Battiest story of the (no longer young) year: Nominations open!

This next tale doesn't unlock the entire secret inner chamber of media cluelessness. But it does shed light on lots of the different things that can and do get out of tune between the first spark of an idea and the final product that lands in a puddle near your doorstep.

Some particularly amusing (and sharp) questions have been raised of late about how errors happen and when and how they get fixed (or not). One of the entertaining things about this one is that almost nothing in it is correctably "wrong." Generally a correction requires a story that says X, whereas 1/X or Y or some other not-X is actually the case: We said the mayor was indicted on three counts of second-degree bestiality, but the court record shows it was two counts of third-degree bestiality. Sorry!

Conceptual errors, meaningless constructs cooked up for the occasion, statistical fabrications, and blithe ignorance of context, on the other hand, don't generally get corrected because they aren't provably "wrong" in the X vs. not-X sense. Hence, corrections tend to look like this:

In a brief report, Cows moo with an accent etc, page 10, August 23, we quoted John Wells, professor of phonetics at the University of London, as saying, "This phenomena is well attested in birds." He assures us that what he said was the correct: "This phenomenon is well attested in birds."

Given the chance, Prof. Wells would probably rather have assured the Guardian (and Press Association and the Beeb) that however well the phenomenon is attested in birds, cows do not moo with a freaking accent. He puts it a bit more politely:

I told them that I thought it was highly unlikely, but that there had been serious research showing that various species of bird exhibit geographical variation in their calls. And if birds and human beings have local accents, you can’t entirely rule out that cows might too.

So the surface error is corrected -- yes, professor knows singular from plural -- but the underlying PR fabrication, which relies on ignoring the context of the bird-call comment, goes unchallenged.

Anyway, on to today's specimen. It started life as the No. 4 entry on the "news digest," AP's capsule description of the top current and upcoming stories, moving about 1:45p Friday for Saturday ayems. The AP sent half a dozen takes between about 4 and 10p Eastern time, but that's not what readers see. This is news in its natural setting: An edited version, and it looks like a pretty early one, from a major regional daily (it'll take a few days for Lexis-Nexis to get a handle on how many papers ran it, but that will be worth tracking too).

U.S. war deaths now equal 9-11 toll
2,973 dead in conflicts initiated as an answer to people killed here

WASHINGTON -- Now, the death toll is 9-11 times two.

U.S. military deaths from Iraq and Afghanistan now match those of the most devastating terrorist attack in America's history, the trigger for what came next.

See a couple of pretty massive assumptions building up here? One, that deaths in war vs. deaths in the event that precipitated war is a valid measure of something, and two, that these are somehow all the same war. That's not falsifiable, in the sense that opinions aren't falsifiable, but it's a pretty strong ideological marker for a news agency to be laying down. (Reminder to editors: A blunder might be the AP's (or McClatchy Washburo's) fault, but when you hit the "send" button and the plate is cut and the press rolls, you own it too.)

The latest milestone for a country at war comes without commemoration.

It also might come without knowing exactly who is the 2,973rd man or woman of arms to die in conflict in Iraq and Afghanistan, or just when it happens. The terrorist attacks killed 2,973 victims in New York, Washington and Pennsylvania.

Whoa. We know it's happened, but we don't know when it's going to happen? [Matters get worse a hair after 10p, when the 5th lede -- broadly, the fifth revision, or sixth version, of the story --notes that the milestone has been passed: It came without the precision of knowing who was the 2,974th to die in conflict. ... The Pentagon reported Friday the latest death from Iraq, an as-yet unidentified soldier killed a day earlier after his vehicle was hit by a roadside bombing in eastern Baghdad. It sounds, in short, as if we know exactly who and where, even if the name hasn't been released yet.]

This is not the first time that a war that was started to answer death has resulted in at least as much death for the country that was attacked first.

AP original: "Not for the first time, war that was started to answer death has resulted in at least as much death for the country that was first attacked, quite apart from the higher numbers of enemy and civilians killed." So the AP syntax has been de-purpled a little, but has any meaningful editing been done? Let's see.

Once your head stops spinning, unpack the construct a little. Since we're mentioning it, we obviously think it's a valid measure of something, but what exactly makes this variable important: War that's (1) "started to answer death" results in (2) "as much death" for (3) "the country that was attacked first"?

So the US civil war is right out (no Union KIA at Fort Sumter), since it wasn't started to "answer death"? How about that pesky World War I? It must have crossed the line with the first two Austro-Hungarian combat deaths (equaling Franz Ferd and his morganatic frau). Do any wars that meet condition 1 not also meet 2 and 3? Or are colorless green ideas asleep furiously at the switch again?

The body count from World War II was far higher for Allied troops than for the crushed Axis. Americans lost more military personnel in a succession of Pacific battles than the 2,390 people who died at Pearl Harbor -- the attack that made the U.S. declare war on Japan.
The U.S. lost 405,399 in the theaters of World War II.

First, somebody should remind the AP that World War II was well under way for some years before Pearl Harbor. And yes, the Allies had more combatant deaths than the Axis (and about twice as many troops mobilized). But our writer's playing apples and oranges. Take out Soviet KIAs -- Moscow wasn't a part of the US war with Japan, after all -- and the picture's quite different. Japan had about five times as many battle deaths as the United States (1.5 million to 292,000, per Dupuy and Dupuy; AP appears to be counting both battle and nonbattle deaths among troops) and three times as many as China. Germany had less than half the Soviet total ("proving" the point?) but more than three times the combined US-British-French total (guess not).

We've got a lot of numbers thrown solemnly about here, but not the faintest idea of what they mean or what points they do or don't support. That lends a certain poignancy to the graf (deleted here) that the AP uses to set up the following quote. "Now this," the AP intones. To which the reader is justified in replying: "Now what?"

"There's never a good war, but if the war's going well and the overall mission remains powerful, these numbers are not what people are focusing on," said Julian Zelizer, a political historian at Boston University. "If this becomes the subject, then something's gone wrong."

He has a point, but -- as with Prof. Wells above -- it's worth wondering whether he and the AP are talking about the same thing. Are "these numbers" the ongoing cost in US lives or the spurious correlation between the civilian cost of a terrorist attack and the cost in military deaths of a related war and one whose stated purpose (depending on which way the wind was blowing) was rather, um, different?

Concluding sermon: Deaths in these wars are worth our notice (probably moreso than one would judge from their play in the press generally). Made-up statistics aren't. Surely there was some news somewhere from Iraq or Afghanistan that could have used this space to actually further public understanding.