A dark and dreary road

In 2003, Dave Shea put up a single web page and invited designers to restyle it. The HTML never changed. Only the CSS did. He called it the CSS Zen Garden, and the intro text described the road that had brought the web to that point as dark and dreary, littered with the relics of an earlier way of building.

If you were around then, you know what the relics were. Layouts built out of tables inside tables. Spacer GIFs. Browser hacks stacked on browser hacks. Sites that worked in one version of one browser and nowhere else. Pages that were technically finished and practically abandoned, because nobody could maintain them and nobody tried.

The Zen Garden didn’t argue with any of that. It didn’t publish a manifesto about standards. It planted one plot, tended it in public, and let hundreds of people show what care looked like when the structure underneath was sound. The demonstration did what no argument could. Within a few years, the table-layout era was simply over.

We’re in a moment that rhymes with 2003. The tools changed, the relics changed, and the road looks familiar.

Now it’s littered with generated artifacts. Storefronts assembled in an afternoon from a prompt, launched, and left. Landing pages that exist because they could. Themes with forty apps bolted on, each one doing something the last developer didn’t know how to do in code. Copy that reads fine and says nothing. Pages that are complete in every technical sense and orphaned in every human one.

None of this is a scandal. In 2003, table layouts weren’t a scandal either. They were the pragmatic thing, the fast thing, the thing that shipped. Disposable pragmatism always feels reasonable in the moment. It only looks like a dark road when you turn around and see how much of it is abandoned.

The lesson people usually take from the Zen Garden is about separating content from presentation. That’s true and it’s the smaller lesson. The bigger one is about how change actually happens in a craft.

It didn’t happen because the argument was won. It happened because someone made a thing that was cared for, kept it working, and let people see it. A tended plot next to a littered road is its own argument. You don’t have to say anything.

We think that’s the right response now, too. Not a position on AI, for or against. We use it every day and it makes us faster. The response is to keep things. To be the people whose storefronts are visibly better in year three than year one, and who can show you the history.

There’s a nostalgia in this that we don’t want to hide. If you learned the web in the years it took to understand what the medium even was, you remember that building things was hard, and that the difficulty was part of what made the work mean something. You had to figure it out. The figuring out was where the taste came from.

When production becomes automatic, that whole layer goes quiet. The artifact still appears. The understanding that used to come with it doesn’t. We’re not against the speed. We’re against forgetting that the understanding was the point, and that somebody still has to hold it for the store to keep getting better.

So our plot is small and it is tended. Twelve or so storefronts, kept for years, with dates on them. It won’t win the argument. It isn’t trying to. It’s just there, next to the road, growing.