One problem with what is being reported on the WikiLeaks Iraq war logs is the civilian death numbers are being reported as estimates. They are not. There are documented. They are a minimum, not a reasonable estimate. The two can be very different, especially in a situation of massive violence.
In 2006, The Lancet published a study that gave the best estimate of the Iraq invasion and war as causing 655,000 excess deaths. This was much higher than the estimate by Iraq Body Count. A footnote in The Lancet piece provided an explanation and I tracked down the analyst: PATRICK BALLBall is a co-author of the book "State Violence in Guatemala, 1960-1996," and wrote the chapter "On the Quantification of Horror: Field Notes on Statistical Analysis of Human Rights Violations" in the book "Repression and Mobilization." Questioned about the disparity between the Lancet study and figures from media reports and efforts like IraqBodyCount.net, Ball said: "I've found a similar disparity between reported deaths and likely deaths in other conflicts that I've studied in Guatemala, Kosovo, Peru and Timor-Leste. Methods such as media reports typically capture violence well when it is moderate, but when it really increases, they miss a great deal. There are a series of biases regarding what gets reported -- we get very good reports about journalists killed, but not rural peasants; we know about big landowners, but not grassroots union organizers." Ball is director of the human rights program at Benetech, a firm that uses technology for social good, and works extensively on human rights data analysis. See: http://accuracy.org/newsrelease.php?articleId=1370&type=exactmatch&se... Earlier today I contacted Ball by email; but he's in the Congo and not terribly available.