I have taken the survey.
tanagrabeast
Seven Years of Spaced Repetition Software in the Classroom
A Year of Spaced Repetition Software in the Classroom
A Second Year of Spaced Repetition Software in the Classroom
Travel Through Time to Increase Your Effectiveness
Did you get IRB approval for these human studies on children?
I’m not sure which is more absurd: the IRB approval process or the very idea of high school. I’ve often asked people to consider a thought experiment where everyone on Earth suddenly forgets that our educational system as we know it ever existed. Would we really reinvent it just like it is now? Hearing how it worked, would we scream in terror and cancel anyone who had taken part? (Status quo bias much?)When I was studying stand-up comedy, I actually developed a bit in which I play-acted a researcher proposing high school to an ethics board. It went like this:
RESEARCHER: “I was thinking we could stick 35 sleep-deprived teenagers in a room for an hour and expose them to academic stimuli. After that, we’ll do some tests on them.”
BOARD: “I see. Tell me more about your subjects.”
RESEARCHER: “Well, they’re minors, obviously.”
BOARD: “Okay…”
RESEARCHER: “And most of them will be enrolled against their will.”
BOARD: “And how long will you need them?”
RESEARCHER: “6 sessions a day for four years.”
BOARD: “Wait, hold on. Sample size? How many kids are we talking about, here?”
RESEARCHER: “All of them.”
BOARD: (mutterings among themselves) “Well, it sounds like everything is in order...”
Are you familiar with Direct Instruction, which is reminiscent of the Mennonite school?
Someone (probably on LW) pointed me to Direct Instruction a few years back, so yes, I’m acquainted with it. Because of the emphasis on staying fully reviewed on all relevant prior knowledge, I saw it as having obvious promise for technical subjects like math, in the hands of the right teacher. I was less convinced it made a good fit elsewhere, perceiving (perhaps unfairly—I didn’t dig too deeply) some big negative trade-offs:
Like with my whole-class Anki, it seems heavily reliant on the teacher’s high-energy snake-charmer charisma. This makes it difficult to sustain for much of a class period and demands a great deal from a teacher who tries to do it all day long, day after day. This also makes it difficult to broadly among teachers with different personalities.
It sounds brittle with regards to roster variance. Specifically, it seems pretty insistent on having everyone in the room up to speed. With careful tracking/grouping of students, this can be achieved, but in practice, kids move in to your school part way through the year and aren’t on the same page. Or you only have the one or two teachers for that grade level math, so the slowest kids are in the same boat as the sharpest. I would think that one or two stragglers would grind the class to a halt, and that this would be statistically inevitable in larger classes. (I don’t know if this makes DI math worse than the status quo, where plenty of students are fall behind and get lost, but with less fanfare and hold-up for everyone else.)
Have you ever tried SRS for muscle memory?
No. I’m not seeing how that would work, or how that would be relevant to what I do, but I’m certainly curious. Do you have examples?
Why does every employer ask for a list of references, then not call them?
You think that’s bad?
A local school district called the “references” of a prospective employee for a tough-to-fill position. These references, her former bosses, uniformly advised against hiring this person.
The district hired her anyway.
After a long and difficult process, the district eventually fired this employee. Then, the principal that had done most of the legwork on the firing got a call from yet another district. Surprised to have been listed as a reference, the principal vociferously cautioned the new district against hiring her.
They hired her anyway.
That’s a fair question with a surprisingly complicated answer.
Before I get to it, let me explain that I had a weekly ritual for reminding them about the potential for personal Anki use and the value I placed on it. Every Monday morning started with a written reminder on the board and an oral reminder from me that the class deck had been updated, uploaded, and was now ready for them to download. I invited them to pull out their phones and download the latest deck at that time, and gave them a couple of minutes to do it. Depending on the season, I would also remind them that “Anki is school EZ MODE”, and “Anki is our study guide” for the mid-term, final, etc.
Of course, way more phones came out than were ever downloading my deck, but I felt the cost was acceptable.
This brings up another limitation of Anki: the fact that it costs 25$ on iOS. From a survey I gave the year before, I know that more than half of my students use iPhones. I made sure they knew they could be using it for free on their desktops, linking their desktop profiles to Ankiweb, and studying for free from their phone’s web browser… but that smells like effort. Effort AND cost? They rejected it out of principle, offended by the suggestion.
Anyway, I didn’t directly poll my students because I felt like I already knew the answer from the downloads and didn’t feel it was worth the risk of lowering the prestige of Anki further. You see, by about 6 weeks in there were maybe 1 or 2 students per class still downloading. If I asked the class to “raise your hand if you’re using Anki on your own” those 1 or 2 people would have felt like losers, and maybe not even raised their hands. The same effect would have been only moderately reduced with a written survey, because the first thing students do after a survey is ask their peers what they wrote. Either way, they would have gotten the impression that “nobody is using Anki on their own”. You know how studies show that teens think their peers are having way more sex than they really are? I chose to let my students think their peers were getting way more Anki. Never underestimate the power of social proof on teenagers.
Another fun dilemma: It occurred to me that whole-class Anki use was probably cannibalizing personal Anki use. I suspect that many of the ~25 people who downloaded my deck during the first two weeks soon stopped because it felt redundant. I was left to choose between two scenarios:
Scenario 1: We use Anki together in class, and about 110 of my ~180 students get something out of it, but only 3 people get the most of it by using it on their own.
Scenario 2: We don’t use Anki in class. 25 students (optimistically) get the most out of Anki. Nobody else gets anything from it.
Even if I felt like total learning was higher in Scenario 2, I’m strongly incentivized to choose option 1. I know who those 25 students would be. They were destined to pass the state test and get As in my class with or without Anki.
Guess which scenario looks better on my annual performance review?
As a counterpoint, let me offer my own experience rediscovering cryonics through Eliezer.
Originally, I hadn’t seen the point. Like most people, I assumed cryonauts dreamed that one day someone would simply thaw them out, cure whatever killed them, and restart their heart with shock paddles or something. Even the most rudimentary understanding of or experience with biology and freezing temperatures made this idea patently absurd.
It wasn’t until I discovered Eliezer’s writings circa 2001 or so that I was able to see connections between high shock-level concepts like uploading, nanotech, and superintelligence. I reasoned that a successful outcome of cryonics is not likely to come through direct biological revival, but rather through atomically precise scanning, super-powerful computational reconstruction, and reinstantiation as an upload or in a replacement body.
The upshot of this reasoning is that for cryonics to have any chance of success, a future must be assured in which these technologies would be safely brought to bear on such problems. I continue to have trouble imagining such a future existing if the friendly AI problem is not solved before it is too late. As friendly AI seems unlikely to be solved without careful, deliberate research (which very few people are doing), investing in cryonics without also investing in friendly AI research feels pointless.
In those early years, I could afford to make donations to SIAI (now MIRI), but could not afford a cryonics plan, and certainly could not afford both. As I saw it, I was young. I could afford to wait on the cryonics, but would have the most impact on the future by donating to SIAI immediately. So I did.
That’s the effect Eliezer’s cryonics activism had on me.
This looks like a valuable book, and as a teacher I will probably read it soon. That said, at the high school level, it often feels like we are already swimming in “best practices” but being pushed under by crushing workloads. Better practices generally mean higher loads—sometimes not in the long term, but always in the short term.
Think of it this way. When you have 180 students per day, anything you do that relates to individuals gets multiplied by 180. Did you design a killer rubric that lets you read and give useful feedback on a submitted paragraph in just one minute? You’re amazing, but you will still need three solid hours to go through them all. And remember that this is on top of all of your other lesson planning and parent communication and extracurriculars and meetings and administrative paperwork etc etc.
And you have school again tomorrow.
In the same dangerous motion of not quitting after my first year, I privately swore to doggedly accumulate true effectiveness without sacrificing my personal life on the altar of public education. In the eyes of many, this makes me a bad person. How dare I draw boundaries around teaching as though it were just a job? Six years later, though, the tortoise is clearly winning this race. The corpses of the hares smolder by the side of the road; they were never as fast as they looked.
I will no doubt find some useful gems in this book, but they will be vastly outnumbered by the tears I shed for all of the great techniques I won’t see any realistic way to implement.
Thank you for writing this review.
I don’t give nearly as many tests and quizzes as you might think, as they are costlier than commonly appreciated. Not only do I have to write them and score them, I have to dedicate precious, precious class time to them. Lots of class time, because we have to wait for everyone to finish, and in my classes some students will always take forever. More than that, even, because if I give anything that smells like formal assessment, I’m required to send students with accommodations to a special testing room where they can get extra help and time. So not only do I have to work out the logistics on that, some of my most needy students might be missing from my class for an extra day.
My testing minimalism raises some eyebrows. I get away with it because my students consistently beat expectations on the tests people care about the most.
You’re absolutely right, though, about assessments acting as an additional review. I do give some regular “take-home quizzes” in the form of vocab/terminology matching sheets or crosswords just to have something to put in the grade book.
Also, that team-based review game we play on the interactive whiteboard has very high overlap with my Anki deck. This did help keep some vocab words that were languishing on the bottom of the Anki deck in circulation. Not every class has the same leeches and due cards on a given day, though, so it’s not really practical to tweak the game that precisely.
By the way, gwern, let me thank you directly for your web page about spaced repetition. This experiment never would have happened without it. My “heartfelt presentation” is basically the CliffsNotes version of your research, and I attribute you at the end of it.
Donated $105, making my contribution the true baseball bat in the infamous $110 question.
May we get these things right more often.
As an American teacher of high school English, with a passion for spaced repetition software, I feel like it is my duty to respond to this post. My answer may surprise you.
If your goals are simply to understand more of what you read and to write more effectively, trying to skill up your general English skills strikes me as rather suboptimal.
Sure, a mastery of common word fragments will improve your ability to make at least some sense of unfamiliar words that use them—I certainly teach these—but you probably already know the most useful ones. I’m also unconvinced that etymology deepens comprehension much; usually, we want to understand someone, not somewords; this comes from understanding what that person intended to communicate, not from unlocking obscure arcana behind the words they happened to use.
Most of what is known to help reading comprehension is language independent, as is most of what is known to help you write better. I certainly don’t think Paul Graham’s skill as an essayist has much to do with his English; if he knows a second language even marginally well, I’m sure he would write in it nearly as effectively. To wit, he eschews esoteric explication. Writing is a craft, not a lookup table.
The strongest predictor of how well someone will do on a comprehension test of a given passage is how much they already know about the topic of that passage. A knowledge of the domain-specific vocabulary for that topic is either the second strongest predictor, or the same thing, depending on who you ask. General purpose vocabulary is farther down the list, and as an educated native speaker, you, again, are unlikely to find much low-hanging fruit in that area. So rather than take another level in English, I would suggest you consider which domains you want to be able to understand more of, and just start reading more in those domains, looking up words as needed. The language you do it in is almost irrelevant.
Consider: in the 6 credit hours of theory and practice for teachers of English Language Learners my state requires all teachers to take, I was taught that teenagers acquiring English as their second language are best off when they can continue learning domain specific concepts in their native language while waiting for their English to mature enough to transfer this knowledge over. Otherwise, they gain conversational English fluency but miss out on their first, best chance to learn foundational abstract concepts in, say, Science, Math, or Social Studies, leaving them without the ability to talk or even think about these subjects in any language.
With all the above in mind, when it comes to Anki cards and vocabulary, I am convinced that a great example sentence is much more useful than a great denotative definition. Connotations matter, and a visualizable, narratable context goes far both in conveying the extra implications of a word and in providing hooks for one’s memory. Still, you’re unlikely to absorb the deep flavor of the word—the full intent of one who wields it fluently—without encountering the word many times in varied contexts.
I say this in part because I acquired a sizable Spanish vocabulary from a time living in Spain decades ago, and there are to this day a number of words common to my internal monologue that are Spanish simply because they capture the flavor of the concept more perfectly than my closest English equivalents. But this is only the case for words that I encountered on enough authentic occasions to build that full connotative sense. Ones I merely studied out of the dictionary never reached that level, no matter how well I mastered them from a recognition and recall standpoint.
As any programmer will tell you, leveling skills in one language will have knock-on effects on your abilities in other languages, whether they are similar or not; the similar ones give you skills that transfer very directly, while the dissimilar ones broaden your conceptual toolset for approaching programs in general. If a problem might be more tractable within the intricacies of language suited to it, by all means, go deep into that language. But if you’re trying to understand say, an algorithm or a data structure, study that.
At the smaller end of the spectrum, I’m using this as an opportunity to zero out several gift cards with awkward remaining balances—my digital loose change.
Or at least, I started to. There’s a trivial inconvenience in that the donation button doesn’t let you specify increments smaller than a dollar. So I wouldn’t actually be able to zero them out.
This is, of course, a silly obstacle I have more than compensated for with a round donation from my main account. The inconvenience will just have to be borne by those poor retail cashiers with “split transaction” phobia.
Greetings.
I’m a long-time singularitarian and (intermediate) rationalist looking be a part of the conversation again. By day I am an English teacher in a suburban American high school. My students have been known to Google me. Rather than self-censor I am using a pseudonym so that I will feel free to share my (anonymized) experiences as a rationalist high school teacher.
I internet-know a number of you in this community from early years of the Singularity Institute. I fleetingly met at a few in person once, perhaps. I used to write on singularity-related issues, and was a proud “sniper” of the SL4 mailing list for a time. For the last 6-7 years I’ve mostly dropped off the radar by letting “life” issues consume me, though I have continued to follow the work of the key actors from afar with interest. I allow myself some pride for any small positive impact I might have once had during a time of great leverage for donors and activists, while recognizing that far too much remains undone. (If you would like to confirm your suspicions of my identity, I would love to hear from you with a PM. I just don’t want Google searches of my real name pulling up my LW activity.)
High school teaching has been a taxing path, along with parenting, and it has been all too easy to use these as excuses to neglect my more challenging (yet rewarding) interests. I let my inaction and guilt reinforce each other until I woke up one day, read HPMoR, and realized I had long-ago regressed into an NPC.
Screw that.
Other background tidbits: I’m one of those atheist ex-mormons that seem so plentiful on this page (since 2000ish). I’m a self-taught “hedge coder” who has successfully used inelegant-but-effective programming in the service of my career. I feel effective in public education, which is not without its rewards. But on some important levels teaching at an American public high school is also a bit like working security at Azkaban, and I’m not sure how many more years I’ll be able to keep my patronus going.
I’ve been using GTD methodologies for the last eight years or so, which has been great for letting me keep my mind clear to work on important tasks at hand; however, my dearest personal goals (which involve writing, both fiction and non) live among some powerful Ugh Fields. If I had been reading LW more closely, I probably would’ve discovered the Pomodoro method a lot sooner. This is helping.
My thanks to all who share their insights and experiences on this forum.
I expanded MIRI’s pool of quality candidates for their office manager position by submitting my application.
If you can see yourself stepping into that role, please do likewise!
Experts talking shop with other experts is one of my favorite finds when I study!
During my dive into stand-up comedy, I came across this video of some top comedians talking shop. Especially from about the 30 minute mark, when they seem less concerned with entertaining their audience, they get into some juicy minutiae of why a joke might work or not. It really expanded my thinking on the subject.
Are such chats more insightful than an expert teacher would be in a lesson on that same topic? Not necessarily. But you might not find a skilled teacher ever teaching a lesson on that exact topic. I think humans are naturally primed to closely observe expert-expert chats for a few reasons:
• Social proof. We instinctively want to be able to talk like the experts do so we can blend in with them. So we listen carefully to how they talk.
• Authenticity. If this is what experts actually talk about, we feel like it must really matter. It’s not just the lesson of the day.
• The overhearing effect. This is a term I’m making up, but I’ve found it to be an important one exploited by storytellers. We naturally want to deduce the context of overheard language, so we listen extra carefully, trying to fill in the blanks. I suspect this is down to humans’ highly evolved appetite for gossip. The fact that the experts aren’t talking to us is essential for exploiting this effect.
Although… I find that an expert talking to himself, seemingly unguarded, seemingly without conscious awareness that he is being overhead… can also trigger the overhearing effect. When I model a skill to my students, I try to verbalize my inner monologue in a way that will be intriguing to overhear and carry that essential whiff of authenticity.
I’m not sure what expert self-talk looks like in foreign language instruction, but I would be interested to find out. (Any ideas?)
But from my time becoming a reasonably fluent Spanish speaker (since lost), I can describe a few language dimensions I found interesting but neglected by all but the nerdiest supplemental books.
Sentence-level inflection patterns vary, and it helps to be aware of them. For instance, the musicality of typical question sentence is different in American English than in Castilian Spanish. If you can pick up on the melody earlier in the sentence, you can better contextualize what is being
saidasked.The way speakers in different languages produce what seems, on the surface, to be identical phonemes, can be quite different, and understanding this is essential to actually sounding like a native. There can be hours of fun trying to practice a Castilian ‘toh’ sound (as in toma), with its thicker top-front palette tongue contact, vs. the American English cousin equivalent (as in tomato).
Native speakers of language A learning language B often end up predictably adopting many of the same idioms and juicy words from language B into their language A conversations with each other, and they find themselves saying or thinking in those patterns even when their brains are mostly running language A. It could be fun to introduce some of these to novices and make it part of the language A classroom slang—a kind of introduction to thinking in language B.
My questions are mostly about the player side, and about how deeply the DM should model the player:
Should the player be assumed to be implicitly collaborating towards a coherent, meaningful narrative, as is necessary for a long-lived TTRPG? Or might they be the type of player you often find in AI Dungeon who tries to murder and/or have sex with everything in sight?
Should players ever try to steer the story in a genre-breaking direction, like erotica or murder-hobo power fantasy? Should DMs resist these efforts or play along? If the latter, should the DM go a step further to actively intuit what this particular player would like to see happen?
Should players provide input that might be more sweeping than usable in narrative? (e.g. Take over the world!) If so, on what sort of level should the DM respond to these?
Should players be assumed to be ready to end the narrative at the ~1,000-step point?
How many of these people want to die today?
Precious few I expect. Their daily rituals must still carry some intrinsic satisfaction. Perhaps they no longer hold long-term goals because they don’t feel like they have enough time left to achieve them and enjoy their fruits. This does not seem unreasonable, though it may seem self-defeating from the outside.
As I’ve recently commented, I don’t like the idea of living each day as though it might be your last, but if I were 80 years old it might make a certain kind of sense. At the very least, this late-game logic creates a sizable hurdle to getting an elderly person interested in something to the point where they become less apathetic about eventually kicking the bucket—which is all we’re really talking about here.
As a high school teacher, I use this tactic all the time. I have to, or I would be overwhelmed by the many requests from parents that seem perfectly reasonable from their perspective but which become mathematically impossible in the aggregate.
“I think each teacher should check my son’s agenda every day and sign off on whether they did their classwork and whether they have homework.”
“Of course. Not a problem. As long as he brings it to me at the end of every class period filled out and ready for my signature, this should not be an issue.”
Three days later—often less—the practice discontinues with no word from anyone.
Another example, by email: “I would like to meet with you this week about my daughter’s grade.”
I deliberately wait between 4 and 24 hours. And then:
“Of course. I’m available every day after school until...”
9 times out of 10 I’ll never hear from the parent again. Ever. It’s easy to rattle off an email to a teacher when you’re mad at your kid, and it’s easy to let a teacher make an appointment for you, but the trivial inconvenience of deciding on and committing to your own appointment time, combined with the cool-off period I created before responding, almost always leaves the ball dead in their court. And I think they feel too silly about it all afterwards to even talk to me again.
Oh well. Guess it wasn’t that important to you.
Yeah, this is a dark art. Selective application is key. I really am there to help. But I use judicious social engineering to filter many of the demands I could end up committed to. Hopefully, I’m letting the ones through where I can actually do some good.