February 24, 2005

In Answer Chris Ambler's Good Questions

I'm glad that Chris is asking the questions that he does in his blog, the most recent entry being at http://onthenet.ambler.net/blog/_archives/2005/2/23/368633.html.

Chris continues our discussion of how would we resolve the situation were there to be multiple, and presumably different, versions of a TLD, our example being the TLD that I operate, .ewe.

I have suggested the NTP (Network Time Protocol) as an example of a design that deals with a critical element of the net and our economic fabric - accurate time - but which does not require that there be one single authoritative source.  Rather NTP accepts multiple claims of authority and through a process of consensus and heuristics weeks out false tickers (which do exist whether by intent or by error) and produces a result good enough for all but the most precision of timekeeping tasks.  I ask why DNS could not be refined to allow multiple sources and through a process of human and automated selection, weed out "bogus" sources.

In the next round Chris pointed out that the question of time is far less subjective than the question of which version of aTLD to accept.  That is quite true.

My response was that we have a plethora of existing mechanisms - both economic and legal - that resolve those kinds of situations.  Some of these mechanisms work by not resolving the question at all but rather leave it up to the discretion of the internet consumer, or their ISP agents and proxies, to make the choice based on their own needs and desires.

Chris continues by pointing out that the net spans national boundaries and that the mechanisms I refer to are much weaker in an inter-national setting than they are in intra-national settings.

As usual there is much merit in what Chris says.  I would be a dunderhead if I were to ignore the fact that, yes, the mechanisms to resolve disputes across national boundaries are weaker.  But I also recognize that for the most part, if somebody in Olliestan or Fredonia, comes up with their own version of .ewe that A) they are far more likely than I to have direct experience with ewes and B) are likely to be playing to a difference audience of consumers than I am here in California.  In other words, as a practical matter, our conflict will probably be more imaginary than real.

But suppose that the conflict becomes real - then there is the meta question whether we ought to establish a body of internet governance just to save little ol' me the trouble of trying to engage in a rights wrestling match with someone in a barely electrified country on the other side of the globe?  (And isn't that what has been much of the justification so far for ICANN - that it saves a small group of people, the intellectual property industry, a few bucks to enforce rights that most of the rest of us couldn't care less about?)

Chris' final paragraph is where the really interesting stuff is to be found.  Chris asks why each ISP doesn't just create its own DNS tailored to its own customer base and with assurances of quality, as measured by the preferences of that customer base.  My answer is this: Yes!  This is what I have been suggesting: that the pieces needed to create a root zone are so small and easy to move about that pretty much anyone, from individual techies to ISPs, can construct their own DNS services for their own use.  Any inconsistencies with other DNS systems will tend to be resolved by the natural pressure to remove undesired inconsistencies (recognizing that some inconsistencies, such as the removal of names that tend to lead to porn or similar materials are a kind of inconsistency that many people actually want to have.)

In these discussions I'm always reminded of Monty Python's Hungarian Phrasebook/Tobacconist Shop sketch in which a Hungarian tourist tries to use a rather defective Hungarian-to-English phrasebook.  I consider a DNS system that provides bad name translations, with badness being measured from my point of view as a user of those name translations, as a DNS system that is much like that Hungarian Phrasebook - something to be tossed into the dustbin at the earliest opportunity and replaced with a system that delivers results much more to my liking.

Posted by karl at February 24, 2005 12:35 AM