How I Develop My #AcademicRunPlaylist (updated 12/2024)

Lots of people have asked me how I find so many talks for my #AcademicRunPlaylist, and until now I’ve only given abbreviated answers. Here I’ll explain my methodology, which hopefully some folks find useful.

To start I’ll add a caveat to this whole thing - I have no idea if this is the best way to do things or even how effective it is. All I can say is subjectively in my roughly 2.5 years of doing this it has provided me with a plethora of talks that have dramatically expanded my horizons, given me great ideas that have significantly changed my work, and connected me with some of the most fascinating scholars in the world.

I started with a list of topics I’m interested in. Initially this was a small list - AI, organizational behavior, management, networks, etc. Over time it expanded to a sizable number of adjacent fields (which honestly I’ve found to be more useful than my core interests) - psychology, economics, law, bioethics, anthropology, cognitive science, robotics, etc. I added new topics once I had nearly exhausted talks to listen to in disciplines already in my playlist.

Academic Primacy

As soon as I identify a topic, I look at lists of top academic departments in that field globally. I search for the department channel first, and then if it doesn’t have one I’ll go up to the school/university level. YouTube used to have a “channels” tab where universities would typically list affiliated YouTube channels that often had names that won’t come up if you only search for, say, “MIT,” and that was immensely helpful for me. Unfortunately they got rid of it in November 2023, hopefully it returns at some point.

Nearly 100% of the channels I subscribe to are at academic institutions - a few are think tanks, but I have very few private company subscriptions. This primarily boils down to issues of bias and objectivity. Academics are also biased of course, but their work is much more public and they generally care a lot less if a particular approach doesn’t work than someone in industry. 

I’ll still listen to panels if there are folks in industry or government on the panel, but it’s very hard to interrogate if what they’re saying is true. If someone at a tech company says their LLM can do X, to what degree was that influenced by internal corporate prerogatives? IMO management is even worse. If a CEO says that they did Y and it’s the reason they were successful, I’ve learned almost nothing. Perhaps I could generate a hypothesis from that, but nothing more. Reasonable people can disagree about these points, this is just my approach.

I’m also conscious of the fact that the departments in global rankings are woefully under representative of the general population. To get a more global perspective, I’ve also gone through rankings of top schools in the developing world. At first I focused on only English-speaking institutions, but I periodically check the channels of top institutions in countries like Brazil and Mexico that occasionally have English talks. I’m also fluent in Japanese, and so I subscribe to many of the top institutions there.

Importantly, I only subscribe to a channel if over the course of a year they have at least 3 talks from separate events that I would listen to. The reason is that there are just too many channels otherwise and I’d be constantly scrolling through uploads every day - this way I focus more on channels that consistently upload longer talks. I only listen to talks that are at least 15 minutes long, since if it’s shorter than that research suggests you can’t really focus on the talk itself and you’ll learn significantly less.

For technical topics like AI or robotics, I won’t add talks that are more than a year old. My thinking here is that this research ages much more rapidly than, say, psychology, and so if I spend my time watching an older video in the AI space it’s probably significantly less useful than me watching a video on another topic.

I do subscribe to a few podcasts, but they tend to be less about the research itself and so I’m very picky about subscribing to those.

Chance Encounters

Sometimes when scrolling through Mastodon I’ll see interesting talk announcements, and when I do I immediately search on YouTube for the hosting lab and the speaker. I look for talks over 20 minutes long (it’s a YouTube filter), and then sometimes I’ll discover interesting labs that I never knew existed. Whenever I find a channel like this, I go through the last year of videos like I mentioned above and if I find enough good ones I’ll subscribe to the channel.

Technical Questions

I’m often asked how I make time to listen to all of these talks. A big part of the reason is that I run a lot. During the pandemic I started to go for longer and longer runs, and typically once a week I’ll go on a run that lasts at least 7 hours and I’ll normally get in at least 3 runs of 4 hours. I found early on that I didn’t like listening to music during runs, so I started listening to talks. 

If you’re going to listen to talks on YouTube, you pretty much have to pay for YouTube premium since it’s the only way you can download videos or listen to them in the background. On the plus side you don’t get any ads.

To set up the playlist itself, whenever I find an interesting talk I add it to my Watch Later playlist. At night I'll download talks for the following day, and I tend to prioritize recent talks and variety - I generally don't only listen to talks from a single channel in one day, with the exception of the times when I listen to long events like NBER meetings. I probably spend about 5 minutes a day going through my subscribed channels that have uploaded new videos checking if any of them are interesting. For every talk I listen to I also save it to a public playlist so others can subscribe to it - I only started doing this in December 2022, but you can find all of the talks on YouTube that I've listened to here.

As an aside, listening to talks as regularly as I do will absolutely eviscerate your YouTube recommendations. I'm assuming this is because so few people do this, but at this point YouTube thinks that I'm partial to very long videos that have at most a few hundred views and static camera angles, so I'm often recommended town meeting videos, religious services, and random academic lectures (occasionally these are interesting TBF).

If you’re going to listen to talks on runs I also can’t recommend the Bose Tempo sunglasses highly enough. You can get them with prescription lenses, and they’re basically the same price as nice sunglasses anyway but they connect to your phone and play sound for around 8 hours at high volumes. Your ears stay clear so you can still listen to the environment, which I like. If I’m inside I’ll either use earbuds or Bose Soprano glasses with clear frames.

Sharing the talks on social media has helped me in a few ways. First it’s a great refresher at the end of the day for me to review what I’ve listened to, which likely helps with learning. It’s also nice because when I tag people they often reach out, and I’ve started a number of collaborations through that mechanism. Lastly it seems like some people also find these talks interesting, and by filtering them first (I don’t post talks I don’t like) hopefully I’m able to surface talks that will be useful.

If folks have other tips on this I’m all ears!

Update 12/2024: Audiobooks

Earlier this year I finally ran through the last ~10 years of talks that are online, so I've started adding audiobooks as well. The list of topics I've hit is similar to above, although my process is necessarily a bit different for identifying books to read/listen to. The rules of the road are also similar to my academic talk methodology - only books by academics (for the same reasons).

As far as identifying specific books, there are a few methods I use. If I listened to a talk by someone that I enjoyed and they plug a book during the talk, I add it to my queue. Sometimes in passing someone during a talk will also cite a book on an interesting topic, and after checking that the author is an academic I'll add it to the queue. Similarly I'll occasionally see book recommendations from folks I'm connected to on various social media channels.

To expand my reading pool I'll google course syllabi on different topics and add required reading by academics to my list. Essentially I query "topic x course syllabus required readings," which works pretty well. Most public universities also have the required readings for their courses publicly searchable, so I've been slowly working my way through those as well.

Every time I read a book I'll add at least one book to my list. As far as the order I listen to them in, I prioritize audiobooks but I've used eBooks as well (more on that in a bit).

On the technical front, I've got five apps that I use that are helpful here: Libby, Audible, Kindle, Adobe Reader, and Alexa. By far the best is Libby (probably only works in the US), which if you log in with your library card gives you free access to a huge number of eBooks and audiobooks. I always check if the book I'm looking for is available here first (obviously, because it's free). Next I'll check Audible, which I have a subscription to. They also have a bunch of free audiobooks, and a ton of the books that haven't been available on Libby are free on Audible. If Audible doesn't have it, I'll look on the Kindle store. For public domain books I'll get the PDF.

The eBooks are a bit tricky for to listen to. If you use Libby to rent one you have to use the "Read with Kindle" option, which then allows you to add the book to your library just like a regular Kindle book. With any Kindle book in your library, you can open the Alexa app and type in "Read book X to me from my Kindle library" and it will start reading it from the beginning. Sometimes you have to type the full title and author for it to work, and while theoretically you can pause and resume the read to me function sometimes it starts reading from the beginning again. You can try to enter the exact chapter, but that sometimes doesn't work and you have to use the Next key to get back to where you were. I think Next does the next page or next section, it might depend on the book. Unfortunately it also seems like Alexa requires internet access, so if you're running in an area with bad connectivity you'll get cut off (when this happened to me and I reconnected it started the book from the beginning). Some books don't have read to me enabled, so you've got to read them the old fashioned way.

Finally, for books where you have a text-based PDF Adobe Reader does a serviceable job. There's a read aloud function, but be warned it also reads things like the page number so it's really a last resort.