Not only has this idea been around for a while as he says at the beginning, it has been implemented a couple times over already. And not just the theoretical version he did that doesn't work because of compression, but variants that try to squeeze out the maximum, add error correction, use audio too etc.

a few ones pop right up when you search things like "using YouTube as storage" but thanks to YouTube search being useless garbage nowadays I can't find the particular one I'm thinking of. It was rather long but really went into the technical details and challenges.

I found this three year old explainer video: https://www.youtube.com/watch?v=tu2O-iHL16Q

You mean this one?

https://github.com/DvorakDwarf/Infinite-Storage-Glitch

Better fork with more functionality: https://github.com/KristianAskk/Infinite-Storage-Glitch

Is there any YouTube search engine that works like old youtube?

All fun and games until Google bans your account for ToS violation.

Seriously, if you actually have important stuff in your Google account I would warn against doing this. You’ll lose all of it more than likely.

You can use steganography, at the cost of (much?) lower bandwidth. You can apply it to one of those “10 hours of x” videos. And of course, use a dedicated Google account.

I wonder if there'd be a stenographic way of hiding the data in a normal looking video so you wouldn't get banned for beach of TOS

Of course there is but the people doing this likely aren’t going to advertise their methods as that’ll get them undue attention.

I remember ArVid was invented in pre CD-ROM era. You could store 2gb of data on VHS videotape. https://en.wikipedia.org/wiki/ArVid

That reminds me of VBS -Video Backup System for Amiga. I probably still have it somewhere in a drawer.

http://hugolyppens.com/VBS.html

The old Discrete Flame/Fire/Inferno systems would store backups of its sessions to tape. At the head of the back up was black and white static that was the data for the project. It would then lay down as actual video all of the mattes used in the comps as well as the actual video clips used. I only ever saw this on D1 tapes. Not sure if that's just because that's what was used, or if the uncompressed vs compressed format of DigiBeta could be used as well. Good lord, how I don't miss the bad old days of early digital computer video formats and the nightmare of pushing the boundaries to make it happen. All of those hard drives clickety-clackety pushed to their limits to record SD!!!! Now, I have an external bus powered SSD in a shell slightly larger than a cigarette lighter that can do 8K resolution video. Early 90s vs 2020

Why not use the audio channels too. Is that too small a bandwidth in comparison to video?

If the video is relatively static ... I am assuming compressed video would be very light-weight even for long videos.

audio is 3, maybe even 4 orders of magnitude less bandwidth.

2 reasons it's not practical.

Storage is cheap, but compute isn't. Encoding is lot of compute and losing data by abusing TOS not worth the risk.

Eh, I consider computing a pretty minimal expense for long-term storage that I always throw my shit through zopfli before stuffing them on my NAS or uploading them to S3.

Is there any limit on the subtitles/captions you can upload on a video? That might be a more reliable storage method.

There's basic sanity checking on subtitles. The main one is that they have a very slow maximum throughput, so you can encode far less info in subs than the video itself.

And how many different languages can you upload too? Base64 encode all your subs?

Actual code (C++): https://github.com/Brendan-Kirtlan/Video-Encode

Obligatory Tom 7's "Harder Drive: Hard drives we didn't want or need": https://www.youtube.com/watch?v=JcJSW7Rprio

I just use reddit. Make your own subreddit, encode whatever in to Base64 / Base16

Save as a post / comment and then run a scraper to obtain data.

Don’t forget to ensure it’s on archive.org.

This has been around since at least 2015. I'm not sure if there were other projects demonstrating it at the time, but here's the implementation that I wrote back then: https://github.com/Valkryst/Schillsaver

In this day and age, it's much easier to use DockerHub or similar to store an unlimited amount of data.

This should be exhibit 1 why trying to make a business depending on developers as customers is probably a bad idea, given that online storage is cheap and easy.

this ai-speakers are so boring, that I can't follow the video.

Did anyone state otherwise?

I posted a similar video yesterday. This seems to be doing the rounds.

https://news.ycombinator.com/item?id=38319393