Let me say publicly that DonBoy’s answer exudes a combination of intuitive genius and confidence that make me think DonBoy is going to do big things in his life. -- Steven D. Levitt (Freakonomics blog)
Thursday, October 13, 2005
Olbermann on the Timing of Terror Warnings: A Skeptical Response

Keith Olbermann devoted a segment a couple of days ago to the recurrent charge that the government's terror warnings are, more often than you'd expect, suspiciously timed to deflect attention at times of otherwise bad political news for the Bush administration. (Video and transcript available via Crooks and Liars; hat tip to Mark Evanier.) As I've said before, I'm not as sold on this as many left-bloggers are, mainly because the fact is that for a couple of years, the Bush administration has had little in the way of good news (aside from that getting-re-elected thing), so it's hard to get a sense of what would be a counterexample to the theory -- although as a matter of fact there is one big fat one that I'll get to below. (And believe me, I'm not claiming "They would never do that". But I'm also not with the group that thinks that saying "Oh, but they would never do that", with heavy sarcasm, makes a point that has no limits.)

Anyway, Olbermann's list continues to fail to convince me that the case is as strong as he makes out, and his faux-self-deprecating bit about "maybe you could also line up the warnings with Walmart openings, I don't know, how ever could we evaluate this claim? It's beyond me" doesn't make me any more believing. The fact is, here's what you'd need to do: go over every interesting political event of the last few years, and mark each one good or bad for the administration, possibly with a rating to indicate its strength in whichever direction. (You'd want to have a panel do this that doesn't know why they're doing it, so to speak; I don't know that the panel has to be itself politically balanced, as long as we can get a consistent rating across the panel of the relative goodness/badness of the events.) Then you mark the "terror events", formal and informal -- note that you have to exercise some judgement as to what counts. (Olbermann has used a mixture of warnings, formal and informal, and arrests, which suggests that the field to choose from might be bigger than is obvious at first glance.) Also, you need to decide how close and/or far from a "trigger" the "terror event" has to be for it to be a hit -- after all, if you end up with half of the calendar being "a good time for a terror alert", then half of anything's going to count as a hit. Then see what you've got. This is not unlike how you'd evalute the performance of a self-proclaimed psychic, or, less contentiously, that of a weather forecaster or an economic forecaster. There's a reason for the contemptuous saying "[Such-and-such economist or leading indicator] has predicted 7 of the last 5 recessions". Ask yourself this, if you're read lefty blogs at all regularly: how often, in the last year, have you read the comment "Sounds like it's about time for another terror alert"? How often has that comment actually been followed by one? (Yes, I don't know either. Fair point.)

(As a side-shot -- Olbermann refers to the logical fallacy called "post hoc, ergo propter hoc" as if its name is "the Logical Fallacy", which doesn't ring a bell with me.)

Here are some specific objections to some of Olbermann's examples:

In Number Three, he lists Powell's speech to the UN (the one with the later-discredited photos of "bio labs", and the prop vial of biochem weaponry), plus anti-war protests, as an instance of a cause. But Powell's speech wasn't a setback for the administration at all -- at the time it was generally considered a success. If you're going to expand the allowable trigger times to "when the administration was trying to sell the Iraq war", that's at least a six-month window, so any "terror event" during that stretch would count as a hit.
(For that matter, you could also reasonably count all of 2004 as "election year".) Not very impressive in the cause-and-effect department.

Number Five on the list is the big fat counterexample I mentioned above:
December 17th, 2003. 9/11 Commission Co-Chair Thomas Kean says the attacks were preventable. The next day, a Federal Appeals Court says the government cannot detain suspected radiation-bomber Jose Padilla indefinitely without charges, and the chief U.S. Weapons inspector in Iraq, Dr. David Kay, who has previously announced he has found no Weapons of Mass Destruction in Iraq, announces he will resign his post.

December 21st, 2003. Three days later, just before Christmas, Homeland Security again raises the threat level to Orange, claiming “credible intelligence” of further plots to crash airliners into U.S. cities. Subsequently, six international flights into this country are cancelled after some passenger names purportedly produce matches on government no-fly lists. The French later identify those matched names: one belongs to an insurance salesman from Wales, another to an elderly Chinese woman, a third to a five-year old boy.
You'd think, though, that it would be useful to recall when happened 3 days earlier, on December 14th: Saddam Hussein was captured from his hidey-hole. That was probably the best thing that happened to Bush between the fall of that statue in April 2003 and Election Day 2004. And they threw it away -- including the suggestion that "we're safer because we did this" -- to counter the mere claim that the attacks were preventable, the resignation of David Kay, and a legal setback in the Padillo case? I really don't buy that.

Number Six:
March 30th, 2004. The new chief weapons inspector in Iraq, Charles Duelfer tells Congress we have still not found any WMD there. On the 31st, after weeks of refusing to appear before the 9/11 Commission, Condoleezza Rice finally relents and agrees to testify. On April 1st: Four Blackwater-USA contractors working in Iraq are murdered, their mutilated bodies dragged through the streets and left on public display in Fallujah. The role of civilian contractors in Iraq is widely questioned.
Rice agrees to testify? That's a huge setback? How about the contractor murders -- that sounds like it's likely to increase support for the war, under the not-very-logical but tempting theory that it shows that we're fighting against awful people who would do awful things. (Or, the theory of "pre-taliation".) It doesn't make all the sense in the world, but it was (and continues to be) a common pro-war talking point. Note also that we're counting "Rice agrees to testify" as a trigger, but not "Bush and Cheney agree to testify", or either of the dates of the actual testimony. This is my core point again -- too many possible triggers leads to too many "hits".

In Number Ten, I think he may have cause and effect mangled, and missed a good opportunity, by his own standards:
Last Thursday. At 10 AM Eastern Time, the President addresses the National Endowment for Democracy, once again emphasizing the importance of the war on terror and insisting his government has broken up at least 10 terrorist plots since 9/11.

At 3 PM Eastern Time, five hours after the President’s speech has begun, the Associated Press reports that Karl Rove will testify again to the CIA Leak Grand Jury, and that Special Prosecutor Fitzgerald has told Rove he cannot guarantee that he will not be indicted.

At 5:17 PM Eastern Time, seven hours after the President’s speech has begun, New York officials disclose a bomb threat to the city’s subway system - based on information supplied by the Federal Government. A Homeland Security spokesman says the intelligence upon which the disclosure is based is “of doubtful credibility.” And it later proves that New York City had known of the threat for at least three days, and had increased police presence in the subways long before making the announcement at that particular time. Local New York television station, WNBC, reports it had the story of the threat days in advance, but was asked by "high ranking federal officials" in New York and Washington to hold off its story.
If I were making this list, surely I'd include the DeLay indictments as a trigger event; the Rove testimony is one of a series of events that gladdens the hearts of those following the Plame case, but I don't think resonate (yet?) with the general public. (Another reminder that you have to count all the bad events.) And the terror speech should of course be an effect, not a cause. Maybe Olbermann means that, but it's not apparent from the writeup. Note also that it seems that it was NYC, not the Feds, who made the most of this, although I should also mention that there are those, such as Steve Gilliard, who thus hold that this is a perfect example, but of malfeasance of on the part of Michael Bloomberg, currently running for re-election. Also, it may be relevant, as I pointed out earlier, that this event is responsible for probably the only not-completely-made-up terror warning email to be found at

In Number Twelve, it's the terror event that I think is stretching it, rather than the trigger:
May 11th, 2005. Later that day, an instructor and student pilot violate restricted airspace in Washington D.C. It is an event that happens hundreds of times a year, but this time the plane gets to within three miles of the White House. The Capitol is evacuated; Vice President Cheney, the First Lady, and Nancy Reagan are all rushed to secure locations. The President, biking through woods, is not immediately notified.
The first that most people heard of this, the day it happened, was "false alarm at the Capitol". That's not going to terrify anyone. And especially so if it really is unusual that the plane made it so close.

Number Thirteen:
June 26th, 2005. A Gallup poll suggests that 61 percent of the American public believes the President does not have a plan in Iraq...

June 29th 2005. The next day, another private pilot veers into restricted airspace, the Capitol is again evacuated, and this time, so is the President.
Now we're counting bad polls? And the terror event is another immediately-obvious false alarm. Sorry, no.

Finally, here's a final counterexample: on the eve of Bush's 2005 Inauguration, there was a claim that terrorists with "nuclear oxide" were headed to Boston. Governor Mitt Romney left DC and came back to Massachusetts, spending the night "in the state" to show his confidence in the countermeasures, or something. Since "in the state" turned out to be "in the bunker in Framingham, 20 miles from Boston", this was most definitely not reassuring. His inaugural sounds like a moment when an adminstration would not want to remind the country of how unsafe it is, no?

Seriously: run the test as I described it above, with all events considered, and see what you get. You might get something. You might even be able to quantify it. But this particular list isn't doing it for me.

UPDATE: I note that there's a dispute as to originality of this list.

Powered by Blogger Weblog Commenting by
free website counter