Seven Days on the 1999 Web: What a Guestbook Taught Me About the Modern Internet

The idea started, as most of my ideas do, at about eleven at night when I should have been asleep.

I'd been staring at my own site — this one, the one you're reading — and feeling the familiar itch. The fonts were wrong. The layout needed work. There was a plugin doing something I didn't understand to a page I'd forgotten existed. And I thought: what if I just... deleted all of it and wrote a page by hand?

So I gave myself seven days. One folder on a server. Hand-written HTML. No framework, no build step, no analytics dashboard, no CMS. A guestbook. A hit counter. A page of links. A little animated GIF that said UNDER CONSTRUCTION, because obviously.

The rules I set myself

They were deliberately strict, because otherwise I'd have cheated by Tuesday.

• Everything written by hand in a text editor. No generators.
• One stylesheet, kept under a hundred lines, and no CSS framework.
• No JavaScript unless it was doing something a 1999 site would have done.
• A working guestbook — people could actually leave a message.
• A visible hit counter on the homepage.
• No social sharing buttons. No cookie banner, because there were no cookies.

I gave myself a budget too: under 30KB per page, images included. My normal homepage is roughly two megabytes before the ad scripts wake up.

Day one and two: the strange joy of doing it badly

I'd forgotten how quick it is. Open a file, type some tags, save, refresh. No terminal tab, no dependency install, no waiting for a hot reload. The first version of the site took about forty minutes and it looked like a school project from 1998. I loved it immediately.

What struck me was the honesty of the thing. When you write a page by hand, you know exactly why every element is there. There's no mystery div. Nothing is loading that you didn't ask for. If something's broken, it's broken because you typed it wrong, and you can find it by reading the source like a normal person.

By day two I'd added a page of links to other small sites, which is a thing I haven't done since I was a teenager. It felt oddly generous. Not "here's my content, engage with it" — just: here are some places I like, go and look.

Day three: the guestbook problem

The guestbook was the whole experiment in miniature, and it nearly broke me.

In 1999 you got a free guestbook from a service that probably no longer exists, and you pasted in a snippet of code. In 2026 that means either a third-party embed that tracks everyone who walks past it, or a tiny bit of server-side scripting and a spam filter you have to actually maintain.

I went with a form and a small script. Within six hours I had fourteen spam entries advertising things I won't repeat here. I added a simple question — "what's the name of the site?" — and the spam stopped dead. Then the real messages came in.

And here's the thing that genuinely surprised me: the messages were lovely. Not performative replies, not quote-tweets, just people saying where they'd found the site and what they thought and, in two cases, recommending a band. A guestbook is a comments section with no algorithm, no ranking, no reply-guy incentive, and no notion of going viral. It's just a wall by the door.

Day four: the hit counter and the tyranny of numbers

I'd assumed the counter would be a joke. It ended up being the most uncomfortable part of the week.

For seven days I knew exactly how many people had visited. And I refreshed it. A lot. More than I'd like to admit. I'd check it after posting a link somewhere, watch it tick up, and feel that little dopamine kick — the same one the big platforms are engineered to produce, only I'd built it myself out of vanity and a text file.

Except there was no retention dashboard, no average read time, no scroll depth, no audience demographics. Just a number. It was somehow both addictive and completely uninformative, which is a decent summary of most analytics I've ever looked at.

What 1999 got right

The small web, basically.

Ownership. My pages lived in a folder I could zip up and move anywhere. No platform could change the terms, throttle the reach, or decide my site was now a video app.

Permission-free publishing. I wrote a thing and it appeared. No review queue, no reach penalty for a link that takes people off-site.

Links as a gift. The links page was my favourite thing I made all week. Pointing at someone else's work, cost-free, with no expectation of reciprocity.

A page that ends. There's something restful about a document with a bottom. No infinite scroll, no autoplay, no recommendations. You read it, you reach the end, you leave. That's it.

What 1999 got wrong

Plenty, and we shouldn't pretend otherwise.

Most of it was inaccessible — tiny fixed-width tables, white text on busy backgrounds, image maps, no alt text. A lot of it was unreadable on a phone, and some of it was simply broken. Frames were a war crime and I will not be taking questions. And the gatekeeping was real: if you couldn't hand-code, you weren't publishing.

I also don't miss the search situation. In 1999 finding anything meant a directory and a prayer. Modern search has plenty of problems, but I'm not nostalgic for AltaVista.

The bits I'm keeping

I didn't keep the hit counter. I did keep three things.

First, a links page. It's small, it's hand-maintained, and it's already the most-read page on the site after the homepage.

Second, a guestbook — properly moderated, but the same shape: a wall by the door rather than a feed.

Third, and most usefully, a page weight budget. Everything I publish now gets checked against it. It has changed how I write: fewer embedded things, more actual sentences, less stuff loading for the sake of it.

The week didn't make me want to live in the past. It made me want to be more deliberate about the present. The old web wasn't better because it was old — it was better in a few specific ways that we quietly agreed to give up, and I don't think we had to give all of them up.

Anyway. The guestbook's still open if you fancy it.