November 14, 2002



Professor Marc Herold consists his current minimum total of 3,141 Afghan civilians killed thus far to be "authoritative and unrefuted." Last post, we looked at the 76 certain fatalities he still has confidence in in the July-November, 2002 period. Only 34 stood up to even moderate scrutiny. Now let's look at the three month period just before that.

Herold's database counts exactly 50 civilian fatalities between April and June of 2002 in Afghanistan, in ten separate incidents. They are:
1) 29 April: 4 armed tribesmen "executed" by Australian SAS troops
2) 6 May: failed hit by a CIA Predator on warlord Gulbuddin Hekmatyar results in 10 madrassa students and/or bystanders killed by mistake
3) 8 May: a madrassa director shot by Special Forces in Pakistan
4) 12 May: 5 alleged Taliban killed by Special Forces
5) 13 May: the same 5 Taliban killed a second time (a clear example of double counting: Herold even acknowledges they're likely the same incident in his notes)
6) 17 May: 10 killed in a wedding party in Paktia after drawing aerial fire with celebrations
7) 23 May: 3 killed in a Special Forces raid... a 100 year old man, a 70 year old man (likely the same man), and a 6 year old girl who falls down a well.
8) 26 May: unexploded ordnance kills 8 in Pakistan
9) 1 Jun: 3 looters killed by Special Forces
10) 2 Jun: man who points weapon killed by Special Forces

Okay, #5 is rejectable immediately. No. 3 and no. 8 didn't take place in Afghanistan. Nos. 1, 4, 9 and 10 are immediately questionable as possibly not being "innocent" civilians. Nor, one suspects, can the U.S. be blamed for a well death, or the death of the same senior citizen twice on 23 May. So that's 29 of 50 that are rejectable on their face. What's left?

Well, it seems certain one elderly Afghan gentleman was killed in that May 23 Special Forces raid (#7). The Americans claim they were being shot at, but given the man's age it is reasonable to presume this was an unintended consequence.

There is, however, reasonable doubt about the CIA hit (#2). Herold doesn't mention it, but Newsweek covered the same incident, and rather than the 10 fatalities (children in a madrassa that was hit by mistake) he reports as his provable minimum, they came up with a rather lower number in their June 12 issue: zero, in fact.

"...managed only to injure a few local farmers who were tending crops at the scene."

Now, Newsweek's reporter on the spot could of course be wrong, and Herold's other sources, which in this case include, interestingly enough, both Ted Rall and Eric "Wormtongue" Margolis, neither of which were in country at the time, could be right. But it's fair to say that that figure of 10 is still contestable, and hence should by rights be excluded from Herold's hard minimum estimate until more data is available.

Actually, no, let's look at what they said, while we're at it. Wormtongue, interestingly, says of the incident only that a "number of [Hekmatyar's] companions" were killed, not 10 schoolchildren... it's uncertain which source Herold is drawing his figure from. Rall, who calls Hekmatyar "evil... sort of an Afghan John McCain" says "the missile hit a group of cars that had nothing to do with Hekmatyar," again not madrassa schoolchildren. Schoolchildren, farmers, companions, people in cars... it's fair to say there's enough doubt about this incident to take this number right off the chart. (It should be remembered also that Hekmatyar, despite Wormtongue's fondness for him, is by all impartial accounts a psychotic mass murderer and the biggest non-Taliban threat to Afghan stability... if it had been successful, his assassination would have undoubtedly saved far more lives than were lost, if any in fact were.)

Okay, so that just leaves the 10 killed at the wedding (#6). How reliable is that? Well, here's the quote by the N.Y. Times' David Rohde that Herold cites, as one of five sources for this one.

An Australian reconnaissance team managed to insert itself between two warring Afghan tribes here without either one noticing. But when the team was discovered, a firefight erupted with suspected Taliban and Qaeda fighters that led to an American air attack that killed nine Afghans. Officials in Khost, the province capital, said the strike focused on the wrong gunmen, killing pro-American fighters. Other Afghan officials said the plane attacked after villagers fired in the air during a wedding celebration.

So, again, there seems to be reasonable doubt about the circumstances on this one. It could be wedding guests... or it could be "pro-American fighters" killed by mistake. The first would fit in Herold's list. The second would not. Based on this quote alone, there's no conclusive evidence one way or the other. That's reasonable doubt... and another 10 stricken off Herold's absolute minimum number of the civilian Afghan fatalities due to U.S. military action.

So out of 50 listed fatalities in this period, judged by Herold to be absolutely certain on the evidence, it is possible to independently confirm... one. The old guy in the village. Personally, I don't doubt the number is higher than that. But Herold's evidence only even comes close to meeting a reasonable standard of proof in that one, singular case.

To date, we have looked at 126 of Herold's alleged deaths. We have established that there is considerably more than reasonable doubt about all but 35 of them... a 28 per cent success rate for the New Hampshire women's studies professor. The number of confirmed civilian dead in Afghanistan continues to drop, from 3,141 to 3,050, and as will be demonstrated shortly, has a lot farther to fall yet. Next time: we look at the January-March period.

UPDATE: I'm trying to focus on the bigger numbers here. If you think for a minute I'm being too dismissive of Herold's other smaller claims, here, for instance, is the entire evidence he has for another one in that list, that the Australian SAS "executed" (Herold's word) 4 Afghan civilians on April 29... a story by the London Times' Anthony Loyd (also referred to here):

A special forces source involved in the shooting described a small number of armed men, probably Afghans, stumbling across a six-man team of Australian SAS. Surprised, the men raised their weapons and were shot in the chest by the SAS.

Loyd does not use the word "executed." Herold adds that. (The Australian defence ministry has claimed the Afghans fired on the SAS first.) Regardless, lacking any more evidence about the Afghans' intent, it does not seem the incident warrants automatic inclusion in any list of "innocent casualties." The others I'm glossing over are all pretty much like that, too.

Posted by BruceR at 05:49 PM



"However regardless of that answer, my data stands as authoritative and un-refuted."

--Marc Herold, Nov.1

Time to go back to Marc Herold's methodology, it seems. His current minimum civilian death number in Afghanistan, from Sept. 7 to today stands at 3,141, down from his original estimate last Dec.10 of 3,767 for just the first three months. Herold says that's due to his culling of earlier bad data, suggesting he has a high level of confidence in the remainder. So let's look at this another way... how many of those remaining 3,141 alleged deaths are still easily questionable?

Part 1: July-November, 2002

Taking the latest period first, Herold's total of 3,141 currently still includes a hard minimum of 76 in the period July 1, 2002 to present. Of these, exactly 60 are from the much-publicized July 1 "wedding party" incident involving an AC-130. Herold nowhere mentions the official inquiry, which conclusively established the actual death toll at 34. Other causes of death in this period include the assassination of a Taliban leader by an unknown party (1 dead), Pakistani border guards firing on an Al Qaeda party (4), the death by bomb of a brother of a former Taliban leader in a raid (1), a July 27 firefight between Special Forces and Al Qaeda remnants in Paktia province (5), another death of a local warlord that America denies all responsibility for (1), and the victims of another Special Forces ambush in August (4). None of these deaths would seem to fit most people's traditional understanding of accidental civilian deaths (only one is apparently by aerial bombing).

Conclusion: For the period above, the safe conclusion is that, barring further investigation into the culpability of the participants, nearly all of the non-AC130 deaths are only questionably "civilian." Prof. Herold certainly does not possess enough information to authoritatively categorize any of them as definitely non-combatant fatalities.(The same may apply to some of the 34 proven killed by AC-130 as well, but there is a stronger case for them being innocents than in any of his other cases in this period.) The upshot: Herold's minimum death count exceeds the provable minimum civilian death count in this period by 42. Evaluation of the evidence for this period alone drops Herold's provable minimum to-date from 3,141 to 3,099.

Tomorrow: we look at the reliability of Prof. Herold's estimates for spring and summer, 2002.

Posted by BruceR at 02:21 AM