Spring Cleaning

When I moved my blogs (this one and The Turbid Plaque) back to Wordpress in September 2019, I was on the rebound. My previous two blogging platforms, a static site generator called Pelican and the flat file CMS Grav, hadn’t worked out. Neither was an inherently bad system, as far as I could tell, they just weren’t for me.

Unfortunately, WordPress had developed a new personality problem in my absence, and I soon learned that it was only going to get worse. Manifested initially as an annoying feature called the Block Editor, it’s actually part of an enormous pile of tech hubris the WordPress developers un-ironically call Gutenberg. The idea seems to be to turn WordPress into a competitor for site builders such as SquareSpace. I’m sure that pleases Automattic’s investors, but after being welcomed to the block editor for the umpteenth time, it didn’t please me.

I’m not alone. A WordPress plugin whose sole function is to disable Gutenberg has a five-star review rating and over 700,000 installations. It definitely helps, but it’s not clear for how long: Automattic has said they’re going to make Gutenberg unavoidable by the end of 2022.

I decided I needed to do a complete overhaul of the blogs, which encompassed two sub-projects: picking and learning a new blogging engine, and figuring out what to do with an archive of 550 posts stretching back 16 years. Yes, I’m old.

Picking a blog engine in 2022

My previous experience with static site generators had revealed a lot of advantages to that approach. Instead of relying on a script and database running on the server, a site generator runs on one’s local computer. It typically takes posts written in a very simple markup language called Markdown, along with images and other materials stored in local input directories, and assembles the entire site into static HTML files in an output directory. Once the HTML has been generated, it’s a simple matter of uploading it to the server. Static HTML, the original Web format, is fast, reliable, relatively archival, and nearly unhackable.

In principle, working with a site generator should be fast and easy. In practice, my experience with Pelican seven years ago wasn’t so smooth. Site layout adjustments that seemed like they should be simple turned out to be hard, and each time I wanted to post something I had to go through multiple awkward-feeling steps to do it.

Site building software and I have both grown since then. Several popular site builders are now available, including at least one with a full graphical interface that looks a lot like a traditional server-side blog engine. Meanwhile, a few years of full-time Linux use have made me much more comfortable at the command line than I used to be. Heck, I’m even using GitHub now.

After some initial testing, I settled on Hugo. Learning the Hugo layout system took a few hours of focused attention and frequent reference to the documentation, but there’s an inherent elegance to it that made the whole process fun. The application is also fast enough to watch the effects of changes in real time: edit something in the configuration file or a template, and the change appears instantly on the locally-served test site in your browser window.

As with most command-line programs, the steep part of the Hugo learning curve was right at the beginning. Now that my sites are set up, and I’ve gotten accustomed to the workflow, putting up new posts is easy. I write a draft in my favorite text editor (currently VSCode), save it to the posts directory on my desktop, then run hugo server and visit https://localhost:1313 to see how it looks. After final edits, I halt the local server and run a very simple Bash script that builds and uploads the new site in seconds.

It’s obviously still early days, and I’m sure I’ll find various annoyances as my relationship with Hugo deepens, but I’m optimistic.

Trash or treasure?

With the new blogging system in place, the next step was to figure out what to do with the archive of old posts. I’d previously left them online as static pages (thanks to Pelican), thinking that people might want to be able to find them in the future. While that was certainly true for a few posts, the vast majority got no traffic at all. Worse, even the ones that people did still visit were missing images and full of dead links.

Twitter poll.

Poll results were an ego boost.

I posted a poll to Twitter asking what I should do with these old posts, which elicited a couple of very helpful replies. Those, and the consensus of various content strategy sites I found, suggested that the hardest approach is also the best: sift through the archive manually, figure out what’s worth keeping, throw out what isn’t, and update broken links, images, and other content as necessary.

Once I buckled down and started doing it, the process was educational. A majority of my past posts were brief, of-the-moment items that outlived their utility years ago. I don’t feel bad about deleting those. Journalists’ work is traditionally short-lived, passing briefly before readers’ eyes before turning into a crab feast tablecloth or lining a bird cage. The effort of creating those short posts was small, and there’s no sense saving them forever. The handful of longer essays - things I actually thought about and crafted - seemed to deserve more care. It took a good chunk of spare time, but I’ve now gone through those, fixed the broken bits, and included them in the site move.

My excavation of the archive also revealed an interesting timeline, with post frequency varying a lot from year to year.

  • 2006: First blog post on the old domain
  • 2010: Acquired alandove.com, moved all posts there
  • 2013: Acquired turbidplaque.com, separated the science and personal blogs
  • 2011-2013: Transcribed old letters on walterandina.alandove.com
  • 2014: Posting rate plummeted
  • 2015: Switched to Pelican (January)
  • 2017: Switched to Grav
  • 2018: No posts (!?)
  • 2019: Moved back to WordPress (September)
  • 2020: Posting increased, mainly due to reviews and the weekly Filtrate series
  • 2021: Posting rate declined again after weekly Filtrate clogged

I think the low post counts in 2014 and 2018 stemmed from struggles with the blogging platforms.

The other thing I learned was the value of evergreen posts. “How to” articles, reviews, and deeper essays are relatively easy to keep up to date. Those pieces are much harder to create than offhand comments and link summaries, but after sifting through hundreds of posts to pull out a handful worth keeping, I see how the effort pays off. I’ll still post some short, frivolous pieces, of course - I can’t stop being the wisecracking class clown - but they may vanish from the site once the novelty wears off.

Now if you’ll excuse me, I’m going to see about straightening up a few things in the basement.