Technological Overthinking #5 | Finding a new read-it-later app

This is a follow-up from my previous post about the read-it-later app Omnivore, and why its existence and shutdown shows off the good and bad of open-source projects. Diving into the world of self-hosting software isn’t for me yet, which means it’s time to search for a replacement hosted service. After reading a variety of lists and reviews of recommended read-it-later apps, and learning from what others in the Discord found to be useful, I decided that writing down a checklist of what I was looking for in a replacement service would probably be the best way to test the options.

What I needed a replacement read-it-later app to have/ not have.

• A priority on web browser features/settings (i.e. that its not using the “app-first with a limited web version” approach)
• A Firefox extension that would let me easily save links/pages.
• An Android app that reliably saves links from web browsing and other apps e.g. Discord messages.
• A paid hosted syncing service.
• A focus on reading text, not on audio.
• No sponsored content or advertising mixed in with my saved links.

What I would prefer the replacement to have/not have:

• No AI features (or at least the ability to fully disable and ignore any AI features)
• Preference for a yearly subscription over a monthly subscription.
• Payment with money rather than data.
• Preference for developers using open source software and/or a nonprofit structure.

With the checklist in place, I started trying out some of the recommended alternatives to see which might be the best fit.

Apps/services I tried:

Readwise Reader (free trial)

Reader was already in the “probably not” pile as it aims to be a much larger program than I need. It aspires to be a read-it-later app, a note-taking app, and a full productivity system that integrates with other systems. For me, that would probably lead me to turn leisure reading into work, which would not be a good idea.

While I was impressed at Reader’s polished and coherent design, I did find that its everything-app nature felt like “too much” as a replacement; for me it felt intended to be a system that needed active maintenance rather than a tool. I have nothing against systems (case in point: this blog series!), but I didn’t want to replace something that had been simple with something much more complex.

The dealbreaker for me was that Reader automatically summarised imported articles with their AI “ghostreader”. This meant that when looking at my list of articles/feed items, I wasn’t actually seeing the start of the article as written by the author, but the AI summary instead. Worse, there doesn’t seem to be a way to disable it. The fact I couldn’t turn the AI off instantly sent Reader to the “hell, no” pile. A read-it-later app that wants to “ghostread” things for me is not where I want to be!

InfoFlow

This app was a project made by a member of the Omnivore Discord, who then announced and expedited it as a result of the shutdown. While InfoFlow has promise, it is very much in beta; I was met with frequent errors when attempting to save articles to the Android app or even when trying to open the app. I will check it out again in future, but at the moment I would be frustrated in trying to use it as a main reading app

EDIT: The InfoFlow dev announced on 2nd Nov that they would also shut down on 17th Nov. It feels like the developer jumped in over their head by annoucing InfoFlow in a beta state, and by wanting to satisfy every feature request and be everything to everyone, an opinion that seems to be backed up by the Discord community reactions.

EDIT 2: Never mind, the InfoFlow dev updated on 17th Nov, their planned shutdown day, to say that they won’t shut down after all, and will keep going sustainably by relying on each user storing articles in their own cloud storage provider. The open-source rollercoaster strikes again!

Obsidian

I never expected this to be the winning choice as, similarly to Reader, it is built to do far more than what I need. Obsidian is a note-taking and knowledge management app that users often describe as being a “second brain”. While the Obsidian software is not open-source , its approach makes up for this: the software is a customisable wrapper over local files that remain fully under your control. This removes all concerns about notes being “locked in” to a proprietary system, and is overall a really interesting way of doing things.

I was also curious about seeing what Obsidian felt like to use because of the hugely positive comments I’ve seen on Reddit about its features, customisability, and the effect its had on people that use it. (Though users have also reported going much too far down the rabbit-hole and spending inordinate amounts of time customising their systems rather than using them.)

While my brief dabble with Obsidian demonstrated that it was indeed overkill for what I needed, I now better understand why its fan-base is so dedicated. If I had been looking for a storage centre for my entire brain, Obsidian would have won.

Raindrop.io

I tested out Raindrop as I saw that a few different people online and in the Omnivore Discord had chosen it as their preferred alternative.

Raindrop.io feels like a hybrid between a regular bookmark manager and a cloud storage provider. Its aim seems to be to provide a place where users can keep and arrange everything they care about online cohesively, whether its articles, ebooks, PDFs, videos, or highlights from other sites. It also emphasises integrations with other services e.g. syncing RSS feeds, Pocket bookmarks, and even Spotify playlists and Reddit saves. It also connects to services like IFTTT so that users can create more complex chains and can automate saving items such as Liked YouTube videos.

The website shows Raindrop as being very visually-focused, and it is described as “designed for creatives”, so I initially felt like I wasn’t the target audience. There are more text-friendly display options in the settings, however, and I was able to make my test list of saved bookmarks look quite similar to my saved list in Omnivore. This was where I found the dealbreaking problem – Raindrop is not actually a read-it-later service. It doesn’t have a reading view, nor a function to mark items as “read” or “unread”. So I was correct that I am not the target audience, albeit for the wrong reason!

I also found that because Raindrop is more focused on sharing and syncing live collections, it does not have any offline reading or viewing functions. I believe it is unlikely to ever have that option, as it seems like an anti-goal for the service.

It’s a shame that Raindrop is not what I believed it might be from the descriptions I had read, as I was very impressed with Raindrop in other aspects. Firstly, the UI is so polished that I was surprised to find that Raindrop is mostly developed and maintained by a single person. Raindrop has been in development since 2013, and its clear that the developer is aiming for long-term stability and consistency rather than startup-style growth.

Secondly, the free tier is free forever, and does not set any limits on the number of devices, bookmarks, or collections. The main limitation that free users can only upload 100Mb per month of files. The premium features are mixture of expected options – such as full-text search and a 10Gb per month upload limit – and unexpected safeguards such as daily backups and a tool to identify broken links. This tool even has customisable parameters for how thoroughly Raindop should scan a bookmark before declaring that it is broken. While there are AI options in the paid-for tier, they are described as being private to each user and not used for training, at least.

Overall, while Raindrop seems to be excellent at what it aims to do, it just isn’t what I was looking for.

Others

From what I can tell, Pocket has changed very little from when I stopped using it in 2022, so I didn’t feel like I needed to revisit it. Pocket was one of the first well-known read-it-later apps, and for the last few years it has been owned by Mozilla and built-in to Firefox, which has result in both more attention (positive and negative). Pocket has a more social focus than some other read-it-later apps, as it includes ways to follow other users and recommend links to them. ( I never used that aspect of Pocket so can’t comment on its performance, but it has its fanbase.)

Instapaper seemed pretty much interchangeable with Pocket in terms of features, so I also didn’t explore it this time. One thing in its favour, similarly to Raindrop, is stability. Instapaper was founded in 2008 and has been steadily worked on by the same team since 2013, with features being added slowly but consistently. If you are looking for a service that is unlikely to fold, be sold off, or be drastically changed, Instapaper may be the best option.

I have previously trialled a few RSS readers, including Feedly, fraidycat, and Inoreader, as I like the idea of using RSS feeds. The issue I had with all of the RSS readers I tried is that I wanted to be able to remove items from RSS feeds so that my list wasn’t overwhemingly busy. The ideal situation for me would have been that for every fetched item, I could choose either “yes, i want to read this, send it to my read-it-later queue” or “not for me, please permanently delete it from my fetched list”. That sort of two-stage feed just wasn’t available in any individual app I tried.

What I decided on:

The winning service for me ended up being Wallabag. Wallabag is a dedicated read-it-later service that is intended to be self-hosted, but users can also pay for hosted syncing. The hosted sync service (wallabag.it) is run by a member of the Wallabag development team, which for me felt like an extra guarantee that it would work smoothly and reliably.

When checking against my wishlist of criteria, Wallabag meets every one.

My Wishlist	Wallabag
A priority on web browser features/settings (i.e. that its not using the “app-first with a limited web version” approach)	Check. The web browser is the most full featured access point.
A Firefox extension that would let me easily save links/pages.	Check
An Android app that reliably saves links from web browsing and from other apps e.g. Discord.	Mostly check: my phone fails to save a link about 5% of the time, but does so in the same way, and with the same error messages, that I used to experience with the Omnivore app. So this one seems like my phone’s fault.
Paid hosted syncing	Check. Hosted syncing is just 11€ per year, or 30€ if you choose the “support Wallabag” option
A focus on reading text, not on audio.	Check
No sponsored content or advertising mixed in with my saved links.	Check
No AI features	Check
Preference for a yearly subscription (or one-time purchase over a monthly subscription	Check- yearly payment is available
Payment with money rather than data.	Check.
Preference for developers/ using open source software and/or a nonprofit structure.	Check.

I was also surprised to find that Wallabag has the two-stage RSS feature I had previously looked for and given up on. In Wallabag’s web interface, you can choose for RSS feeds to either automatically become part of your read-it-later queue, or to stay in a separate Feeds section where you can manually delete or save each item. You can even choose an option for each feed you are suscribed to. After choosing to save or delete an item, it is not then re-fetched when you refresh the feed. This is exactly what I had wanted to find before, and having the option for either behaviour here is an excellent choice by the developers.

So far, the only feature from Omnivore that I miss, but not enough for it to cause a problem, is the newsletter handling. Each user had a within-Omnivore email address which could be used to sign up to newletters and have them automatically delivered to their Omnivore queue instead of their email inbox. I don’t have a replacement for this feature, so I might just go back to having them arrive to a dedicated folder in my inbox.

Bonus feature:

As I was first writing up these notes a few weeks ago, Wallabag announced that they had created an Omnivore importer. The system that the Wallabag team have developed isn’t optimal because it is time-consuming for people who used Omnivore heavily; however, this is partly Omnivore’s fault. Exporting data from Omnivore produces a JSON file for every 20 saved articles, and the Wallabag importer requires uploading each JSON file separately. As I have just over 12,000 articles in my Omnivore library, importing everything will require uploading 600 files one at a time. So far I’ve done about 2,500 articles and, barring some link rot, they have arrived.

The option to mark incoming articles as having already been read also didn’t seem to work. But still, this is far better than not having the option at all! And to be done in just a few days, for free, is just another example of the community-focus and sharing spirit that I talked about before.

Overall, now that I have chosen a definite winner, I can consider this question successfully overthought.