The .self TLD Is Real Now: ICANN Application Filed

A nonprofit founded in Colorado eight months ago wants to hand every self-hoster a free domain and put "surveillance capitalism" in its mission statement. That's the pitch behind .self, a proposed top-level domain from the Human-Centered Computing Foundation, and as of this month it's cleared its first real hurdle with ICANN: acceptance into the Applicant Support Program, the discount track for applicants who can't stomach the standard $227,000 gTLD application fee.

If you self-host anything, a namespace built specifically for that purpose sounds appealing on the surface. It's also, right now, mostly a single-page PDF and a set of promises. Here's what's confirmed, what's still vapor, and what a working self-hoster should do about it, which for now is nothing.

What HCCF is proposing

The pitch, in HCCF's own words: an internet infrastructure that has become "dominated by algorithms that exploit our data and capture users' attention," with .self positioned as an answer starting at the naming layer. The concrete mechanics, as far as they've been published:

• One person, one domain. Each individual gets a single .self address tied to their identity, not a corporation's.
• Free. No registration fee, and by explicit design, no domain investors. HCCF has stated it doesn't want the namespace turning into a speculation market the way plenty of gTLDs have.
• Self-hosting by default. Every .self domain is meant to point at infrastructure the holder controls, not a third-party platform.
  That's the extent of the public technical detail. There's no published DNS resolution architecture, no identity verification mechanism, and no registry service provider named yet. HCCF has said it's "in the very early stages," which is an unusually honest thing for an applicant to volunteer.

The ICANN process reality check

Getting into the Applicant Support Program is not the same as getting a gTLD. ICANN itself is explicit that ASP acceptance doesn't qualify anyone for approval. HCCF still has to file a complete application, survive evaluation, handle any objections from existing trademark holders or competing applicants, select a registry service provider capable of running real DNS infrastructure at uptime SLAs ICANN enforces, and prove it can handle abuse response and policy controls for an entire namespace.

The application is going in during the 2026 new gTLD round, which opened this year for the first time since 2012, and which already has more than 100 bids publicly announced across every category from city names to novelty strings. HCCF's own site lists Riley O'Donnell and Lucas Silva as CEO and COO. Domain-industry tracking shows strings ending in "self" already occur in existing gTLDs at a rate comparable to niche extensions like .contractors or .tattoo, with generic terms like "yourself" and "myself" accounting for roughly half of that volume, worth knowing if you're wondering whether .self is functionally novel or one entry in a much larger 2026 gTLD gold rush.

Realistic timeline: gTLD evaluation rounds routinely run multiple years from initial application to delegation, even for applicants who sail through with zero objections. Nobody's registering a .self domain in 2026 or 2027.

The skepticism is worth taking seriously

The most useful pushback so far is specific, not cynical dismissal. One detailed writeup flagged the obvious irony first: an organization campaigning for a human-centered, less-extractive web published its founding manifesto as a single-page PDF instead of an HTML page, which is exactly the kind of accessibility miss the project claims to be correcting.

The harder question is underneath that one. A "one person, one domain" model requires verifying that each applicant is one person, not a bot farm or a bulk registrant working around the free-domain, no-investors rule. That means collecting identity information, and plausibly a stable IP address, from every registrant, on an ongoing basis, from a nonprofit with limited resources. HCCF's stated mission is reducing the amount of personal data flowing through centralized chokepoints. Running identity verification and abuse moderation at gTLD scale is itself a data collection and trust problem, arguably a bigger one than the naming-layer problem .self is trying to solve. HCCF's own site acknowledges this is more a verification, technology, security, and abuse problem than a lack of will, which is a fair assessment of the gap between the mission statement and a working registry.

There's also a simpler critique worth sitting with: the shortage on the self-hosted web isn't personal websites. Anyone who wants one can already stand one up under any TLD that exists today. What's missing is discovery and connection between them, and a naming convention alone doesn't solve that.

What's out of scope here

This piece covers the .self proposal and the ICANN process it's entering. It doesn't cover the broader 2026 new gTLD round, which reopened for the first time in over a decade and has its own set of controversies (Amazon's pending gTLD portfolio, a lawsuit against ICANN unrelated to this application, and a wave of city and novelty-string bids) worth a separate look if you're following domain policy generally.

What to do about this today

Nothing, yet, and that's fine. If the appeal of .self is the idea of a domain that signals "this points at infrastructure I run myself," you can have that today under any TLD. If you're already running the kind of self-hosted productivity stack that .self is theoretically meant to represent, a .self address changes nothing about how your Nextcloud or Immich instance works. It's a naming layer, not infrastructure. Worth tracking as the application moves through evaluation, not worth planning around yet.