> The Internet Corporation for Assigned Names and Numbers (ICANN) is the private sector, non-profit corporation
If you made your own DNS root and got folks to adopt it, you could make your own TLDs as well. Folks have tried in the past, none have really caught on (outside of a few specialized services, like .onion for Tor).
Why? Why can't we have as many custom TLD's as we want? Or why can't my blog just end with .helloworld if I decide to do so?
(* Unless it's a legacy IPv4 allocation that predates the RIR system.)
You can do whatever you want, but the internet runs on consensus and if consensus devolves into everyone doing whatever they want then DNS doesn't add anything beyond just using raw IP addresses.
Is ridiculous and an example of something I think should be avoided, but even it has added value beyond a raw IP address
It's essentially the same attention-consensus-economy game that web startups do, but earlier, more primitive, and deeper.
They're now arbitraging the trust which got them to this point.
"At no real cost..." True. Except the cost of running one of the 13 server addresses. And IMO it could be a dog and no one would notice. I think the A server (198.41.0.4) is really all anyone needs. The cost of a new TLD to ICANN is the cost of editing a text file.
"By what authority..." None.
", why are they allowed..." A question I have been asking for over 20 years. Answer: Because we let them?
You can say no to ICANN. Run your own root on 127.x.x.x. You can edit the root.zone to be just as you want it. Want to delete a silly TLD (e.g., .loans)? Edit a text file. Want to add your own new TLD? Edit a text file. The cost? Editing a text file.
I recall a former Board member of ICANN admitting he himself ran his own root for many years.
ICANN's ability to make millions in profit from TLD's relies on an interesting prerequisite. All DNS admins have to use a root.hints file that points to the (13) addresses serving ICANN's root.zone. Often they have no idea this root.hints file even exists, let alone have the guts to edit it. The root server addresses to use are chosen by the authors of the DNS software, e.g., the software automatically downloads root.hints from ICANN to bootstrap itself.
If admins or users choose to use a different list of root server addresses (e.g., 127.x.x.x, 10.x.x.x., etc.), all bets are off.
So how do you stop ICANN from making millions posing as a pseudo licensing authority for registries? One way is to stop using ICANN's root.hints and use a different root.zone that you control. If enough people do this then one day ICANN has no relevance.
Right. Not gonna happen. I'm probably one of only a small number of users who will ever run their own root.
So far without any success :)
The Tor network can "register" and address ( call it a pointer, because it's basically a hash ) for .onion. That means if you run the the tor service, you have a pointer, and this is probably a good solution.
So if we'd all had a distributed global "dns"-like network, where each and every system gets a unique id this would partially be solved, for free, but the requirement for master nodes (like tracker servers for torrents ) would probably still be present.
The problem with this approach is that obviously you can't make a choice for xyz.mydomain, because there will be at least someone else who wants xyz.mydomain, and in this case, who and how would decide which of you can have it? Right now this decision maker is money, which of course is an issue, but at least a solution.
So: does anyone have a distributed, fair solution, that is able to solve disputes and act as a replacement for dns? So far I'm not aware of any.
https://www.iis.se/vad-vi-gor/pdt/#PDT_eng
Disclaimer: I work for the Internet foundation in Sweden.
For .coop we had to have fully redundant servers (ie < 99.9999 uptime) on 4 continents
So if you had applied to bid, and lost out on .blog, you would have gotten an equal share of the money that Automattic paid to be the owner of the TLD.