- by Paul Murphy, -
On most workday mornings my email contains a newsletter promising "unbiased opinion" --an editorial oxymoron whose intent is probably to deny the presence of commercial advocacy or personal bigotry in the materials presented.
We tend to think of bias as necessarily negative, but in fact a bias is just a predisposition to believe or disbelieve new information about someone or something. Positive bias is rarely recognised: if something we read is consistent with our own beliefs we don't typically see it as biased even if those who hold contrary opinions do.
I have a friend, for example, whose ardent embrace of all things Microsoft has been completely undampened by fifteen years of lost files, forced upgrades, untimely crashes, and DLL hell. To him, my enthusiasm for MacOS and Unix is bigotry while I consider his inability to value experience over peer pressure a perfect illustration of bias in operation.
Ultimately this is a question of values: he values the social lubrication that comes from shared pain over technology; I don't. Unfortunately this spills over into how we respond to opportunities to gain new information or insights by making us search out and accept information consistent with our views while rejecting, often on the basis of source rather than content, the reliability and interest of contrary information.
If, for example, we both saw an article starting with the words:
"Our next generation OS will be the most secure ever released for commercial use", said
whether either of us would continue reading would depend on whether the speaker represented Microsoft or the BSD developer community. His decision would be wrong, of course; after all Longhorn is what you'd expect from an attempt re-invent PICK by hacking the JET drivers on the basis of a BeOS brochure, but that's not the point. The point is that it's impossible to be sure that what you choose to know hasn't been heavily influenced by pre-existing biases.
In other words: "I am what I read" has the corolary that "I read what I am" and correspondingly that neither you nor I can ever really know if our opinions are predominantly attributable to a selection bias in what we read or not.
I got my first Sun workstation, a 3/160 with a 20" screen running SunView, in 1985 and have upgraded to newer Sun gear roughly every four to five years since. It's fair to suspect, therefore, that I might be a little biased in Sun's favor and correspondingly upset when an anonymous AP hatchet job starting with the words "Troubled computer maker Sun MicroSystems.." and ending with the usual nonsense about Solaris/SPARC being unable to compete with "industry standard" servers using Linux appeared in the September 30/03 Toronto Globe and Mail.
MCI is "troubled" - they're in bankruptcy; Tyco is "troubled" - their former CEO is charged with stealing $600 million from the shareholders. Martha Stewart is "troubled" --but Sun just made an extremely conservative accounting change having no effect on the business, their cash position, or their earnings potential. So why did this particular writer unleash such highly prejudicial language without troubling itself with facts? was it bias? incompetence? advocacy? market manipulation? or just plain malice? I wish I knew, but in truth I'm not even sure how much of my own angry response reflects bias rather than knowledge.
On the other hand I'm quite sure the people who send me flamemail every time I comment on the disgraceful cost and performance offered by IBM's mainframe Linux have it wrong. In my unbiased opinion people who use company gear, and company time, to send me stuff like:
Paul,
I researched your article. I've determined that you made up the research
and nothing in the article has any truth. I couldn't find any of the
intermediary evidence you suggest. It was a slam dunk to find the lies.
Did you have a motive for writing this article?
Randy
and no less than eight follow-up messages, have personality problems that go well beyond bias.
Or then again, maybe not; after all, if I had that seriously troubled AP writer's email address...