Rethinking My Website and Blog#

How we read and write shapes how we think

Twitter chaos made us rethink our engagement on the web. Personally, I felt smarter and happier pre-Twitter. During this period I engaged by reading and writing blogs. I drifted away from my practice of blogging, but I’m intentionally turning back to that direction.

With this in mind I rebuilt my personal website and blog. This article discusses the reasoning behind the new design.

I had a few objectives in this transition:

  • Optimize for slightly longer attention spans

  • Mix evergreen and timely content

  • Flatten organizational hierarchy

  • Experiment with automation to integrate with social media

Motivation#

Why I Stopped Blogging#

I used to blog a lot more, but I stopped for a few reasons:

  1. My Twitter follower count increased, encouraging me to engage directly there

  2. I grew multiple blogging venues (personal blog, community project blog, company blog) which had longer review periods and delays, which discouraged writing

    I find that I write 10x more if publication is instantaneous

  3. My blogging infrastructure was embarrassingly out of date

    this was ancient, didn’t have https, metadata for social sharing, good style, etc..

But I miss it for a few reasons

Longer Attention Spans#

I write short posts to a short attention span. However, I think that the 280 character limit of Twitter goes too far. Twitter encourages punchy communication devoid of depth or complex thinking and forces lots of skimming. I feel that my brain is largely inactive when on the web. I’m pushing against this.

I’d like to author more thoughtful content. I’d like to encourage readers towards deeper thought. Blogging helps.

Evergreen content#

I frequently point colleagues to posts I wrote years ago. Here are a few:

And also some more recent posts I’ve made:

I value these posts 100x more than any Tweet because I get continuous value from them. I’ve shared these manually with hundreds of people, and they get plenty of traffic from strangers.

New Blog Implementation#

I have a new website and blog! 🎉

Source lives at mrocklin/mrocklin.github.io and is hosted at https://matthewrocklin.com.

There are a few choices I made. Let’s go through them and why.

Basics: Sphinx and Markdown#

I am most familiar with Markdown to author documents and Sphinx to render them. I stole my baseline implementation from Chris Holdgraf. This uses various technologies like the PyData Sphinx Theme, MyST and ablog, which I’ll let Chris describe.

Blog posts (timely) v. Docs (evergreen)#

Looking at my old blog there are two different categories of articles:

  • Timely: One-off content describing a recent experiment or thought that I wanted to share right away

    Examples:

  • Evergreen: Deeper thoughts where I’m trying to communicate with / teach colleagues

    Examples:

Both categories are important to how I work, but are typically served differently.

  • Timely/ephemeral content is well served by blog infrastructure with chronological ordering, tags, and RSS feeds.

  • Evergreen/educational content is well served by a layout more similar to the documentation of a software project.

I want both. I have both.

If you’re reading this on my website you’ll see a sidebar to the left, similar to what you would find in the documentation for a software project. This sidebar lays out the evergreen content. New posts don’t go here automatically, I have to place them in the sidebar intentionally. This whole site could be seen as “documentation to working with matt”.

At the same time, every post does have appropriate blog metadata to be included in the top-level articles/ page, which lists things chronologically, as well as the Atom/RSS feed. It’s like a documentation site where every new page gets an RSS entry.

Atom/RSS Feed#

I miss Google Reader. This technology encouraged people to read and share long-form articles. It made the content and sharing primary while commenting/liking was an after thought.

I don’t use an RSS feed reader today, but I’d like to start again. I know that others do and so I’m resurrecting an Atom/RSS feed is at https://matthewrocklin.com/atom.xml.

Flat directory structure#

One of the things I disliked about my Jekyll blog (and lots of static site generator default configurations) was the focus on chronology and dates in titles and URLs. URLs looked like this:

  • https://matthewrocklin.com/blog/work/2022/07/19/scipy-mission

I then shaped this site to be in categories with URLs like the following:

  • https://matthewrocklin.com/engineering/scipy-mission

I don’t care about the dates of things that I read. They just get in the way. But then I realized that I didn’t care about categories either, and so now everything is just flat.

As a reader you probably don’t care (you got here by clicking a link) but flat structures appeal to me personally.

Automated Social Sharing#

I’ve hooked up IFTTT to Twitter and LinkedIn. I haven’t found a good Mastodon solution yet.

The results here aren’t as good as I would do for Twitter, but it does mean I don’t have to log in. Also, I never used to engage at all on LinkedIn, and so this will help greatly over there.

Final Thoughts#

This is an experiment. It may fail! That’s ok!

People should experiment more with how they communicate. I encourage you to give your own communication habits some consideration.

Think for yourself   How I Conduct an Interview