May 25, 2004

Webhits on Google per gigapage: A replacement proposal

Most spelling reform proposals are totally unsuccessful, and some are disastrously so. I now want to withdraw mine, making it perhaps the most short-lived and unsuccessful in the history of the universe. Keith Ivey points out to me that under the standard conventions for physicists published here by the Physics Laboratory of the National Institute for Standards and Technology (NIST), unit names like the hertz, joule, pascal, and so on are always lower-case despite being named for persons, though with abbreviations "the symbol or the first letter of the symbol is an upper-case letter when the name of the unit is derived from the name of a person": it's Pa for pascals, Wb for webers, Hz for hertz, and so on. That would suggest "ghit" for web hits using the Google&tm; search engine, but "Gh" for the abbreviation. (The reason I suggested g-hit as the pronunciation is that "git", with a velar stop [g], is a word in British English meaning stupid or obnoxious person. But it won't matter any more; read on.)

I now realize that I am not content with any of this. It now occurs to me that the best analogy for Google hits as a measurement term is not hertz or joules or pascals, but degrees Celsius. Degrees are the units, Celsius is a specific scale of measurement. Words like Celsius are capitalized because they are names of people. Google should be capitalized because it is a corporation name But hits, like degrees, are not named for a person or corporation. According to NIST ‘the correct spelling of the name of the unit °C is "degree Celsius" (the unit "degree" begins with a lower case "d" and the modifier "Celsius" begins with an upper-case "C" because it is the name of a person.’ However, I'm not done yet. Read on.

Putting this insight (hits are analogous to degrees, Google is analogous to the Celsius scale) together with Phil Resnik's suggestion that we really want to count web hits relative to a given search engine, and Mark Liberman's scientific refinement pointing out that we really want to measure web hits per billion documents, we get to the following proposal. The basic unit should be the webhit, abbreviation wh. Webhits measured using Google are Google webhits, abbreviation whG; webhits measured using AltaVista are AltaVista webhits, abbreviation whA; and so on. Web hits per billion pages, i.e., per gigapage (Gp -- David Nash points out to me that giga- is conventionally abbreviated G-, not g-), will be a unit obtained by division, and the NIST standard would be to say webhits per gigapage when spelled out in full but wh/Gp when using the abbreviations. Measurements of webhits on Google per gigapage would be expressed in whG/Gp, and so on.

So that's my current proposal: measuring currency of (sets of) strings on the web by whG/Gp. Though we probably haven't heard the last of this thread. We may need a national standards committee to make recommendations to a central council standing executive committee which will then report to the membership at large through delegates to a national congress... yawn... I never wanted to become a national standards administrator. I'm a grammarian. Grammar is fun. Grammar is exciting.

Posted by Geoffrey K. Pullum at May 25, 2004 04:48 PM