The Probability of Success With SRS
Within the confines of the how-to-learn internet, spaced repetition sits atop a pedestal of best practices and is lauded as the key to learning efficiently. But there are those who are less convinced that it deserves to be so highly regarded. Sometimes I wonder what a direct face-off between those factions might look like:
🔔 Ding-Ding!! 🥊
For-Faction points out that spaced repetition is scientifically sound, says everyone should be using it! Against-Faction deals a powerful blow by pointing out organizing flashcards takes too much @#$%ing time! For-Faction rebounds, shakes it off and comes back with 4.0 averages and national placement in standardized testing! But Against-Faction ducks this blow, weaves, and points out – WHAM! – testing is not the end-all-and-be-all of learning and education…!
For some it feels like spaced repetition is indisputably the correct way to go about strengthening memory and learning material, but there is another frame that hints at a different picture: For-Faction typically walks away with language learners, medical students, and others that are taken with its arguments, whereas Against-Faction is composed of countless others who find it more hassle than its supposed benefits are worth. Due to spaced repetition’s fame and regard as a great learning innovation of our time, it’s worth noting who walks away a devotee and who strikes it from their list of must-have tools.
What Is: A Reframe 🐝
You pay an up-front, perhaps regularly occurring, tax on your time and effort in exchange for the increased likelihood that you will have target knowledge on-hand, at a later time, in long-term memory.
What does this mean? Sometimes it’s as simple as a boolean (true/false) value: Do you know it or not? When you speak to someone in a second language will you be able to remember the words you need to convey your message? Ensuring your vocabulary is in long-term memory is crucial for the skill of speaking a second language.
But there is more to it than just plain old booleans.
“Time and Effort” 🐝
Sometimes the gains are less obvious. What if you aren’t studying natural language vocabulary, but a programming language’s syntax? Many will quickly retort that this is a waste of time. Indeed, why memorize information when you can instantly look it up? But there’s a rub: “Nearly” instantly. All those little lookups take a few seconds or even a minute and that time adds up. The question then becomes, how much time will performing those lookups take as compared to paying a spaced-repetition-time-tax?
Since looking up information itself is a sort of recall, you could make an argument against spaced repetition if you’re only dabbling in the language or doing a quick weekend side project.1 Time spent practicing flashcards could rival the entire time spent on the project itself and that’s kind of silly. But if you’ve got a programming language on your resume and plan to work with it for the forseeable future, paying a few up-front minutes memorizing all the fundamentals will save you time later on in many-a-moment of actual problem-solving. This also frees up your mental cycles to focus on the problem at hand instead of on the forgotten details.
Spaced repetition allows you to re-schedule the time and effort necessary to solve a problem or perform a skill, so that you’ll have more time and effort at the precise moment of solution creation or skill execution.
Imagine how differently we would approach studying programming languages if we felt the same amount of embarrassment forgetting syntax as we feel at forgetting words while speaking to other people.
“Target Knowledge” 🐝
Of course, there is more to this memory game than mere time and effort. There is also the question of content: What is it exactly do you need to remember? This is a much trickier question than it seems at first glance, not least of which because we can’t know everything.
Let’s look at language learning again: How do you know what words to study? At first, it’s really easy. “The words in my textbook.” Ok, ok, but how did the vocabulary lists in those textbooks come to be? Vocab lists and word lists are generally created based on a bunch of facts and statistics about words that describe how likely you are to encounter them.
This can be expanded to pretty much any domain to answer the question, “What do I need to know?” Well, “How likely am I going to need to know it?”
“Increased Likelihood” 🐝
Remember, spaced repetition gives you the ability to intentionally increase the likelihood that you’ll remember or know some fact at a later time. Let’s combine this with the likelihood we will encounter some fact in the future:
With this vantage point it’s easy to see that our goal is to increase the size of the intersection between the likelihood of encounter with the likelihood of remembering. This can become silly, of course. Someone who memorizes all the presidents might then go to parties and intentionally steer the conversation in directions that will allow them to show off their presidential knowledge – “I came across that thing I knew, whoopee!” But we all know that is not an organic representation of the real likelihood of enounter.
Let’s revisit word lists: They are typically built from statistics such as word frequency and topic relevancy. From the data that describes words, vocabulary curators can build lists that will realistically help language learners in the real world. Knowing a bit about where you want to go can inform you on what you need to know.
It is possible, and I think likely, that the dominance of spaced repetition in medicine, language learning, and test prep is due to the fact that the sheer number of items in the I-know-I’m-going-to-need-this-category is so vast. “In order to do X you must memorize 1000 Y” as opposed to, say, “In order to do X you might need to know 5 Y or 10 X or 2 Z…Not sure which though!” The time saved in churning through dozens, hundreds, or thousands of flashcards is worth the up-front cost in time and effort because you have to have so much memorized in order to do anything.
As for other domains, spaced repetition likely falls short because the likelihood that you know now what you will need to know in the future falls off a cliff. Life is a vastand spaced repetition only helps insofar as you have a high level of confidence that the target material is going to be necessary to know, or at least be somewhat useful, down the line.
Even in language learning one could make the arguement that the more you learn the less spaced repetition becomes helpful. At some point you are probably better off just reading, watching, or listening to more content, which have additional benefits such as providing the natural context of those words you’re memorizing3.
And The Winner Is…? 🐝
I don’t think there is much debate about whether spaced repetition is effective at drilling facts into your brain. What’s more, it is an innovation that makes your memory deliberate. When you are deliberate, you give yourself the ability to steer. When you can steer, you can aim. And if you can aim, you can reach your goals. You cannot reliably reach what you cannot aim at and there is something to be said for that.
But I also think with great innovation should come great consideration: As others have said, spaced repetition a hack. Our minds naturally evolved to accrue knowledge in the manner of forgetting curves. So, it might be worth considering the consequences of bending this knowledge acquisition process to our perceived needs.4
After all, life is complicated. How do you really know that the set of facts you learned yesterday are the set of facts you should know tomorrow? At the start of a journey, do you really know the exact coordinates of each and every step will be?
Perhaps a good rule of thumb for spaced repetition is this: Anything you need to know to get “up and running” you should drill’n’memorize, but leave the rest up to nature’s algorithm.
Although, one could argue that it’s still worth it. Just make sure the spaced repetition time tax is a reasonably small proportion of the allotted project time. ↩
These venn diagrams could be broken down further. They are limited to these two categories for simplicity’s sake. ↩
Clozes can provide context, but only a limited context compared to an essay or book. ↩
At the very least, it probably raises the likelihood of suffering from “theory induced blindness”. ↩