The correct way to handle this mess is to simply keep things messy. BGP tables are a big mess, for example, because a lot of companies keep their IPs when going from ISP to ISP.
But, assuming that Verisign can’t keep third level domains (as a DNS implementer, I don’t think third level domains is a huge deal; see thread below):
* Third level names where only one person has the second level domain should be transferred to whoever owns that single third level name.
* Third level names where multiple people have the same second level domain should be put up for closed bidding: Only current owners of .name domains with a given second level domain name (e.g. last name) will be able to bid for the second-level domain. So, if one has john.smith.name and joe.smith.name, Joe Smith and John Smith will be in a bidding war for smith.name.
If the issue of .name not being in public suffix is a real issue, Verisign can handle that by disabling new third-level .name registrations, and provide Public Suffix with a list of those registrations (just send all the owners a privacy notice, making it clear that the existence of the name will be made public for security reasons). More reading: https://github.com/publicsuffix/list/issues/2306 (There seems to be issues with this list being too long to keep in the Public Suffix because there’s too much software out there which can’t handle it. That seems strange to me: Even here in 2026 where RAM costs far too much, Deadwood can store a list of 240,000 blacklisted entries in under 10 megs; there are about 22,000 three-level .name domains and I could store that list in a way that could be very quickly looked up in about a meg of memory)
Now, personally, I think Verisign can keep these messy third level names, and are doing things this way so that Neil Fraser has to compete with every single 2-bit cybersquatter out there for the rights to fraser.name.
As an aside, it’s trivial to have DNS servers handle multi-level domains without having to have a zone file for every level; e.g. https://this.is.a.long.name.maradns.org works, and there’s no zone file for name.maradns.org, long.name.maradns.org, a.long.name.maradns.org, and so on.
Also, since people have brought up the “org fuckery” without providing details: https://bluecatnetworks.com/press/the-org-domain-sale-explai...
You know dang well if .org was owned by an investment entity, they would had jacked up the prices as much as they could get away with.
The auction plan is gross. It is unfair to say "This thing you thought you were buying the right to exclusively control (subject to continued renewal payments)? We're gonna transfer it next week to whichever Fraser pays us the most. Happy bidding!" It wouldn't be any more fair to do that than to announce to all .com owners that there will now be an auction between them and everyone who ever typed that name into a search at the registrar.
What they should do, is nothing.
I agree.
As a DNS implementer, the action plan is unnecessary. There are, what, only 22,000 or so .name domains. One can write code to do two lookups for firstname.lastname.name: If firstname.lastname.name is found, return the NS delegation. Otherwise, if lastname.name is found, return that NS delegation. Finally, if neither is found, return NXDOMAIN.
One argument is that this is hard to implement in the real world (it’s about one day, at most one week for a skilled DNS developer to pull off; probably half a day to be honest, and yes I have written code like this), so then yeah if that’s a real concern let’s have a closed auction. I’m opposed to the auction, based on my experience that this isn’t hard to implement.
If it’s an issue, just stop all new firstname.lastname.name registrations, and only allow lastname.name new registrations. Then we only need to deal with this corner case for about 22,000 domains, which we can keep in a special hash and would take about four megs to store.
Love the practical explanation of how straightforward it should be. Especially with the finite set. Hell, they could send an email to all 22,000 saying "We're sorry, we have to raise your renewal fees by $12 a year. We hate to do this, but we're having to do a lot of extra work to support this unique type of domain." they'd bring in $264,000 a year, surely that should be enough to pay a single skilled developer to manage this and also be available to help smooth over any future complications.