I Tracked Everything I Read on the Internet for a Year

Since early July 2021, I've been keeping a record of every article I read on the internet, which at the time of writing, is just a few months over a year's worth or articles.¹ Every article I read, I put on the list - and as it turns out, I read a fair amount. I thought I'd take a moment to do some stats, talk about how the list works and why I do it.

Some stats

At the time of writing, there are 1024 articles in my list. On average, I've read ~2.4 articles in each of the 421 days since I started the list. The majority of those articles come from the Hacker News digest email I get every morning or the long list of RSS feeds I subscribe to², but I also find articles in a smattering of other places, like Reddit or a news app of some sort.

There's nothing interesting about the graph of the number of articles I've read over time. It's a straight line.

If you look closely, you can see where I hit my "personal best" of 15 articles in a day on 2022-04-06. My calendar says that day in particular was during a holiday, so I can only imagine the 15 articles was the result of a few lazy hours in bed one morning.

What's a little more interesting is the data about the sites I was actually reading.

So far, I've read pages on 596 unique websites.³ I only read articles on 116, or 19.46%, of those websites more than once.

The top three most-read websites in my list are: 1. the BBC (www.bbc.co.uk), 79 articles (!) 2. Xe Iaso's blog (xeiaso.net), 26 articles 3. Ars Technica (arstechnica.com), 24 articles

Some other notable mentions are rachelbythebay.com (9th, 9 articles), simonwillison.net (14th, 7 articles), and jvns.ca (16th, 6 articles).

I wish there was an easy way to filter by "independent websites", but there isn't. There's quite a lot of commercial domains on the top end of my most-visited list, I'd like to see a ranking of my most visited independent websites.

How it works behind the scenes

Somehow, I have to add an article to my reading list. The easiest way I came up with to do this was with a JavaScript bookmarklet.

It's pretty simple - it gathers the URL, page title and meta description and image, then modifies window.location.href to redirect me to an endpoint on this website. That endpoint processes the input data then redirects me back to the original article. If you're interested, the source code can be found here.

It's done this way because some websites restrict what HTTP requests can be made with JavaScript on their site. I originally tried to use the Fetch (or was it XMLHttpRequest?) API to save the data, but quickly ran into the a dead-end and pivoted to using the current redirect-based method.

Since I use Firefox on both the desktop and my Android phone, my bookmarklet auto-syncs between my devices. It's fairly easy to add to my reading list on my phone, which is pretty handy as that's where I do a large portion of my reading.

The endpoint

All the data for my reading list is stored in a big CSV file in a Git repository. It used to be that the endpoint on my website would clone the repository, make a new commit and push the changes, but now all of that is handled completely within GitHub Actions. I wrote about making this change in my last post.

The static site

The static site that displays the list is generated using a small static site generator I wrote in Golang and deployed on GitHub Pages. When a new item is added to the list, the generator automatically runs and deploys using the magic of someone else's computer.

The static site uses a variation of this website's stylesheet, just with tweaked colours. Otherwise, everything is identical.

If you've so much as looked at the reading list site, you'd probably agree that it's a bit of a disaster. It's a single huge page with every entry in it, making it difficult to navigate and near impossible to search. I'd like to fix this - at some point, I'm going to enhance the reading list to split the content over a few pages and attempt to add a client-side search, but that's a project for the future.

But... why do all of this?

Honestly? I don't remember why I started doing this in the first place, but I certainly like having it around.

Often I'll find myself wanting to share an article I read a few weeks ago with someone, and with my reading list, I can do that with a quick Ctrl+F. It's really handy, and for that reason alone, I'm going to keep doing it. There's not really another, practical use for it.

That said, having a list of all the articles I've read and being able to fiddle around with that data whenever I want to very much satisfies my inner nerd.

I guess I keep up with my reading list, in part, because I enjoy it.

1. May it be known that I had every intention to write this article earlier than I eventually did. ↝
2. These are also served to me in email form. I get an email every morning which is generated by Walrss, an AGPL-licensed, Docker-ready app that you can self-host if you really want to. ↝
3. The graph of unique domains over time is also an uninteresting, near-straight line ↝

This article is referenced elsewhere on the internet:

• Hacker News
• Automating a Reading List :: zanshin.net