The Sequences on YouTube
tl;dr: Do you have any thoughts/advice/warnings/ideas for someone who wants to turn the Sequences into a series of YouTube videos?
Naturally, this post also exists as a video.
I will be:
Turning all 50 Sequence Highlights posts into 6-15 minute videos
Hopefully 1-2 a week
Just a guy talking into a camera, no fancy effects yet
I’m looking for:
Someone who can edit videos. I can do it myself, it will only be slightly more complicated. (I’m running on a Chromebook.) I’ll be searching for interim solutions and I’m not sure what I’ll end up settling on, but the more options, the better.
Ideas. This community is supposed to be good at arriving at good idea through First Principles regardless of whether they end up looking weird or not. If you want to help, think about this for 5 minutes, by the clock.
I have:
Zero experience in video.
(But I do theater?)
(And I’m motivated.)
A Blue Yeti mic
I’m doing this because:
YouTube has a much larger audience potential than this website currently has. More people learning about the Sequences is good.
People who come here and haven’t read the Sequences may do more damage than good. Only, not everyone is interested in sitting down and reading 8 hours of blog posts before being a-okayed to speak.
This project has been tried before using Minecraft railcarts and a robot voice, but it didn’t curate many views. This is not the same project. There are plenty of longform high-views videos on YouTube (eg. video essays or RealEngineering videos) and this wouldn’t even be longform. Short attention spans are not the perceived problem here.
Ideas that might work:
The Sequences are more than a decade old, and Eliezer and others have noted how they are flawed and outdated in certain specific ways. So:
I might change the writing a little if better for video. The writing in itself isn’t sacred to me, I’m just here to facilitate the transmission of ideas.
Of course, I am pretty foolish and should be running any changes by the LW audience before doing anything drastic. (And Eliezer, if he deems this project worthy of some attention.)
Having LW users write short afterwords to videos might be a good idea to address some problems and concrete applications of lessons without editing the original text. I could also plug CFAR.
It’d be nice to be able to ship new viewers into a space for discussion, but linking to LW might be a bad idea for garden maintenance. Discussion seems good to me, and if it’s not an LW post specifically bred for discussion written up for every video, then perhaps a Discord channel? Seems suboptimal.
I’m reaching out to the Rational Animations email to ask them if they would be willing to draw thumbnails for the videos. If that doesn’t work, I’ll get a Midjourney/Dall-E subscription. Thumbnails are really really important on YouTube.
I’ll be using the LW fontfor footnotes and title text on video. Currently there are no planned special effects or anything, just me speaking into a camera and some text appearing from time to time for footnotes, table of contents, and emphasis. If you have any idea for cool (and time-cheap) special effects, hit me with them.
There will be a general principle of experimentation here. If you suggest an idea to me, I will do it. God knows there’s a lot of posts in the Sequences.
I’m very pro-sponsorships. If I ever gain credibility enough, I can do EA org adverts for free. The plan is to get people interested who are not yet in the rationalitysphere, so these sponsorships would actually be useful.
Warning:
The first few videos will necessarily be terrible, especially, hopefully, by the standards of the 47th video.
So… just… keep that in mind.
Video will be filmed on a Pixel 6 phone, which should be fine given that means 1080p quality. People on YouTube care more about the quality of the audio than the visuals anyway, and the audio will be perfect.
That’s all! Keep in mind that this is highly experimental: I really hope the channel can sustain some growth, but I don’t have great calibration for predictions like this (*ahem* my Manifold profit is currently negative).
I’m competent in basic video editing and am willing to help you out. Contrary to some of the other comments, I think that there’s substantial value in converting the sequences as-is to YouTube videos considering the size of the platform. Plenty of people use YouTube who don’t read blogs or listen to podcasts. You could take the project and run with it, constructing animations, diagrams, and visualizations. But even absent those potential enhancements, I support disseminating the message as widely as possible.
Yep, “floating head” videos as they’re called are actually quite popular. They are also by far the easiest way to start a channel and starting a channel is better than not having one at all.
I would also like to add animations, diagrams and visualizations, and find some other ways to make content much more engaging. Up until now most of my efforts were directed toward having the courage to publish this in the first place. Now, I can move on to actually getting down to the logistics!
Given that a podcast already exists, I think you might get more bang for your buck if you did some animation on top of it. Otherwise, the only thing you are adding is putting it on youtube and having a camera of your face. This would probably be (much) harder, but also probably much higher reward if it worked.
Maybe a collaboration with rationalanimations would help? Not really sure, but good luck if you try to do this!
Yeah, it’s really important to avoid having it look like a lecture. You could also try learning animation yourself and seeing if you’re decent at it (or even really good at it), that would make it easier to collaborate with rationalanimations/@Writer since they’re bottlenecked (there’s also tons of great Yud and Scott Alexander stuff, and rationalanimations has been growing/sped up lately but even 5 videos per quarter won’t get nearly enough of them animated in time).
There’s also galaxy-brained high-risk high-reward ideas like @dkirmani’s attention-based “zoomer-friendly” format of the sequences. The Sequences are great material to work with in general because all you have to do is something that Yud didn’t think of.
What I recommend is reading the CFAR handbook or tuning cognitive strategies and having this be one of the real-world challenges you test yourself against as you read and grow and test yourself, just like with the Sequences except you’ve already read them. Sequences-based optimization ideas are probably well worth pursuing on the margin, so don’t get intimidated, but the final winning idea makes you and many others win big so it’s worth it to iterate and try things and fail many times and learn many lessons until something incredible happens.
It’s a lot of work to learn to create animations and then do them for hours of content. Creating AI images with Dall-E 3, Midjourney v6, or SDXL and then animating them with RunwayML (which in my testing worked better than Pika or Stable Video Diffusion) could be an intermediate step. The quality is already high enough for AI images, but not for video without multiple tries (it should get a lot better in 2024).
For ayone trying to keep up with AI for film making, I recommend the youtube channel curious refuge https://www.youtube.com/channel/UClnFtyUEaxQOCd1s5NKYGFA
Where is the podcast?
And is anybody aware of any other videos of the Sequences? Might make sense to start with those missing.
Here’s the podcast (should be on any podcast app): https://podcasts.apple.com/us/podcast/rationality-from-ai-to-zombies/id1299826696
Suggestion: do them out of order.