Read other letters about this article
It's important because, if we are going to talk context (proximity to September 11) for the comment about prisoners and the Geneva Conventions, it matters when the other statement about government abuses was, relative to the April 2004 watershed event: the broadcast of the Abu Ghraib pictures on April 28, 2004. People may not like to compare the two events, but I would wager both changed many people's thinking just as much.
As for jschultz's comments, there are no people during war who are not entitled to protection under Geneva. None at all. There are differences in the protections offered. jschultz mentioned spies and saboteurs, the only privilege they are denied that POWs are granted is the privilege of sending letters through the ICRC/IFRC.
Actually, despite the best efforts of Douglas Feith and the Reagan administration, we are still signatories of the 1st and 2nd 1977 additional protocols to the Geneva Conventions, although we have not ratified them (Douglas Feith was the primary lobbyist in getting Reagan to send them to Congress with a recommendation against ratification, it's a Israel/Palestine thing). Under the Vienna Convention, we are therefore beholden to obey them, even though we are not required to apprehend offenders. They detail the rights of prisoners in conflicts that are not between two nation states. They are almost the same as those of POWs, something Feith is vehemently allergic to. It is also not true that POWs and other prisoners cannot be interrogated. They can't be coerced or abused.
I, for one, do not forgive Eric Holder his opinions of January 2002, although I do believe it is one item in a mix and will look at the whole mix, especially his recent statements and what he says at confirmation. I was an early opponent to the idea that we needed to immediately decide (as was pushed by the media in the immediate, still in September, aftermath) on how to re-balance civil liberties and security, and therefore also an opponent of the PATRIOT Act. I was pretty vocal about it (although I didn't comment on blogs or usenet about such things then, I wrote perhaps a few hundred letters to the editor between 2000 and 2006 or so, almost never published, the NYT doesn't like my writing much).
Among other things, I was boiling over about it by January 2002, because I had spent days tracking down an article (not)published by a grad student because the IEEE had removed it from their conference proceedings, because they felt it was disrespectful of the President so soon after September 11 (the paper was delivered in conference on September 11). I said as much at a conference, and several people pulled me aside to give me "a word of advice" not to say such things, "one had to be careful." My arguments at the time were that setting the two conditions up in such a way presented a class of optimizations known well in signal processing as "rate-distortion curves". One has to know that one is getting all the civil liberties one can for the current amount of security before one can postulate that any change in one must change the other.
That isn't a legal or political viewpoint, but if logic is the underpinnings of such viewpoints, perhaps it should have been. It is provable, given some pretty common linearity assumptions.