Posted by pavel_lishin 21 hours ago
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.
What they should do, is nothing.
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.
I don't know what that history is, but did it really make a tld used by only 22k people more appealing?
While changing email is inconvenient, I don't understand the point about IoT devices. IoT devices should not depend on the Internet at all for obvious privacy and security reasons. If you are using IoT devices with "cloud" accounts, then this is a blessing in disguise. Put that garbage in the trash and rebuild around HomeAssistant, Zigbee, RTSP, etc. I find it hard to believe that someone hosting their own website would fall for the cloud IoT scam.
For instance, I own a .house domain that I use for a bunch of stuff that I've programmed. It would be a pain in the ass to go change that domain out. Now take that to next level and you're a business that's deployed a few thousand devices that need to call home.
I guess all this is to say - IoT doesn't just mean cheap botnet honeypot IP cameras. Take a look at https://www.balena.io/cloud for instance
Imagine he's set up some IoT devices at his parents' home, and those devices use services that he hosts somewhere on the internet. It would be silly to hard code the IP addresses in there, right (unless he operates his own ASN)? So he would use DNS to allow those devices to find his server(s). This would be the case whether the servers are at his home, at his office, or in a rack at a data centre.
It's a pretty bizarre exception to the normal, intuitive ways that domains work.
I'll admit that it's a crappy situation and I would be frustrated in his place. But if I were in his place, I probably would have also thought it prudent to have a backup plan.
What exactly is non-standard about an ICANN-approved TLD? Yes, the multi-level structure is a little odd, but given that ICANN approved it in the first place, one has a reasonable expectation that they would work as advertised.
Even though .science was launched in 2014 (more than a decade ago), I still consider it a non-standard TLD and still deal regularly with difficulties around its use. (For instance, you wouldn't believe how many online services reject email addresses than end in .science because they use regexes that exclude TLDs with 7 letters.)
Likewise, I've registered .lol and .fun domains but never would have assumed that just because they're available now, they will be available in perpetuity.
In that sense, .name as a third-level TLD is even more non-standard, because the standard way domain registration works is by choosing a single "second level" domain name, then adding subdomains.
Why wouldn't you assume this?
> the standard way domain registration works is by choosing a single "second level" domain name, then adding subdomains
The entire ccTLD systems of the UK, China, Germany, France, Japan, and many others would beg to differ.
.de doesn't use third level registrations.
Whereas with .name, you are choosing what appears to be a subdomain, followed by what appears to be a domain name. But under the hood, you do not own all subdomains for that domain name; you own only the combination of first_name.last_name.name.
(And yes, I am aware that you can also register last.name by itself, but only if there are no third-level domains using it, so for practical purposes it doesn't change my question above. Plus, when the OP originally purchased his domain, you could only buy third-level domains for .name.)
1) benedict.cumberbatch.name
2) drstrange.co.uk
I expect the average person would assume that for the domain 1, .name is the TLD, cumberbatch.name is registered by a private entity, and benedict (and any other subdomains associated with cumberbatch.name) is a subdomain fully controlled by that private entity.
Whereas for domain 2, I think the average person will view ".co.uk" the same way they view ".com", even though technically it is both a ccTLD and a second-level domain.
The only people who will make the assumptions you state are people who are tech-savvy enough to be familiar with those concepts, but not well-versed enough in them to understand the nuances we’re discussing.
And none of this does anything to advance the argument that .name is somehow nonstandard.
Uh, 99% of people would assume a .name address is a scam. Hate to break it to you.
For the first couple of years of .name's existence, it only allowed registration of third-level domains, and the ability to register second-level domains was added later (and only if no third-level domains existed for that second-level domain).
The author is in no way at fault here, and I don't think I would have assumed there was a heightened level of risk if I were him.