My dear faithful reader-fans who have survived my diverse ramblings since I first rambled way back in Computer Weekly circa 1956 will know that one of my most piquant threads has been of the psephological persuasion. As you know, psephology is the formal study of elections, apparently trivial but dripping with deep, dark paradoxes. There are endless papers in the MAA and AMS MathLit searching for the “perfect” electoral methodology. When voters are asked to record their “preferences”, it seems that the “normal” logical/algebraic relational laws break down. Thus, the transitive “A prefers B and B prefers C implies A prefers C” often fails when the total votes are counted (see Referecnes).
Or re-re-counted, to refer to the current (as I write in November 2000) Florida Presidential fiasco that will decide “the lesser of two weasels” as Jay Leno stated. My Brazilian correspondent, Rainer Brockerhoff, is alarmed/amused by the fact that the leading “wired” nation (aka “Fellow 'Mericans”) requires “hand counts” while the billions in Rio and São Paulo are “cast” electronically.
Apart from the quirks (popular vs. electoral college) of choosing the leader of the free-world, we are subjected to and influenced by statistically dubious polls daily.
I helped install/program IBM for George Gallup in the 1960s. The in-joke was the over loaded quiz: “Given the rampant inflation and unemployment, would you vote again for that Commie bastard Harold Wilson?”
Evans Data and similar “opinion” gatherers are more subtle, yet we need to be cautious. What if 39.78% (note the spurious accuracy) of developers say they are “scale 6 of 10” interested in “looking at C# when it's defined”?
Finally, as my last “Last Word”, I bid a fond Abschied.
I'll continue to promote the Linux agenda in my other columns.