Actually, I think ICANN's institutional authority is very circumscribed. The UDRP only happened because NSI was willing to go along with it as part of exiting the transitional uncertainties that also yielded ICANN. [I know you think this isn't so, and that NSI had to go along with intense pressures of the time; what I mean is that the pressures weren't coming from ICANN -- they were coming from sources that would exist with or without ICANN.] I don't see either of David's examples as something that ICANN actually has the authority on the ground to implement. We might be able to converge on actions ICANN can take that affect the larger world in some way. Controlling the chartering of new TLDs is one of them. I think it is using overly valenced language to describe a content-based process for determining what new TLDs should come into existence as a form of "censorship." It might be better for ICANN to simply hold a lottery or somesuch among all applicants, a content-neutral way of awarding new TLD slots if it insists that there can't be hundreds or thousands. But yes, I still believe that a decision to, say, exclude .XXX -- aside from the harm to any settled expectations of the would-be .XXX operator -- is not in the same category as speech-affecting decisions like those to filter Internet traffic itself or to arrest people who petition the government for a redress of grievances. It's true that I've weighed in a few times on how much domain names "matter," and have offered the answers "not much," or even "not as much as many people say." (This is still short of "running around" telling people to go away, but that's a matter of degree, and of perception.) While you and I may not come to agreement on that, I think it would be very helpful to the cause of drawing people into the debates to be very sober about where they might matter and where they won't. To me, phrases like "lever of global control" are too suggestive of something larger than I think even the people who invoke them mean to say. When they're used, and then fire up politicians and nearby public interests organizations, their interest then wanes when they fail to see the connection they've been primed to see. I don't think a consensus that ICANN can choose what TLDs to add using content-based factors -- perhaps one of "avoiding controversy" -- translates to a consensus or practice overall about censorship on the Net or even more broadly. And too often people hint darkly at "control of the Internet" in the same breath as ICANN, when we all know that that is a devastating simplification. I see the CDA as a very different creature. There the full gun-toting power of the world's most powerful government to put people in jail had been brought to bear directly against individuals. Put something on the Net later deemed harmful to minors, without checking for a credit card, and you can become a felon. That's an act in a completely different category from anything that ICANN could conceivably do, and certainly from deciding what TLDs to introduce. The introduction (or not) of .XXX or .protest or dot-anything else is neither a meaningful subsidy or barrier to the production of information thought related to that TLD's signified concept. (I'd still be happy to see a .protest come about! Certainly harder for companies to argue trademark confusion there.) By the way, I completely see how the way the situation is viewed can vary by discipline. Some OII colleagues in the social sciences have had a hard time even parsing a sentence like "ICANN doesn't matter." Their response is: whether it matters is an empirical question that depends on ... whether people think it matters. So long as people are showing up at the meetings, ipso facto it matters. I think that may be a genus of your institutional flypaper point: it's a common meeting place for people who want to exercise authority in these areas, especially because there are not obvious others. On what I see as the threats to speech/freedom: there are the known but still powerful ones, and the less known and therefore even more dangerous ones. In the first category I'd put increasingly sophisticated government surveillance, including dragnets/vacuum cleaners, deployed against a comparatively unsophisticated public that will not be using PGP and hushmail. I'd also put the straightforward blocking/filtering of Internet content undertaken by national governments in that category. In the less known category I'd put peer-produced systems of surveillance and reputation aggregation, and also the rise of "tethered appliances" -- devices like iPhones and BlackBerries that are eminently more controllable by their vendors (and not much modifiable by anyone else), and thus able to serve as constricting points of information flow. I'd also raise our current crop of natural monopolies: Google as a search engine -- its rankings are far more subtle and powerful than whether a new TLD is chartered, and yet no one calls for Google to be governed by a board representing the world and the public interest -- and those other "private sheriffs" in a position to shape what we can do, see, and express on the network. Apart from writing about it, my own zone has been in prototyping tools and proto-institutions that can better the status quo, such as stopbadware.org. Helen Nissenbaum's Track-Me-Not would fit into that zone, too. And I think we can persuade some private parties -- whether the likes of Microsoft or Cisco -- to build speech- and privacy- protective technologies into what they develop. Of course, there are many ways up the mountain! Best, Jon