Luxury of simplicity
An evergreen blog topic is “writing my own blogging engine because the ones out there are too complicated.” With the risk of stating the obvious:
Writing a blog engine, with one customer, yourself, is the most luxuriously simple web application possible. Complexity lies in:
- Diversity of use-cases: applications that need to work on multiple devices, different languages, work with screen readers, might need to work offline, or on a particular network.
- The real world. Everything about the real world is complicated: time, names, geography, everything. Governments can’t simplify this very much, so they build extremely complicated technology so they can serve every citizen (in theory). Companies can “define their customers” and simplify this a bit. Individuals can simplify this a lot: just me, my timezone, my language.
- The problem area. Something like how Microsoft Word decides cursor position, or Excel calculates formulas - there are actually hard problems out there, with real complexity to solving them, with no simple solution.
Which is all to say, when I read some rant about how React or Svelte or XYZ is complicated and then I see the author builds marketing websites or blogs or is a Java programmer who tinkers with websites but hates it – it all stinks of narrow-mindedness. Or saying that even Jekyll is complicated, so they want to build their own thing. And, go ahead - build your own static site generator, do your own thing. But the obvious reason why things are complicated isn’t because people like complexity - it’s because things like Jekyll have users with different needs.
HTML-maximalists arguing that it’s the best way probably haven’t worked on a hard enough problem to notice how insufficient SELECT boxes are. Or how the dialog element just doesn’t help much. Complaining about ARIA accessibility based on out-of-date notions when the accessibility of modern UI libraries is nothing short of fantastic. And what about dealing with complex state? Keybindings with different behaviors based on UI state. Actions that re-render parts of the page - if you update “units” from miles to meters and you want the map scale, the title element, and the measuring tools to all update seamlessly. HTML has no native solution for client-side state management, and some applications genuinely require it.
And my blog is an example of the luxury of simplicity – it’s incredibly simple! I designed the problem to make it simple so that the solution could be simple. If I needed to edit blog posts on my phone, it’d be more complicated. Or if there was a site search function. Those are normal things I’d need to do if I had a customer. And if I had big enough requirements, I’d probably use more advanced technology, because the problem would be different: I wouldn’t indignantly insist on still using some particular technology.
Not everyone, or every project, has the luxury of simplicity. If you think that it’s possible to solve complicated problems with simpler tools, go for it: maybe there’s incidental complexity to solve. If you can solve it in a convincing way, the web developers will love you and you might hit the jackpot and be able to live off of GitHub sponsors alone.
See also a new start for the web, where I wrote about “the document web” versus “the application web.”