Thursday, May 9, 2013

Google 'irony'

For a long while it was only a publication devoutly to be wished, but finally The Troubled Path of the Lock-in Movement is now available for download courtesy of SSRN--just register, and have your way with them!

Troubled Path quickly relates this ominous note;
Now, more than a decade after the settlement in the Microsoft [anti-trust] case, arguments based on network effects are again being put forward as a basis for antitrust action or other industrial policy. Google has been a prominent target.... A paper appearing recently in this journal... “Search Engine Competition with Network Externalities,” observes that search providers use information about past searches to improve subsequent ones, and asserts that the practice creates a network externality.... From that assertion, the authors derive theoretical results and advance this policy proposal: “All search engines should be required to share their (anonymized) data on clicking behavior of users following previous search queries.”...we note that the paper follows the pattern of much of the lock-in literature: A network effect is posited and argued to give an incumbent product an advantage that is sufficient to lock it in, even where it is inferior to an available alternative.
Though the authors, U of Texas, Dallas's Stan Liebowitz and NC State's Steve Margolis, don't mention it, they are well aware that Google's chief economist today is Hal Varian, former Berkeley professor, textbook author, and co-author with Carl Shapiro of the best seller Information Rules.  And in that otherwise useful tome there is a very badly handled explanation of the debate, still current, over path dependence by lock-in due to network effects--i.e network externalities as the paper cited in Troubled Path has it.

In Troubled Path, Liebowitz and Margolis relate the sad (really, scholarly malpractice) history matters that have followed in the wake of the work of then Stanford economists Brian Arthur and Paul David these last two and a half decades.  A story of denial, double-talk, and duplicity in service of attempting to rescue Arthur's and David's story from the 1980s (a story so good it had to be true) of market failure.  Popularly dubbed The Economics of QWERTY.

The details of that intellectual scandal will have to wait for another day.  Instead we're now going to show something of Hal Varian's small role in promoting a false history that threatens to rebound to his new employer's (and his) detriment. In Stan Liebowitz's 2002 book Re-Thinking the Network Economy there is this mild criticism of Information Rules:
Shapiro and Varian...present the QWERTY example on pages 185 and 186.  After presenting Paul David's strong lock-in story as if it were true, they mention in the last sentence or two that something appears wrong with the story because computer keyboards are so easily reprogrammed that strong-form lock-in would seem to be impossible.  Yet on page 233 they are back stating that QWERTY is an inferior design. They do try to cover themselves by citing in a footnote the Liebowitz and Margolis 1990 presentation of the typewriter history, but it appears to be more an example of covering themselves against claims that they missed a relevant article than attempting to present full information for their readers.
Long story made short:  When that criticism was noted in a review of the Liebowitz book on the usenet discussion site sci.econ (around Thanksgiving 2002) Hal Varian arrived to defend himself with;
 ...we did not cite the L&M piece in a footnote; it was in "further reading", where we cited it in parallel with the David piece. 

The quibble between 'in a footnote' and 'in further reading' apparently being important to Prof. Varian, but it wasn't exactly dispositive to the issue of whether or not he realized that Paul David's story of the history of the typewriter keyboard was false.  

Over the next two weeks the debate raged, Varian refusing to concede any error in his (and Shapiro's) presentation of the facts.  Apparently that frustrated Santa Clara's David Friedman who confronted Varian pointedly and repeatedly over his failure to be responsive to Stan Liebowitz's main argument;
I haven't read your book, have read "Fable of the Keys," and don't see how this answers Stan's point. If you repeated historical stories which they show to be false, without telling your reader that they were false, or at least controversial, then you were at fault--you were, presumably unintentionally, misleading your readers. Saying that QWERTY can't be as much is claimed doesn't deny those particular stories. 
In Information Rules there was this statement; [QWERTY] offers a fascinating example of collective switching costs and the difficulties of coordinating a move to a superior technology. Friedman asked Varian;
 How do you know that the collective switching costs earlier were large enough to provide a "fascinating example?" It can't be because of the failure to switch, since as you point out we now know that people didn't switch even when it became easy [when computer keyboards came along with their easy switch to Dvorak].
....So if you didn't swallow the Dvorak myth, and if you did accept the arguments made in the L&M article, what was your basis for the claim that QWERTY "offers a fascinating example of collective switching costs and the difficulties of coordinating a move to a superior technology," before the conversion to computers whose keyboards could be reprogrammed?
When Varian then asserted;  I thought that was what we said: since nothing changed when the coordination costs were removed (so far at least) which suggests Dvorak can't really be significantly better than QWERTY, Friedman went at him again; 
How can the Qwerty/Dvorak story be a fascinating example of "the difficulties of coordinating a move to a superior technology," as you apparently said it was, if Dvorak isn't significantly better than Qwerty?
So what was the "superior technology" you were talking about? Wouldn't the correct statement be more like "The Qwerty/Dvorak story provides a fascinating example of the persuasiveness of bad arguments based on careless historical research," since that now seems to be your position? 
And on and on and on and on it went.  Varian never giving a straight answer, no matter how many times he was confronted with his own (and his co-author's) words.  At least he was genial, which we can't say for the people he was indirectly defending, Brian Arthur and Paul David.  Now, more than a decade later, it's Varian's business at risk from the same faulty idea that harpooned Microsoft.  History matters.

No comments:

Post a Comment