(I dropped the staff alias; this doesn't seem like instructions to them.) On Tue, Nov 24, 2015 at 05:29:57PM -0500, George Sadowsky wrote:
1. Can ICANN Inc. make policies regarding which strings they will delegate? What are the degrees of freedom under which they can make them?
Yes, of course it can. It has the control over the root zone, and it can make approximately any policy it likes about delegation from there, for the same reasons that any zone operator makes those policies. If ICANN wants to make rules about strings in $language, and how it will treat those strings when making policy evaluations, it can go nuts. It can decide that it will not make delegations using any string that contains the letter "q" for all I care. But whatever that string is and whatever it means in whatever language was under consideration, that's not the "semantic meaning" of the _domain name_. I'm aware that some people think this is a pointless distinction, but the failure to have made it in the past is how ICANN (from what I've heard, in the person of one of its former CEOs) ended up leaving people with the impression that "variants" would "work", even though it was never technically possible to make such a policy for the whole Internet. We are more than 10 years into the tragicomedy that resulted from such promises, and it's more than a little depressing to me that we're still at the stage where people do not automatically repeat, "There is no semantics to a domain name," as soon as the idea is floated.
2. Can the "ICANN community" make policies regarding which strings they will delegate? What are the degrees of freedom under which they can make them?
No idea. The distinction between the ICANN community and ICANN Inc seems to me to be a problem a long way distant from the decision to delegate a domain name by some administrator. I think this is a matter of how policies are developed inside ICANN, and it's all long before the issue of what a domain name means.
3. Milton might argue that there should be complete freedom to propose any string you'd like, and it should be accepted. Do you agree with that? (Does Milton?)
I certainly have no intention of speaking for Milton. If _I_ had my druthers, there would be complete freedom to propose any string with the guarantee that the answer would be no, but I'm not trying to make policy here. I'm trying to draw a distinction between ICANN making policy for consideration of strings in the DNS, and ICANN delegating things in the DNS. Those are different activities, and one of them has no resulting semantics.
4. Under what conditions, if any, should the semantic content of a natural language string be grounds for refusal to consider it as a new gTLD, in any future nGTLD round?
If you're going to talk about meanings of a string in some natural language, then you're already outside the DNS. That's the _whole point_ of what I'm saying. Names in the DNS happen to get used in natural language all the time. For those users of language, the domain names almost certainly have semantics. This is true even though, in the name "ns1.example.com.", neither "ns1" nor "com" is a word. But keeping the semantics outside the domain name system, and then allowing ICANN itself to figure out however it wants to decide that someone else has made meanings for the strings it needs to worry about when delegating, are a long way from talking about the "semantic meaning of a domain name". For instance, suppose I wanted to delegate really_offensive_string.crankycanuck.ca. I'm the registrant of crankycanuck.ca. This is just none of ICANN's business, and I think you'll agree that it isn't. There _certainly_ are people who have suggested that ICANN should get into regulating that by contract. If the ICANN bylaws discuss the meaning of domain names, then ICANN has to start arbitrating meanings of strings anywhere in the tree. If it's about the use of domain names, then you automatically get the protection of the delegation system of the DNS and you can quite correctly say, "Not our baliwick."
5. The problem is that these strings are read by both computers and by people, and are processed very differently, with very different reactions. Are we to ignore that?
No, but we're not to privilege the political view over ICANN's technical responsibility, either. There is no damage that results from saying "use in some natural language" instead of "meaning of a domain name", except to those who want to use ICANN as a regulator of the entire domain name system. Since the whole system was designed precisely to prevent such kinds of centralization, I regard that particular damage as entirely salubrious.
These are questions that will come up and will be important in the future. Do you think that there is unanimity regarding the answers? It doesn't matter what you or I believe if there is no unanimity, because these questions will recur and the resulting policies may not be what we or others might want.
I believe there is _not_ unanimity, because like others on this list I believe that a number of people would like very much to make ICANN a locus of contractual regulation of the Internet. I think there is unanimity among anyone who correctly understands the limited and narrow but important and legitimate role for ICANN as the operator of the root zone. Best regards, A -- Andrew Sullivan ajs@anvilwalrusden.com