper ELI5 card, since they will otherwise kill the speed and fun of it.
(I have to publish this post 50% unfinished because I need to publish something before tonight's Inscribe deadline)
I wanted to quote a certain short, vaguely remembered post by Caplan or Hanson on teaching by example, but I couldn't find it. I'd be thankful to any reader who remembers it.
This contradicts the "one word per answer" rule in the letter, but not in the spirit. If I am a history buff trying to memorize the exact years of a war for which I have lots of context that makes me care about it, your brain will save both numbers in one slot, and the double question is absolutely fine. If you have so little context that it feels like memorizing two random numbers, you should either create two cards or (much better) ask only about either of them, or the year in which the main fighting happened, or (even better) internalize that memorizing numbers that feel random comes from having misunderstood what kinds of things one should use Anki for. What does "number don't feel random" feel like? Being a history nerd, it took me a couple of second to understand David Bowie's song title "Aladdin Sane (1913–1938–197?)" the first time I saw it. On the other hand, not being a biology nerd, I always fail to memorize Cell Biology by the Numbers, or to be particularly good at estimation games. Any reader who feels the call to write anything making those number feel non-random will gain my eternal gratitude.
Back when I was more alienated from my emotions, I thought about work like this: You have a budget of energy to spend, and the goal is to efficiently convert that energy into impact. You identify the activity that generates the highest impact per unit of energy, and then you spam that button for as many waking hours as you can stomach. Is it any wonder I was always burning out?
What I failed to understand is that energy is not fixed. It’s dramatically affected by what you choose to do — some activities increase your energy over time, others deplete it. This feels so obvious to me now that I feel kind of stupid including it here, but I know many people (they are endemic to the Bay Area) who still labor under the false belief. Whether you enjoy what you’re doing affects your ability to do it year in and year out, and dismissing this as “selfish” doesn’t make it untrue. As my husband puts it, the best use of effortful motivation is to engineer your life to require less of it.
The earliest reference to the quote comes 50 years after Charles 5's death, contains no horse, and refers to what we would now call Dutch. When the saying is used in Spanish, speakers mistakenly think it refers to modern German, hence my translation.
In order to alleviate the confusion of the skimmers, "decks 1 and 2" will be referred to as the "ELI5* [15]and OASR decks", imposing on true readers a tax of 4 extra redundant characters every time a deck is mentioned. As an exercise for home, and assuming that the trade-off is justified on utilitarian grounds, calculate the percentage of true readers vs skimmers that the author expects to exist among his audience.
This result is easily provable for finite collections, but whether it also holds for infinite ones is still an open question. For finite but large collections the result can also be proven, but the construction requires some technicalities to be discussed later. LINK C
If the reader attempts to count the number of facts in the example question, they will realize that "facts" are a fake ontological category. Insistence in using questions with fake ontological categories is prone to cause ontological crises in which the ajahn must redefine their utility function, i.e. tell themselves that the obviously boring activity they are partaking in is the way Anki is meant to be used, and that they are having fun.
This was possible in our second example (other words like "German" or "talk" could have been chosen as easily). The first example shows a question where this is not feasible or sensible, unless one has a particular interest in a specific fact and decides to accept every answer that mentions it.
TL;DR: You've been using Anki right for ages; why don't you start using it wrong?
Using Anki for tasks it's suboptimal for has synergistic benefits for someone already using Anki quite a lot.[1] There are many great introductions to Anki, so I offer undiluted hot takes, and I add whimsicality to teach by example.[2]
Using Anki right
Language learning
Don't. It works, but it's much more inefficient than learning languages the right way: using Input Hypothesis learning allows you not only to have much more fun, but also to use time more efficiently by learning other stuff at the same time.
Isolated facts
A use case
One day, deep in a rabbit hole, you learn about a king who inherited/conquered an empire on which the sun never set. You want to remember the guy's name. He also happens to have a face so funny that its funniness deserves its own Wikipedia page. Coincidentally, that same evening you watch When Harry Met Sally... and laugh at the "white man's overbite".
Human memory is proof God friendzones us and wants us to despair about his mixed signals
For the rest of your life, you will never, ever, forget that "a funny face guy inherited land from his four grandparents some time in the 1300s or maybe the 1600s, and also for some reason this makes me think of that white dude's biting face quote from that funny what-was-it-called movie".
For the rest of your life, if anyone asks you to say the name of that king, you won't remember.
Kronecker's bitter lesson
You will never be able to remember the things you want to remember in the same natural, effortless way in which you remember the things God wants you to remember.
Scaling up the the time or compute you use to try to infer the answers to your Anki questions won't substantially change how well you remember them. Deep learning with Anki is only possible on the basis of Kronecker's finitist understanding of human memory, as an intuitionistic approach to Anki that rejects all cards except those that can be constructed from intuitive whimsical memories. One finitist, Anki-enlightened ajahn puts it so:
This is the whole Anki; all the rest is commentary.
ε-machining your Anki cards
The perfect Anki card should be maximally short but contain all the information from your past life which can be useful for the retrieval of a certain word/number in the future. "Maximally short" means that it should be objectively ambiguous/incomprehensible for others (otherwise you aren't being concise enough); in the next section I advocate for using up to 5 words. Containing all useful information for future retrieval means that, even if you might forget the answer, you should never forget the meaning of your question. If you don't immediately know what you're being asked when reading a question, the question is wrong. It might happen sometimes that you read a question, don't get it, and after some seconds (or after reading the answer) you remember the original God-given intuition and why you wrote your card. This is a good sign! It means the intuition is still present in your mind, but your maximally short question now is a bit too short. Add a couple of words to make it obvious to future you what they're being asked.
Rarely it might happen that you have completely forgotten what the original connection was and you might have to change the question to a boring rote memorization on. This happens almost never to me, so maybe the key is rather to get an authentic feeling of when your brain has formed a monkey-brain connection, and avoid lying yourself into thinking two things are monkey-brain connected just because you wish them to be.
Epitomize Laconically in 5
Your question shouldn't have more than 5 words, and the answer should be one single word or number. Some strategies:
The specifics of a card that is perfect for you will depend on your idiosyncrasies, and you'll need to test to get a feeling of them, but here's a good starting point:
1500s 4 grandparents -> never sunset
Some instrumental convergence theorems
You can't memorize the coffee if you're bored
Another advantage of images, colorful Unicode symbols, and ELI5 in general, is that short, whimsical cards are fun, especially when you answer 50 in a row and get 45 right. On the other hand, you won't memorize anything if you get bored of Anki. So never ask yourself if you really need an image or a symbol, but instead try to always have them unless reading the question you think about the meta level of how weird the question is (even asking yourself if it sounds weird is bad; change the question) instead of staying at the object level of thinking "oh yeah, I remember that! What was his name?".
For some people, sexual imagery or racial slurs or blasphemy or whatever your taboo is might become associated in your brain with things you are trying to remember: always use them (and possibly never share your deck with anyone)[4], under pain of being pruder than medieval priests.
Add any pun that comes to mind, or write "grandparentZ" if it makes you chuckle to imagine Charles 5 dressed like a rapper and dancing all around the globe in a never ending night party.
Embhodid ajahncy
Bodhi (enlightenment) requires developing an equal vision that sees no difference between a priest and a dog, or even between a priest and a dog eater. Anki enlightenment requires queering the serious-ridiculous binary; it requires not thinking of cards as having "serious content" and "added jokes", but seeing the underlying monkey-brain connections that underlie and pervade everything.
Regarding feelings, Anki enlightenment is more demanding than bodhi, which only demands that we feel indifference towards our emotions, like the Mahabharata king who would experience nothing if one of his arms were cut while the other was perfumed. Anki enlightenment demands from us that we transmute every emotion into undifferentiated joy, be it the pride of getting an answer right, the titillation of a bad pun, or the flow state reached when quickly answering many ELI5 questions right. But, furthermore, it also demands that we become attached to having fun while using Anki.
Principle of Excluded Boredom (PEB)
A less advanced Anki user might ask whether things like "grandparentZ" don't contradict the well known maxim that every peculiarity in a card makes it less likely that you'll remember the relevant information when real life prompts you in a non-descript manner.
But that reader obviously isn't using incremental reading to memorize the previous section about intuitionism[5], or else they would remember that contradiction is not the way to prove theorems in Anki science, in which, way above the law of noncontradiction, we find the following Principle of Excluded Boredom:
From PEB, Shanto's Axe and the Yes Free Lunch Theorem follow, as well as the concept of OASR.
Ganto's Axe
The hidden Cesàro cost (HCC)
Similarly to the idea of not using idiosincratic questions being a true but ultimately non absolute statement overcome by the PEB, the hidden Cesàro cost is a concept for which the aspiring Anki-ajahn must learn both its truth and its insufficiency. We start with its truth:
Over a lifetime using Anki, every question will have come up a certain number of times, the higher the bigger the number of times that the answer was wrong. The summation of all the seconds spent with one specific question is its hidden Cesàro cost.
Anki-ajahns who meditate on the HCC of a card sometimes develop detachment from it, and delete it if the conventionally defined threshold of
The Yes Free Lunch Theorem (YFLT)
The HCC is not only buddhavacana, but is in fact a very advanced teaching that every Anki-ajahn should internalize. And yet, the HCC is not true in the absolute sense.
The HCC wrongly presupposes that mental energy is limited.[6] In fact, having a (low) percentage of your deck be about trivia from the latest Netflix series on Brouwer's turbulent love life, mysticism and fist fights with Hilbert is a very efficient way of learning serious stuff.
Memo-in-Sup or the Johnson-Lindenstrauss lemma
What if you want to memorize non-atomic, structured data, like 10 facts about Charles 5, or a quote? Three strategies:
Memo-in-Sup consists in using one card for more than one fact, at the price of exponentially longer (and thus less fun) questions, and a certain memorization error. For example:
Or:
Answering these questions right won't make you remember all 10 facts, nor produce the whole quote out of context. But it will help produce some part of both. A different point along fun vs error trade-off is found in the dual card where we ask about the whole thing, but accept not completely right answer. See the second OASR example for more.
Etymological aside
The L in the LJ lemma comes from the Lindenstrauss family, a prime[9] example of non-atomic data (a nuclear family), where each member had published papers listed in Mathematical Reviews. The surnames of four mathematicians could thus be memorized with the use of a single Anki card. The eponymous Johnson, Samuel, wrote:[10]
Latitudinarian uses of Anki
Having a deck with exclusively short, fun questions that you feel intrinsically curious about is not only a lot of fun, but also extremely effective. Prune the deck to keep it fun, and be strict with what "getting an answer right" means to keep it effective.
But what should you do if, as if often happens, you also half-extrinsically would like to memorize some stuff but like, don't really care super much about it and forcing you to ELI5 them would be boring? We already saw one example of this in the Memo-in-Sup section, where we traded off some fun and some error tolerance to gain some memorization. The generalization of that trade-off is called "half-assing". When considered as a problem in bounded rationality, half-assing is also called "optimally assing".
Joy Division
Since optimally assed questions are less fun than the ELI5 questions described in the previous section, at this point (but not before), it becomes necessary to divide our cards into two decks: the maximally joyful one, and the other one.
It is recommended to name the decks "1" and "2"[11]. While decluttering the names of your decks adds very little fun per se, an aspiring Anki-ajahn should be suspicious of rationalizations of why one needs names longer than one character. They are symptomatic of you not having the right attitude, which will cause a decrease in fun in other situations.
Decluttering the number of decks is even more important. Most of the thoughts the come to mind about the creation of further decks arise from attachment, and the true Anki-ajahn sees those thoughts arise with indifference, neither repressing them nor acting on them. Some exceptional legitimate uses of new decks will be discussed here and here.
An attentive reader might notice the contradiction between claiming that deck ELI5* should only contain maximally fun questions, and our having accepted a trade-off in fun in the Memo-in-Sup section. The attentive reader has indeed noticed a point of Anki doctrine where exact science is not possible, and only heuristics exist. For more information, see YFLT's converse.
Optimally-assed SR
Definition
OASR is built around the following maxim:
OASR shares some properties with Memo-in-Sup, and indeed is its dual: Both are trade-offs in memorization error / fun space, and while Memo-in-Sup preserves strictness about what "getting it right" means, and within that constrain maximize the funness for a non-atomic fact, OASR sacrifices the "getting it right" premise, and because of that is compatible with lower values of fun.
OASR has another advantage: As mentioned in the Samuel Johnson quote, it is often helpful to know where some knowledge is found. Having a vague recollection is often enough to find the card in your Anki collection[12].
If you only remember "some funny fact about grandparents", typing "grandparents" in Anki's search function will provide a bounded list of candidates to inspect, whereas Claude won't be able to be of any help. For a generalization of the search function, see LINK D.
Scheduling principles of OASR
Kinds of cards fit for OASR
The second and third kind are sometimes known as "long-form OASR". Here is one example of each:
Three reflections: These questions are long and thus boring. They should only be read in full if your first thought when starting them is "oh yeah, that was a nice quote / bunch of fact, how was it?".
This implies being very tolerant with the answers, to the point where "Easy" means "am I particularly interested in getting this question again, to possibly read it in full then?". If an answer is provided at all, basically anything should be accepted. Getting 1 fact right and 2 kind-of-right should automatically implies pressing the "Easy" button.
For some questions[14], marking the central word of the quote with a different color[15] is recommended, so answers can be evaluated (if at all) based on how close they are to that one word. "Something about horses" thus counts as right. (Saving time and keeping it fun still have non-zero importance in the OASR deck).
More Anki theorems
Rebirth in a new deck
The joy & strictness division between the ELI5* and OASR decks is essential. The doctrine of impermanence teaches that the division must be constantly updated; one should always be on the lookout for cards that now find themselves in the wrong deck.
If one gets an ELI5* answer wrong, and one notices that the level of enthusiasm is low (asking oneself automatically means it's low), but high enough that one doesn't want to delete it, the card actually has an OASR nature.
If one gets an OASR card right and gets a smile at the question/topic/whatever, the card has an ELI5* nature despite the appearances. One will need to refactor the question as an fun question, which might involve fragmenting it into several cards, using less grammar, etc. If the card is about a question, one should use a cloze question and hide one single word, no exceptions. The word should be uncommon enough that one never answer with a synonym that forces one to waste time thinking about whether the answer counts as right or wrong (if this happens with a ELI5* question, either the question was ill posed or the answer is not fit for ELI5*, and one has to rethink how to refactor the card).
The untrained mind often sees a transmigration from ELI5* to OASR as a demotion to a lower kind of life, but one should become unattached to those feelings. Also, when moving in that direction, one should not reset the card; from OASR to ELI5* one may do so.
YFLT as tertiary choice
Having gained the tools from the OASR section, we now see that the choice is not binary but tertiary: a card can be kept in ELI5*, sent to OASR, or be deleted.
The analysis also becomes more accurate with the added variable of Google-independence, which measures how difficult it is to find a certain information with a given memory of it. For example, the fact that the Significs group was founded by van Eeden is something than can be googled in five seconds; it has zero Google-independence. We mentioned the example of trying to find Charles 5 from "some funny fact about grandparents"; this has infinite Google-independence.
In ELI5*, the lower the Google-independence, the higher the threshold of enthusiasm to delete a card, so that e.g. the question "Founder of Significs group" should be deleted as soon as it stops being maximally fun. The questions about Charles, however, should be sent to OASR while there is some wish of having a vague memory of them, even if they aren't maximally fun.
In OASR there is no place for cards with low Google-independence; if the user isn't particularly interested in memorizing them, but they can be easily googled, the user can simply google them if they ever need the answer.
YFLT's converse
The Yes Free Lunch Theorem proved that maximally fun cards should be kept in ELI5* and not deleted. In the ELI5 context, we skipped asking the converse question, and in the Joy Division section it was announced that the question doesn't have an objective answer. Now we have the tools to dive deeper into the question:
A scientifically correct answer would specify the levels of boredom and Google-independence that require a question to be sent to OASR or be deleted. Such answer, however, doesn't exist, and each Anki-ajahn is forced to find heuristics of what works for them. An ajahn able to skim Memo-in-Sup questions quickly, and answer them without reading them whole, will be able to keep them as relatively-fun ELI5* cards, whereas an ajahn with a tendency to read cards whole will be forced to restrict the ELI5* deck to only proper ELI5 card, since they will otherwise kill the speed and fun of it.
(I have to publish this post 50% unfinished because I need to publish something before tonight's Inscribe deadline)
Beware! Gigabrain applications of this principle allow you to Ponzi-scheme yourself into anything.
I wanted to quote a certain short, vaguely remembered post by Caplan or Hanson on teaching by example, but I couldn't find it. I'd be thankful to any reader who remembers it.
This contradicts the "one word per answer" rule in the letter, but not in the spirit. If I am a history buff trying to memorize the exact years of a war for which I have lots of context that makes me care about it, your brain will save both numbers in one slot, and the double question is absolutely fine. If you have so little context that it feels like memorizing two random numbers, you should either create two cards or (much better) ask only about either of them, or the year in which the main fighting happened, or (even better) internalize that memorizing numbers that feel random comes from having misunderstood what kinds of things one should use Anki for. What does "number don't feel random" feel like? Being a history nerd, it took me a couple of second to understand David Bowie's song title "Aladdin Sane (1913–1938–197?)" the first time I saw it. On the other hand, not being a biology nerd, I always fail to memorize Cell Biology by the Numbers, or to be particularly good at estimation games. Any reader who feels the call to write anything making those number feel non-random will gain my eternal gratitude.
For completely unrelated reasons, I won't show any real card of mine in this post.
The predecing burn is taken almost verbatim from Socrates to Hippias (Ἱππίας ἐλάττων 369a).
The earliest reference to the quote comes 50 years after Charles 5's death, contains no horse, and refers to what we would now call Dutch. When the saying is used in Spanish, speakers mistakenly think it refers to modern German, hence my translation.
Long questions should preferably be formatted to facilitate quick reading, like in this example.
On
Also relevant to this post, thought not specifically to memorization, are the following quotes:
In order to alleviate the confusion of the skimmers, "decks 1 and 2" will be referred to as the "ELI5* [15]and OASR decks", imposing on true readers a tax of 4 extra redundant characters every time a deck is mentioned. As an exercise for home, and assuming that the trade-off is justified on utilitarian grounds, calculate the percentage of true readers vs skimmers that the author expects to exist among his audience.
This result is easily provable for finite collections, but whether it also holds for infinite ones is still an open question. For finite but large collections the result can also be proven, but the construction requires some technicalities to be discussed later. LINK C
If the reader attempts to count the number of facts in the example question, they will realize that "facts" are a fake ontological category. Insistence in using questions with fake ontological categories is prone to cause ontological crises in which the ajahn must redefine their utility function, i.e. tell themselves that the obviously boring activity they are partaking in is the way Anki is meant to be used, and that they are having fun.
This was possible in our second example (other words like "German" or "talk" could have been chosen as easily). The first example shows a question where this is not feasible or sensible, unless one has a particular interest in a specific fact and decides to accept every answer that mentions it.
Here shown as bold for technical reasons.
ELI5* is meant to suggest that the deck mostly contains ELI5 questions, but also some Memo-in-Sup ones.