Or at least, it did. The main Invidious instance is closing down, and its only developer is leaving the project forever.
But more on that later. We need to go back in time.
Terminology note: When I write "Invidious", I am referring to the open source code that anyone can run. There are several public hosted Invidious instances. If I am instead referring to the main instance, invidio.us, then I will make that clear. Only the main instance is deliberately choosing to shut down.
HookTubeBefore Invidous was HookTube. HookTube was pretty much the same idea from a user perspective, but it was not open source and it used the YouTube API instead of scraping the site. HookTube survived 18 months before it was served a shutdown notice from YouTube’s legal department at some point in 2018. The site still exists, but it now just embeds YouTube, which provides none of the desirable privacy protections. After this shutdown, the site has been altered to advertise the site owner’s various, vaguely fascist "news" sites. It’s not really worth visiting.
CloudTubeAround the time that it closed down, I was using HookTube for my own viewing, and I was very annoyed that it was gone. I quickly read up on the YouTube APIs and started making my own site. Of course mine too would be served a shutdown request as well at some point in the future, but I had a theory to hopefully avoid that: make it open source. If my site was forced to close, anybody would still be able to use my work by running their own copy. I called my site CloudTube. On July 18th, 2018, I opened CloudTube to the public and announced it on r/privacy. I had two major features up my sleeve that would make the site better than HookTube: subscriptions, and 1080p+ video.
On the comments of that post was omarroth, which was my first encounter with him. We had a conversation in PMs about his site, Invidious. Unlike CloudTube, Invidious got all of its data by scraping YouTube, which appears to make it impervious to legal requests for reasons that I don’t understand. We discussed how Invidious exposes an API to allow other services to easily extract data from YouTube, too. This is how I first heard of Invidious. Shortly afterwards, I rewote CloudTube to use the Invidious API.
Both of our sites grew, each offering slightly different features. CloudTube remained the only frontend to have 1080p+ playback. Invidious had instant subscription loading.Omar and I continued to collaborate, particularly when we worked together on the YouTube annotations archive, which ended up being a truly unbelievable success. Our relationship was a purely professional one, though. I never found out a single thing about him as a person. To this day, I have no idea how old he is, whether he has a job, or even what his time zone is.
As 2019 continued, his presence in the project discussion chatrooms dwindled until it became nonexistent, as did his work on adding new features to Invidious. I simply assumed that he was extremely busy with other things in real life, or just wanting to spend time elsewhere. After all, having a successful project does not define you as a person. (Or at least I hope it doesn’t. Ponder that.)And then it was 2020. Omar’s presence was zero. Radio silence, apart from showing up in the chatroom around once a month to answer one question and then disappear. I don’t blame him for this. I’m just documenting the history.
AnnouncementAnd then finally, August 2020. This announcement: Stepping Away From Open Source
What does this mean for the future of Invidious?
While the main instance is shutting down, Invidious is open source and other people can and do host their own instances of it. Being able to access the software should not become a problem.
DevelopmentOmar was the only developer on Invidious. This could partially be attributed to the fact that Invidious is written in Crystal, a considerably less popular language than things like Node and Python. There are Crystal developers out there, but certainly the use of Crystal will be a barrier to trying to get new developers on to the project.
In an announcement of the seizure of the domain posted Thursday by the US Justice Department, the DOJ alleged that WeLeakInfo allowed its users to access "a search engine to review and obtain the personal information illegally obtained in over 10,000 data breaches containing over 12 billion indexed records—including, for example, names, email addresses, usernames, phone numbers, and passwords for online accounts."
I have not seen a single other person in the Invidious chatroom that knows Crystal in my two years of participating there. I believe I have read every message since I joined.
Having a developer is important since YouTube updates its interface every couple of months, requiring adjustments to get scrapers to work again. Next time there is one such update, it is very possible that Invidious may become unusable, and nobody will volunteer to repair it.
To some degree, this has already begun to happen.
APIInvidious’s API is used to power other projects like CloudTube, and FreeTube (which I’d love to write more about, but perhaps not in this post.)
If Invidious dies, then so does everything that relies on it.FreeTube is undergoing a rewrite that will add alternatives to the Invidious API for all extractors. For CloudTube, I’ve talked about it a little on my blog here,, but I intend to follow this up with another post when I have more to announce about the future of CloudTube. I use CloudTube myself, and I would like to keep it running even if just for my own sake, so I hope that the future will be bright.
DataYou’re on Data Horde, a digital archiving blog. In what way does Invidious relate to digital archiving?
Well, finally some good news: not all that much. Invidious does not save videos, so apart from user settings and subscriptions, both of which are private, I do not believe there is anything stored by Invidious that we need to scramble to archive. The results of the 2019 annotations archive are safe and sound on archive.org.
This article is less about digital archiving and more about analysing the life of a service that a lot of people around the data community will miss.
The futureAlternatives to Invidious and ways to move forward from here will be discussed in the future in another post, either here on Data Horde, or on my own personal blog, https://cadence.moe/blog.
Thanks for reading.
— Cadence [they/them]