The indicators of a quality website (somewhat silly)

You know it's a good website when it:

1. Uses subdomains (except for username.platform.com)
   ... especially multiple levels
2. Has a non-standard port number in URL
3. Is served over a plaintext connection
4. Uses very nearly default styles
5. All tag names are written in SCREAMING CASE
6. The path ends in .html
7. The URL has meaningful directories and names
8. The HTML has typos
9. Has commented out draft work

Of course, these are all just indicators of age.

1. The use of subdomains fell out of favor due to IP address exhaustion and reverse proxy software. If your going to be setting up a proxy anyways, it doesn't make much sense to fiddle around with DNS. (which often makes setting up TLS harder)

On my site, the server handling / and /mcu aren't even on the same continent.

At one point, it was standard practice to setup a subdomains per protocol in case one had to be moved to new infrastructure, but HTTP is the last one standing: public access FTP/Gopher/Telnet are just about extinct.

Even in places where they absolutely should be used, subdomains seem to have fallen out of fashion. Many companies will register a new domain for an internal service instead of putting it under their well known name.

The last surviving use is to isolate user provided content: any platform that allows custom HTML/JS/CSS will give each user a subdomain so that browsers know it's not from the main site.

2. Port numbers are another victim of reverse proxies. Additional services are likely to be proxied instead of being exposed directly, especially as they often rely on the proxy to implement rate limiting and blacklists.

Nowadays, it's standard practice to route everything through Cloudflare, which doesn't support port numbers in the URL. This is part due the huge amount of bot traffic from hacked IOT devices: please do not connect your TV/toaster/security camera to the internet.

3. Ciphered connections have been the default for a long time. Many hosting providers don't even offer plaintext as an option.

As a result, the old trick...

# telnet maurycyz.com 80
Trying 65.109.172.162...
Connected to maurycyz.com.
Escape character is '^]'.
GET /misc/quality/indicators.html HTTP/1.0
Host: maurycyz.com

HTTP/1.1 200 OK
Server: nginx
Date: Mon, 05 Oct 2026 16:07:21 GMT
Content-Type: text/html; charset=utf-8
Content-Length: 14891

Accept-Ranges: bytes



        
        

Connection closed by foreign host.

... likely won't work unless the site owner has gone out of their way to change the default to use a "less secure" configuration. (In practice, a hard redirect doesn't do much because an active attacker can just intercept it, and users are still protected from passive attackers because browsers use encryption by default)

4. Very early browsers didn't support CSS, and stylesheets were quite limited for the first few years. The resulting pages might have a custom background, but bear a strong resemblance to default styles.

5. All-caps tag names helps when reading printouts or using text editors that don't support syntax highlighting... both of which are very rare nowadays.

The rest (.html, meaningful paths, typos and comments) are indicators of handcrafted markup served from a directory.

While some antiquarians still do this, most webpages are published using a fully automated content management system: written in markdown (often by an LLM) and then stuffed into a chronological feed and newsletter. This lets people treat their website like social media, complete with inline ads.

Enjoying this post? Subscribe to the newsletter. Directly to your inbox.

You're lucky to get a hand written URL slug instead of:

  https://someone.smallcrow.com/p/title-except-it-cuts-off-randomly

Put together, having most of these factors reliably dates a website to the 1991 - 1998 range, although many have been continually updated since.

However, age is a strong indicator of quality.

Keeping something on the web for decades requires active work, and in most cases, considerable expense. If someone thinks it's worth keeping around, it's probably worth reading.

... and you can be sure it's not LLM slop, which has caused a sharp decline in the quality of pages. Sure, it's spelled correctly, but its a rambling description of a reality that is quite different to our own.

Related:

• /misc/domains/: The consequences of refusing to use subdomains