News Gloss

It's obvious, but is it visible?

Everytime a US politician speaks of the "Dangers of nuclear proliferation", some country like South Korea, Iran or China is mentioned. But the US is the only country that has ever used nuclear weapons. Its publicly available strategy documents point to the federal government's nuclear array as a key component to "promoting US interests around the globe". So, obviously the US is the most dangerous nuclear power, and always has been. Nuclear proliferation mostly endangers one of the major power strategies of the US.

Obvious ... but visible? How would you figure this out when hundreds of times a day, news organizations speak of "rogue nuclear states"? How dare those rogues threaten the monopoly of nuclear death held by the Federal government.

Again, how can we figure this out from text?

I was reading the John Perkins book "confession of an economic hit man", when he wrote of the death of Ecuadoran populist president Jaime Roldos, in 1981. Roldos had stood against profit by US-run oil companies in his country, and the political battle had become critically heated. Then his helicopter crashed, under circumstances so obvious that headlines around the world declared that he'd been "eliminated" by the US. But, inside the US, the press merely reported Reagan's condolences to Ecuador. Reagan's team clearly was responsible for Roldos' death, something that is much clearer 25 years later ...

How can a computer figure this out? First, it would need the text of all of the world's newspapers, going back as far as possible. It's pretty clear that we could pinpoint differences between stories which the world found important, and which the US press did not. This is a generally useful analysis -- the press will cover a story about, say, a newspaper strike, which is not visible in the newspaper itself. Or a corporate magazine might not mention a major scandal which the whole world is aware of. There must be a certain structure to these kinds of blindspots. That structure has not yet been investigated. The analytic tools have not begun to be developed for this kind of work.

posted by Greg Bryant @ 1:34 PM