A universe of dots

TTL 49

Part of my job at Automattic is always to remember how it feels to be a user of our products.

For that, I like to step into the role of an ordinary customer using our own tools for a real project — something I care about.

Currently, that project for me is a photography side project I have been slowly bringing back to life. I will talk about it properly in a future post. For now, what matters is that to do it the way I want, I ended up registering a small family of domain names.

Working on that brought back an old obsession of mine: how much meaning we have lost in domain names, and how much we could still do with that lost meaning if the industry decided to use it differently.

Let me explain.

When every character meant something

In the beginning, domain names were not about brand creativity or clever hacks. They were about structure and efficiency.

Leaving aside the very first .arpa (which belonged to the US research agency that built ARPANET), the core generic TLDs were:

• .com – commercial entities
• .org – non-commercial organisations
• .net – network operators and infrastructure
• .edu – educational institutions
• .gov – US government
• .mil – US military

They were born in a world where computers were slow, memory was expensive, and every character mattered. If every domain name had to end with “dot something”, that “something” needed to carry meaning.

Then came the country-code domains: .us, .it, .uk, .ch, and so on. Again, those last two letters meant something. You did not have to explain that .it was Italian or .at Austrian. The TLD itself was information.

The important thing is: at the start, the ending of a domain name told you something real about what you were about to visit.

Fast-forward a couple of decades, and look at what happened.

And then chaos came

Between the mid-90s and the early 2000s, the web exploded, and with it, the demand for names.

Opening up .com, .net, and .org to everyone solved the scarcity problem, but it also diluted their meaning. Today, nobody sees .org and thinks “ah, this must be a non-profit”, or .net and thinks “this must be a backbone provider”.

Then, small countries realized they were sitting on short, attractive two-letter strings, and started licensing them for global use:

• .me (Montenegro) became a home for personal sites.
• .tv (Tuvalu) turned into a magnet for video platforms.
• .fm (Federated States of Micronesia) became popular with music and radio.
• .co (Colombia) was promoted as an alternative to .com.
• .ai (Anguilla) is now the default choice for anything related to artificial intelligence.

From the country’s perspective, this makes perfect sense. You license the TLD to a private operator, share the revenue, and suddenly your ccTLD is a global product. It is smart public policy.

From a meaning perspective, though, labels that were originally designed to give structure to the namespace slowly turned into branding material. The semantics got fuzzy.

Today, if I tell you “go to p3ob7o.com”, you know almost nothing about what you will find there. You have to click to discover.

Personally, I find that a little wasteful.

Why I like .blog so much

One of the TLDs I enjoy the most is .blog (I might be slightly biased, but that’s not the point 😂).

If I tell you “go to paolo.blog”, you know what you are walking into before you even open the page. It is my blog. You might not know the topic or the tone, but the genre is clear.

If I say “go to paolobelcastro.com”, which is also mine, what do you expect? A personal site? A company page? A portfolio? A blog? All of the above?

You only know once you are there. The domain is not helping you.

Over the last decade or so, we have added hundreds of such TLDs: .store, .shop, .news, .photo , .art, .studio, and many others. Some are generic, some are focused, but a subset genuinely describes a type of content or activity.

With all this variety, it seems like a waste not to take advantage of it.

One domain for every need

The project I am working on right now is a good example.

It has at least four distinct faces:

• a place where I write about the subject
• a place where I simply show the work
• a place where people can buy prints or digital files
• a place where I catalogue the pieces I want to preserve long-term, with proper metadata

Those are very different tasks. Different audiences, different expectations.

The traditional way to do this with domains is:

• pick one TLD, usually .com,
• build a site at example.com,
• and then, if you want to separate the areas, use subdomains: blog.example.com, shop.example.com, gallery.example.com, etc.

Technically, this works fine. From an operator’s perspective, subdomains are convenient: you stay within one registration, you manage renewals in one place, and you inherit SEO authority more easily.

But from a language perspective, you are putting the burden on the left side of the dot. All the meaning must be squeezed into the subdomain or the path: blog., shop., /store, /photos…

On the right side, the TLD sits there as a generic “.com” or equivalent, adding little or nothing.

If I had to design this from scratch today, with the tools we actually have available, I would do the opposite.

Instead of:

• blog.example.com
• shop.example.com
• gallery.example.com

I would love to say:

• example.blog
• example.store
• example.photo
• example.art

Meaningful ending telling you, in one glance, what you will find.

I am, in fact, experimenting with exactly this pattern on my current project, but for the purposes of this post, the pattern itself is more important than my specific implementation.

The product we never built

Now, here is the crucial part.

As an individual, I can go out and register these domains one by one, from different registrars, operated by different registries, at different prices, with different terms. I have done it. I am doing it.

But this is not something you can reasonably expect from the average person who just wants to put a serious project online.

If you show them a giant search box and say “type a name, pick a TLD”, they will pick one. Maybe two, if they are paranoid about brand protection.

What they will not do, spontaneously, is to understand the potential of owning a small, coherent suite of semantic TLDs that map to different aspects of their work.

Creating that understanding is not the user’s job. It is our job, as an industry.

When the big wave of new gTLDs was launched years ago, that was the window of opportunity to build this as a first-class product. All those namespaces were empty. Nobody had taken “their” strings yet. In theory, it was the perfect moment to say:

“If you are a creator, we have this package:

yourname.blog, yourname.photo, yourname.art.

If you are a small business, we have this package:

yourbrand.store, yourbrand.news, yourbrand.blog, yourbrand.support.

If you are a museum, we have this package:

yourmuseum.art, yourmuseum.store, yourmuseum.digital, yourmuseum.events.

The closest thing we have to a “cross-TLD product” today is something like trademark blocking services: you pay once to prevent anyone from registering your brand across hundreds of TLDs. That is useful, but notice the direction: it is about blocking potential use, not enabling it.

I cannot help feeling that we missed a chance there.

The second-best time is now

The good news is that, even today, the majority of the newer TLDs are far from saturated. Together, they count tens of millions of registrations, spread across hundreds of extensions, but only a small fraction of those extensions are “crowded”.

There is still plenty of space to build sensible combinations.

Technically, nothing prevents registries and registrars from sitting around a table and designing opinionated packages for different types of users.

Commercially, yes, a package of five domains will generally cost more than a single .com. But in the context of any project serious enough to invest in branding and infrastructure, domain fees are not the bottleneck.

The true constraint is cognitive, not financial. If you force every user to invent their own system, most will default to the simplest possible mental model: one project, one domain.

If, instead, you propose a clear, ready-made structure – “this name, these four extensions, this is what they are for” – you are not pushing complexity on them, you are offering clarity.

A question for the industry

I want to be careful here.

This is not a lecture about what “we should have done 10 years ago”. I have been talking about ideas like this for a long time, and I have not built such a product either. It would be unfair to pretend otherwise.

So I will frame it as a question, not as a verdict.

We work in an industry where:

• only a small minority of people online own a domain name at all
• most of those who do, own exactly one

At the same time, we manage hundreds of TLDs that are, collectively, underused and rich in potential meaning.

Do we really believe the best we can do with this situation is selling each extension one by one?

Or do we think there is still room to design something more ambitious and more helpful, especially for creators, small organizations, and the kind of projects that could genuinely benefit from having a clearer, richer “family” of names?

If you run a registry, a registrar, or a hosting company, I am curious:

How are you thinking about this?

Are you experimenting with semantic bundles of domains?

As for me, I am quietly testing this model on my personal project that combines writing, photography, a small store, and a curated archive.

When it is ready, I will share the details.

In the meantime, I would love to see us, as an industry, treat those last few characters after the dot as something more than decoration again.

If someone forwarded this to you, you can subscribe.

I also publish on paolo.blog and monochrome.blog.