To put it bluntly, focusing on one number like "800 attacks" to measure the level of violence in Iraq is meaningless. That is the figure that has been getting a fair amount of attention in the last couple of days because of Bob Woodward's latest book, "State of Denial," being released next week. Though the book is not yet on sale, some details have come out, including Woodward's own highlighting of this figure: that American troops are being attacked in Iraq at a rate of approximately 800 times per week, or once every 15 minutes.
It is easy to pick out numbers that either exaggerate the level of violence in Iraq or down play its importance. It is also critical to note that no reliable estimates exist of how many Iraqis are killed because many simply disappear, and there is nothing approaching an accurate estimate of wounded.
All attack counts are extremely uncertain because many low-level attacks or individual attacks on individuals are never counted, intimidation and forced moves are not counted, and there is no way to separate murders, kidnappings, and violent crimes from insurgent actions and ethnic and sectarian violence. The kind of pressures that have led to more than 130,000 Iraqis (22,977 families) being displaced since the February 22, 2006 bombing of the Mosque in Samarra are largely unreflected in "attack" counts.
Furthermore, counts have a heavy bias towards insurgent attacks in Baghdad and the three other provinces where Sunni insurgents are more active. Counts of Kurdish-Arab violence are very uncertain -- which has a major impact on the numbers reported in Kirkuk and Mosul. So is intra-Sunni violence in the "Sunni triangle" and intra Shi'ite violence, particularly in Basra.
The Department of Defense does regularly issue numbers that at least have some value in indicating the rough intensity of the fighting. These numbers do differ in definition over time, which makes trend analysis difficult. It is also important to note that the number of attack says nothing about the intensity of attacks or the casualties that result. Some incidents are far more important than others.
Given all these caveats -- which raise serious questions about the real world meaning of any of the "punchline" figures used in Department of Defense reporting, MNF-I reporting, political debates, and various books -- the most recent data in the latest Department of Defense quarterly report (August 31, 2006) show that:
--There has been a massive increase in the Sunni versus Shi'ite sectarian violence that could trigger an all-out civil war, divide the country, and effectively force the US to withdraw. "Sectarian incidents" rose from less than 200 per month during May 2005 to January 2006, to some 1,200 in February, and have averaged around 2,200 per month ever since. (p. 35). Given the fact that such totals seriously undercount total ethnic and sectarian violence, this more than tenfold increase is a serious warning of the possibility of civil war.
--Figures like "800 attacks per day" have no meaning unless tied to a given time period and analyzed by type of attack and target. The more normal form of US reporting is based on significant attacks per week. These totaled around 470 per week during the period before the constitution was signed (February-August 2005), and have risen steadily ever since. They averaged around 540 during the referendum and election period (29-8-05 to 10-2-06), rose to 625 per week during the transition to the new government (11-2-06 to 19-5-06), and reached about 790 per week during the period of government operation (20-5-06 to 11-8-06). (p. 31)
--This is a rise of some 70% during the period in which a new constitution and new government were supposed to bring stability. These attacks concentrated more and more on Iraqi targets and Iraqi civilians, particularly in the Baghdad area -- which is the only area where counts of attacks on civilians have anything approaching minimal accuracy. The percentage of attacks on civilians in the Baghdad area rose from 15% in April to 22% in June, but such counts and percentages are so uncertain that they have limited value. (pp. 31-32)
-- The US regularly claims that almost all attacks are concentrated in four provinces - Anbar, Baghdad, Salah ad Din, and Diyala - with only 37% of Iraq's population. For example, this included 81% of US-reported attacks during May 20th to August 4th, 2006. The problem is that such counts focus heavily on the insurgency and reflect a systematic undercount of ethnic violence, intra-Shi'ite attacks, and attacks outside urban areas where the US has at least limited confidence in its sources. It simply is not clear what such claims mean.
--There is no correlation between the increase in the number of attacks during these periods and US casualties. (p.32) US casualties actually dropped during the peak periods from 1 April 2004 to November 28, 2004, when they averaged some 22-24 per day. Since that time, they have consistently remained under 20 per day. The US also has reported steady increases in IED attacks that have not led to anything approaching proportionate increases in US casualties.
--In contrast, Iraqi government, security, and civilian casualties per day have risen very sharply. These totaled less than 40 per day until the election period began on November 27, 2004. They rose to 45 per day during the election period. (27-11-04 to 11-2-05). They rose to around 50 per day during the period before the constitution was signed (11-2-05 to 28-8-05. They averaged around 48 per day during the referendum and election period (29-8-05 to 10-2-06), rose to 81 per day during the transition to the new government (11-2-06 to 19-5-06), and reached about 118 per day during the period of government operation (20-5-06 to 11-8-06). (p. 32) This is a roughly threefold increase.
Another illustration of the danger in focusing on attack counts - which say nothing about attack effectiveness is that the number of attacks on Iraqi infrastructure have steadily gone down, but Iraqi oil facilities and exports have continued to experience serious problems, and much of the economic aid effort has been paralyzed or had to be limited to "secure areas.
Infrastructure attacks averaged around 13 per week during 1-4-04 to 28-6-04. They dropped to 10 per week during 29-6-04 to 26-11-04.. They dropped to 5 per week during the election period. (27-11-04 to 11-2-05), were 5 per week during the period before the constitution was signed (11-2-05 to 28-8-05. They dropped to 4 per week during the referendum and election period (29-8-05 to 10-2-06), to 2 per week during the transition to the new government (11-2-06 to 19-5-06), and to one per week during the period of government operation (20-5-06 to 11-8-06). (p. 33).
In practical terms, however, all aid and economic activity became steadily more dangerous, water and electric power distribution continued to present major problems, and oil exports stay below 1.7 million barrels a day. (US State Department Weekly Progress Report, September 20, 2006, p. 22)