03 December, 2010

First Critique of Economic Study

George Kirikos writes:

Now, on to the latest "paper". Briefly, it is junk, and mere intellectual masturbation. It once again lacks any empirical rigour. It also ignores what the
Department of Commerce letter of 2008 *told* ICANN to study:


Take a look at PAGE 1 of that PDF!! Where is the work about whether the domain registration market is one market or separate markets? Where is the work about substitutability? Where is the work about switching costs? And so on? Is ICANN trying to pull a fast one on the public, by not even instructing its paid consultants to do the right study??

The authors have not even attempted to perform a true study of the AFTERMARKET for domain names. They make only anecdotal references to a handful of transactions, rather than doing a *systematic* study of ALL reported transactions. That's the difference between a casual "survey" paper and real original research that is rigorous.  The data *is* available (e.g. DNJournal.com and other sources have years worth of transactions). But, it appears the authors were rushing to slap together something, anything, that ICANN can "claim" is supportive of the new TLD process. Perhaps ICANN was counting on an assumption that a last minute publication would be "overlooked" or "ignored". That assumption is incorrect.

The "empirical" work, if you can call it that, is hilarious to read. These are supposed to be papers by EXPERTS, not newbies. Paragraph 17 (page 9) talks about a "non-random sample of five generic words." You *must* be kidding. Is that a statistically significant sample? Where did these authors study statistics?? I really want to know. Five words. Not even "random". This is basic "anecdotal" hand waving work, not anything experts would call "empirical" work.

The authors throw around buzz words like "principle of revealed preference" to attempt to pretend that there is deep underlying scholarly understanding behind their work. Ultimately, though, they failed to do the real work. They could have done proper statistical tests, the systematic study of the entire universe of domain names (zone files, hello??) broad surveys of real domain name participants/registrants, but instead did not do the proper work. They could have studied *actual* preferences, via real world data. They could call up Amazon.com, and ask them what it would cost for them to switch to a different domain name. Did they? Of course not.

Similar to the "benefits" section of the paper, the "costs" work was a joke. It was a series of anecdotes, without reference to rigorous real world data. Ask Verizon or Microsoft or any company why they own tens of thousands of domain names (99% of which are *defensive* registrations). Oh, I forgot, it was the *job* of these authors to ask them, but they didn't. Ooops.

A proper study would have said "the benefits of new TLDs" would be $X billion, and the costs would be "$Y billion" and would allocate those costs to various parties (consumers, registrars, registries, etc.). The number of new TLDs would have been discussed (rather than "unlimited" or "1,000/year"). The lack of any financial numbers of this nature, or even an *attempt* to get to those numbers, tell us that the study was completely worthless.

No comments:

Post a Comment