It is not a question of never knowing. It is a question of whether the fact that we did not find evidence of WMDs after we invaded was evidence that there were no WMDs before we invaded... What Rumsfeld is saying is that the fact that we did not find WMDs after we invaded is no reason to think that our invasion was not justified, since there might have been WMDs which were gotten rid of when we invaded.
This isn't an epistemological exercise, though. This is about building a case. I can accept that absence of evidence just is what it is, I mean I operate under the same mentality in medicine and science.
But the pre-invasion evidence was crappy and based in large measure on that crazy Iraqi informant "Curveball"; very prominent inspectors like Joseph Wilson and Scott Ritter had argued pre-invasion that the WMD programs had been disbanded years before and that this whole uranium entreaty to Niger was bogus -- and when you combine this with the absence of evidence within Iraq, you build a good case that they were not there in recent years before the invasion either. In fact this case outweighs the contrary case.
In the unicorn case, of course finding no evidence of unicorns is evidence there are no unicorns, just as if we were in India, finding no evidence of elephants would be evidence that there were no elephants. The question is whether finding no evidence in both cases is evidence that there are no unicorns (in the one case) and no elephants in the other case....What is difference?
The difference is that in the case of elephants their existence is corroborated by other evidence, whereas this is not the case with unicorns.
Is the fact that there is no evidence of Santa Claus (sufficient) evidence that there is no Santa?
Asking whether it's evidence is different than whether it is sufficient
evidence. I've been making the case that absence of evidence for WMDs IS INDEED evidence that they did not exist. But it is not sufficient in that it doesn't disprove it beyond all doubt.
Think of this in terms of diagnostic tests and their statistical basis. The most useful idea here isnegative predictive value (NPV), which is:
(true negatives) / (true negatives + false negatives)
In other words, a test with a NPV of 100% means that every negative test is a true negative. A test with a NPV of 90% means that only 9 of every 10 negative tests is a true negative.
Looking for WMDs on the ground in Iraq is not a perfect test -- the NPV cannot be 100% ever
unless you search every possible square inch above and below ground of the entire country, in the walls, in the caves, everywhere -- and you exclude the possibilities of exportation or destruction. In other words, there are a lot of reasons why this test might be negative.
But then you need to ask yourself is this likely to be a true negative or not. Which is more plausible -- that Saddam had the wherewithal and resources to quickly destroy a whole WMD program, its weapons, and its resources beyond any trace? That he would rather destroy such weapons than using them against an invading army? Or that the evidence suggesting that they existed had a low positive
predictive value and it was a false positive??