I’m back from a week in Cornwall. I’d like to say I was rested, but chasing after an 11-month-old baby in a caravan isn’t all that restful. Still, it was sunny, and good for a change of pace ;)
Via b1ff.org, here’s the Nexis search that US Department of Justice White House liaisons ran on job candidates to determine their political leanings:
[first name of a candidate] and pre/2 [last name of a candidate] w/7 bush or gore or republican! or democrat! or charg! or accus! or criticiz! or blam! or defend! or iran contra or clinton or spotted owl or florida recount or sex! or controvers! or racis! or fraud! or investigat! or bankrupt! or layoff! or downsiz! or PNTR or NAFTA or outsourc! or indict! or enron or kerry or iraq or wmd! or arrest! or intox! or fired or sex! or racis! or intox! or slur! or arrest! or fired or controvers! or abortion! or gay! or homosexual! or gun! or firearm!
This Nexis reference says the “w/n” keyword searches for ‘words .. within 5 or 10 words of each other, Ex: “Enron w/5 investigation”‘.
This is just a smidgen away from the concept of a SpamAssassin-style scoring filter. Crazy stuff.
Best of all, it’s buggy and over-sensitive, according to one librarian: ‘If that is really their search string, they were going through 99% unrelated citations. There need to be a very nested set of parentheses to make the terms work, starting with one after the w/7. Fired and sex are OR’ed twice and need to be nested, at least in the case of Fired and the OR’d terms immeadiately following.’
Update: good Slashdot comment thread here. This comment indicates that the above librarian might be off-base regarding the w/7 parentheses, since the OR operator has higher priority. Here is an even better walkthrough of the query statement logic. Finally, here’s an explanation of the “spotted owl” curiosity…