"Count what is countable. Measure what is measureable. What is not measureable, make measureable." -- Galileo

Monday, October 22, 2007

Thoughts from Today's New York Times

Today's New York Times has a piece about methodologies (registration required) for counting site hits as it pertains to the online advertising business. At issue is the discrepancy between server log numbers and sampling panels. In the end,
“The irony is we’ve always called for more measurement,” said Stephen Kim, director for global trade marketing at Microsoft Digital Advertising Solutions. “Now we’re getting it, but many people are somewhat frozen in how to deal with having more measurement.”

Also on the NYT website today was the article on online libraries. Turns out many libraries are turning down offers by Microsoft and Google to digitize their collections and instead going with the Open Content Alliance, a nonprofit effort aimed at making their materials broadly available.

Finally, as an aside to the difficulties of measuring social and economic phenomena, NYT discussed the recent report from the Head of Reconstruction Teams in Iraq. One conclusion was that "the program still has not developed concrete methods to measure the effects of the teams on progress in the country."

What's all this to do with Plone metrics? First, that even with exact server data, we don't know what people are actually doing and sampled panels don't necessarily do a better job. What does it mean if someone at Los Alamos National Laboratory downloads a Plone installer? What does it mean when I have 15 students at my College of Santa Fe class download the installer? Apples and oranges again.

Secondly, the digital libraries observations point out that guaranteed open source is a significant component to IT decisions these days. The Documentum/Alfresco case study will be a fascinating Master's Thesis someday for someone studying the early history of the Internet and the relationship between COS and open source.

Finally, the report from Iraq speaks to the difficulty of measuring effectiveness. This takes me back to my early posts about Bullock's dissertation on integrated measures of effectiveness. I'll continue to pursue that as I slowly fill in the matrix with data and normalize the data.

No comments: