Asked and counted

After reviewing a mere 2056 queries hitting older zdnet blogs on my own site during October I can say with no certainty at all that google users have much bigger vocabularies, and ask less structured questions, than either MSN or Yahoo users.

Back on September 30th I put my old 2005 and 2006 zdnet blogs on my own website too and, yesterday, spent some time looking at the resulting apache logs to see if there's a clear pattern in October search hits from google, yahoo, and msn.

The October sample is relatively small in total (2056) with almost all query hits (92%) coming from google. As a result I'll be revisiting the data after Christmas to see how my preliminary results hold up.

At the top of that list is an oddity: there were only 47 hits from MSN, all of them from Windows users with ten (21%) of them consisting of more or less complete sentences - and, of those six were, like the example below, suspiciously evocative of less than knowledgeable school assignments:


Of 113 Yahoo searches only two obviously followed this pattern:


But my favorite in this genre came from a google user:


Some queries simply don't make sense to me (numbers are frequency counts):

1 against+the+use+of+english+government
1 national+Unix+Windows+pencil

And some seem a bit sad:

1 failing+Linux+install+netvista
1 acpi+failure+intel+toshiba+solaris

In total the search terms used on google matched 521 unique words (excluding plurals) vs. 73 for MSN and 128 for Yahoo (148 combined). The most commonly used words, overall, were:

19 minix
31 and
38 comparison
57 between
66 solaris
75 of
135 linux
197 versus (50 versus; 50 vs.; 97 vs)
270 windows
291 unix

And, finally, although there was nothing obviously odd about the browser and OS usage here there was something suggestive about this with respect to another data set. Some past blogs contain references to images that are only stored on my site and these produce page hits that look like this: - - [31/Oct/2007:19:03:35 -0600] "GET /images/sunray170_dual.jpg HTTP/1.1" 200 129663 "" "Mozilla/5.0 (X11; U; Linux i686; en-US; rv: Gecko/20070723 Iceweasel Firefox/2.0 (Debian-"

There were 584 such hits during October but only 73 of them (12.5%) came from Linux users - about par, except for the surprise: of 423 hits from Windows users, 35% (146) used Firefox. Overall Firefox users accounted for nearly 38% of October hits on my site, but for only 29% of Windows users. What this number suggests, therefore, is that the old blog readership on zdnet leans a bit toward comparison shoppers moving cautiously out of Microsoft's monopoly world.

From all of which I conclude first that if I want my stuff to be popular I should have more A vs. B comparisons - and, (getting off on the right foot, you see :) ) that google users have, on average, more technical interests and bigger vocabularies than msn/yahoo users.