It’s probably not surprising that I’m often particular about which apps and programs I use. While most of the time apps can be relatively interchangeable, I have three favourite programs/apps. For me, that’s defined as programs/apps that a) I have chosen as the sole way for me to do a specific task after a lot of research, and b) I would be genuinely upset about not being able to use any more.

The apps/services are MusicBee (where I manage and back up my music and podcast library and sync it to my phone and mp3 players), Omnivore (where I save articles from my phone, laptop, or desktop to read later and archive), and Listenbrainz (an open-source listen tracker).

Of the three, MusicBee would be the biggest loss. It’s also the main reason I still own a Windows PC. Finding anything that could equal MusicBee would require intensive research, but its local-first nature means that I could continue to use my installed copy even if the program was no longer actively maintained, making it hard to truly lose access to.

Omnivore, on the other hand, is a service rather than a program. It is also now a dead service: on November 30th, the hosted service was shut off and all user data was deleted.

Until Oct 29th, Omnivore was the main way that I kept track of the articles and links I wanted to read and archived the articles I had read. I used it daily and had a library of over 12,000 items. (This combines imported previously-read links, RSS feed items, newsletters, and the links I’ve saved since switching to Omnivore.) For context, I even lurked in the Omnivore Discord server; the only time I’ve ever joined a social medium for a piece of software!

The announcement of Omnivore’s impending shutdown was frustrating partly because members of the Discord had been discussing payment options for months beforehand. The hosted Omnivore service was 100% free to use, and there were no obvious avenues to donate to the project. People wanted to give the devs money, and to pay for access to hosted Omnivore (or premium features such as the high-quality text-to-speech) to ensure longer-term stability, but the devs didn’t take anyone up on that, which I found confusing.

The announcement email on Oct 29th was a little annoying because of its surprise nature and its buried lede. The email opened with “we’re joining forces with ElevenLabs” which sounded promising, but just two paragraphs later was this sentence: “Next, all Omnivore users will be able to export their information from the service through November 15 2024, after which all information will be deleted.” This did not fit with the optimistic tone of the email!

ElevenLabs’ venture-capital-funded nature and focus on AI text-to-speech instead of reading means that many Omnivore users will not move over to their product, ElevenReader, on principle, while many others could not move over to ElevenReader as its feature set is much smaller than Omnivore’s. (Both of these were true for me).

Fortunately, there was one bit of good news in the rest of the message: “ElevenLabs is committed to the developer community and the Omnivore codebase will remain 100% open-source for all users.” This means that Omnivore does technically still exist, in that the code exists. People with the right equipment and skills to host the code on their own server can just about run a  more limited version of Omnivore at home. But for anyone who isn’t technical enough to run their own server, or who would rather their read-it-later app was a tool to use rather than a project to work on, is out of luck.

As a result of spending time looking for a replacement service, and of keeping an eye on how the Omnivore Discord responded to the shutdown, I’ve been thinking about how Omnivore is a strong example of both the good and bad aspects of open-source independent software.

The bad:
The dev team were running the hosted Omnivore service at a loss, because of the costs of storing data and fetching RSS feeds etc. Something had to fold, and that something would be either the zero price tag, the ad-free and tracker-free experience, or the hosted service itself.

Researching a replacement, especially when looking at the length of my list of needs and preferences, helped me remember me why most people stick with the default options and don’t get caught up in the details of getting exactly the right option for every choice.  

Finally, and most importantly, it reminded me that preferring to use open-source, independent software always comes with uncertainty. As someone who is more technical than the average person, but less technical than the average open-source enthusiast, adopting a new tool always comes with the question of how long it will be before I’m forced to jump ship and choose again. How long might it be until the tool is closed down, or bought by a company with very different values, or locked-down and turned into something user-hostile? It creates a feeling of always waiting for something to go wrong, and it makes me wary of recommending any tools or programs to regular people who are used to too-big-to-fail apps and tools backed by the world’s largest companies.

The good:

When the Omnivore shutdown was announced, the response from both the Discord and wider tech communities demonstrated the people-focused inventiveness and willingness to help that marks the best of the FOSS world.

Within hours, people on the Omnivore Discord were advising each other on replacement services, researching tools and services, and even writing tools to help others import their data and articles into alternate sites and services. Members are also adapting the code to remove outside dependencies so that the service can be self-hosted. (When I described the self-hosted version as “more limited” earlier, that was not an insult, just a symbol of how new and how in-progress the hard work currently is).

The lead developer was also very responsive and helpful. He spent time fixing any bugs that prevented people from exporting their data, and then extended the shutdown date from November 15th to November 30th to ensure everyone had time to export their data. Given that many of the people leaving Omnivore were not moving to ElevenReader, he could have said “take it or leave it”, but instead he made sure everything ended as smoothly as possible, which is a very good-natured move.

Other read-it-later apps (both open-source and proprietary) also joined in, pulling extra resources into improving, or even creating, importer tools so that people could safely move their Omnivore data into a larger variety of apps. Obviously some of this is marketing-savviness rather than sheer goodwill, but it worked, and it was a good example of putting the people affected by a change first.

Overall, while it is annoying that such a useful and functional service is now gone, I think the good side of this equation wins out. It’s hard to be completely pessimistic after watching others respond to frustrating news by dedicating their time and attention to making and improving tools that make others’ lives easier.