Ben Edelman of the Berkman Center at Harvard has compiled another study on domain names, this one entitled “DNS as a Search Engine: A Quantitative Evaluation.”  As a matter of disclosure, I admit that I suggested this project to Ben and helped in its methodology.

The study is a first step at proving what many assert: The domain name system is a good addressing system but a bad directory.  The study took 100 famous brands, 100 randon “plain old” brands, 100 random trading names and 100 college names, and used those names as search terms in the DNS (by affixing .com or .edu to them), or search terms in google or as RealNames (while RealNames was still working).

My scan of the results is that the “dot com” assumption (that ACME can be found at .  ACME.COM) works for famous marks (because by now famous companioes have bought back their names), and works miserably for small companies.  Both Google and RelNames have statistically significant improved accuracy ratings for smaller companies.

I understand this to be a pilot study.  I hope there will be helpful feedback to design a more robust study, the point that debates such as “unrestricted TLDs vs. restricted TLDs” and “key words vs. search engines” can be fueled by more facts and less rhetoric.