🚨 NEW VIDEO DROP
A look at different scales of simulation studies and the programming needed to do them
📺 What’s coming up?
Bootstrap; A look into survival analysis; why is statistics hard?
📰 TL;DR
Rereading is useful for picking up knowledge you weren’t ready to receive in the past.
This issue was inspired by an intro from a great 3Brown1Blue video. The particular quote that struck me was this:
I realize second editions are much less of a thing for YouTube videos than they are for books, but part of my motivation here is that there's been more research since that original video that's very much worth discussing.
The key phrase here is “second editions”, or as I thought of them: revisiting old material that you’ve already covered.
I used to think that once you read or consumed something, that was it. You’ve already extracted all the value from it, and you’re free to move on to new material. It’s “inefficient” to revisit things.
But there is immense value in revisiting things you’ve read or seen before, and here’s why:
Your past self simply not have been in the right position to most benefit from a particular piece of information.
I started really internalizing this view the first two years of my Ph.D. For those unfamiliar, the first 1-2 years of a Ph.D is taking courses and enduring qualifying exams. Studying for these often boils down to memorizing concepts and doing the same problem sets over and over again as practice.
Whenever I looked at a problem for the first time, I often did not understand what the problem was even asking for. There are probably several math terms I don’t understand. So I go back to the textbook or notes to learn those terms.
With the terms in mind, I return to the problem and realize that I don’t know what theorem I’m supposed to use. Back to the textbook.
After several iterations of this process, I have (probably) stumbled into a solution. A solution that is probably wrong in some places, but will slowly be refined with consecutive revisits.
What this process highlights is the development of a network:
The nodes represent key ideas, concepts and theorems. The edges represent how they are connected. As you revisit material, you open up more opportunities to add important nodes and edges in this network.
From my perspective, it’s very unlikely to instantly jump to the right panel. More efficient people will just go through this process faster. But no matter what, you don’t get to the right panel without repeated exposure or thought.
Louis Pasteur is credited with saying “Fortune favors the prepared mind”. And I posit the way you get a prepared mind is to revisit things to learn and hone.
So I regularly revisit audiobooks and videos that I’ve seen before. More often than not, there’s a quote that just hits in a different way because I’m now in a better place to acquire that knowledge.
An aside from the language learning world
There’s a version of this in the world of language learning. It’s called the “input hypothesis,” made famous by linguist Stephen Krashen.
The input hypothesis states that people progress in their knowledge of the language when they comprehend language input that is slightly more advanced than their current level. In other words, you learn by seeing slightly harder words/structures/grammar. To make it explicit, once you learn something, it become “easy”; what is “hard” to you is constantly changing and in flux.
A consequence of the input hypothesis is that people interested in language learning should dedicate most of their time getting input. There will surely be stuff you don’t know, but the repetition will slowly develop your intuition for the language. What’s extreme for language learning is the sheer number of repetitions needed to get a good grasp of the language.
Applications to YouTube
What does all this have to do with YouTube? (For me at least.)
For one thing, it opens up the range of content. I used to be afraid of repeating myself in videos. But it’s just a fact that most people who encounter the channel will not see the video where you mentioned a specific fact. It’s okay to have the same idea concept in multiple videos. There’s probably a line there, but there’s a good chance that you’ll be able to get more people to hear the ideas that you want to communicate to them. I have no idea what point in the learning process people are at; the only thing I can hope for is to be at the right place at the right time.
Second, it helps me rectify an early mistake I’ve made with my videos. I have a few videos that contains a dense amount of ideas. That being said, having a single title capture all these ideas is hard, so I’ve essentially made it impossible for potential viewers to benefit from seeing these ideas.
The most prominent example I’ve noticed is in my video on the “Nobel Prize” of statistics. The main topic of this video is this prize, but it also goes into a brief explainer on all the ideas that have won the award: Cox regression, bootstrap, mixed models, and theories supporting MLE. I’ve had to point commenters to this video multiple times because they want info on these specific topics, not the award that they won. So I’m letting myself revisit these topics, when in the past I would have just continued pointing to a URL.
That was a long one. Thanks if you made it this far, see you in the next one.
Christian
Footnotes
🧐 What am I enjoying right now?
Severance, Season 2 and Singles Inferno, Season 4. Something I’m not enjoying is applying for jobs, but it’s all that’s left to do now besides YouTube.
📦 My other stuff
I wrote guided solutions to problems to Andrew Gelman’s Bayesian Data Analysis. It’s for advanced self-learners teaching themselves Bayesian statistics
You can support me on Ko-fi! YouTube and Substack are by far the best (and easiest) ways to support me, but if you feel like going the extra mile, this would be the place. It is always appreciated!
Just wanted to say, as a fellow statistics YouTuber, I love reading your substack updates - always interesting to hear from someone on a similar journey (especially someone who is in the earlier stages and not yet at a 3b1b/Statquest level of fame). Hope you keep it up.
Thanks for this. It's really nice this stood out to you and you're looking to apply it to your own learning and your videos too. It's an underrated idea and thanks for sharing it as I'll be using it in my own learning now.
And I love your videos, man.