Conflicting intelligence clouds assessment of Iraqi insurgency

The head of the Iraqi intelligence service, General Mohammad Abdullah Shahwani, told the AFP news agency in early January that the Iraqi resistance consists of 40,000 core fighters and 200,000 part-timers1. His statement caused a stir, because this estimate of the number of insurgents appeared to be much higher than anything that US sources had proposed.

The fact that Shahwani offered different figures during an interview a couple of days later went largely unnoticed. He told Al-Sharq Al Awsat, a London-based Arabic newspaper, that his estimate actually stood at 20,000-30,000 fighters with more than 200,000 non-active supporters2, figures which are comparable with the most recent US Army estimates.

A numbers game in Iraq

US Central Command (CENTCOM) had maintained an estimate of 5,000 core fighters3 throughout 2003, while coalition sources referred to 2,000-7,000 core fighters and a total of 20,000 active supporters4. By October 2004, however, the coalition revised its estimate to 12,000-16,000 core fighters, with some military intelligence sources putting the figure as high as 20,0005, even if experts working for the coalition in Iraq had made similarly high estimates months earlier.

In part, this increase seems to have reflected an intensification of the insurgency but it was also a reconsideration of earlier estimates, which were now implicitly reckoned to be underestimating the strength of the insurgency. It is unclear whether Shahwani had been misunderstood by AFP; in any case, his interview with the news agency was not the first sign of a discrepancy between military intelligence sources and general intelligence services such as the Central Intelligence Agency (CIA), which, by the end of 2003, was talking of a total of up to 50,000 insurgents6.

In fact, 23 months into the occupation of Iraq by the US-dominated coalition, very little reliable information exists on the size of the Iraqi insurgency. This is in spite of the fact that the Baghdad station has become the largest in the history of the CIA7. This lack of trustworthy data is hardly surprising, because it takes time to build an intelligence network (almost) from nothing. In this regard there are many similarities with the Afghan conflict, both under the Soviet occupation (1979-88) and after the fall of the Taliban in 2001.

The Soviet mistake

When the USSR invaded Afghanistan in late 1979, it found the Afghan intelligence service in a very bad shape; the Soviets had to rebuild it almost from scratch. For their planning and analytical purposes, the Soviets and the pro-Soviet government had to rely on extrapolating the number of insurgents from the level of military activity that they encountered on the ground. The main source for this type of estimates was the GRU, the Main Intelligence Directorate of the Soviet General Staff. The GRU estimated the number of opposition fighters (mujahideen) at approximately 30,000 in 1981 and 40,000 in 19838.

The GRU continued to produce assessments of the size of the Afghan insurgency, based on the same approach. The estimated number of mujahideen increased to more than 85,000 in 1988-89 before stabilising at 55,000 in 1989-91. By 1983, however, the KGB and its Afghan counterpart, the State Intelligence Service (KhAD), had gone a long way towards rebuilding an intelligence-gathering network in the field, which they strove to improve.

By 1985, the KGB/KhAD estimate of the number of active insurgents stood at 105,000, more then double the estimate of the GRU. Throughout the 1980s and early 1990s, the KGB/KhAD estimate was between 50-350 per cent higher than that of the GRU. When inactive armed groups were included in the count, the difference would grow to as much as a factor of 20. Interestingly, however, no later than 1982 the KGB's estimate of the total number of 'illegal' armed groups in Afghanistan was already very close to estimates produced later in the conflict - approximately 750,000.

The problem in estimating an insurgency's strength based on its level of military activity is that it is dependent on guessing how often the average guerrilla group carries out military action. Such a method also fails to take into account the fact that, due to their limited strategic mobility, guerrillas may have to stay inactive for long periods of time before any target is available to engage. This was the case with Fallujah before the US-Iraqi offensive began in November 2004. The coalition estimated that two thousand insurgents were based there; although a few managed to involve themselves in fighting elsewhere, the majority had no option but to wait for something to happen.

The understandable and inevitable difficulty in providing more accurate and reliable data on the Iraqi insurgency explains only in part the often acrimonious debate over its size. If estimates of the strength of insurgents based on their level of activity were simply used to give an idea of how much military activity was taking place, their limitations would not matter much.

Statistics muddy the waters

Unfortunately, statistics are often used to make political statements. In the case of the continuing insurgency in Afghanistan, for example, a February 2004 coalition estimate of militarily active Taliban put the number of core fighters at 1,0009, which might have provided a roughly accurate picture of how many were fighting at any given time. That number was then used by coalition spokesmen and diplomats to illustrate their point that the Taliban have been reduced to a small band of troublemakers.

However, given the scarcity of targets in the areas of insurgent activity in south, southeastern and eastern Afghanistan - and given the Taliban's low strategic mobility - in order to deploy 1,000 fighters close to potential targets, many more must be available. They could be en route to resting areas or the front lines; simply taking a break in Pakistan; or in safe areas within Afghanistan.

Moreover, insurgent activity in Afghanistan has always declined over the winter, since the time of the jihad against the Soviets, as many fighters go into winter quarters and travelling back and forth from the safe areas takes longer. Following the same pattern of self-deception, every winter during the 1980s Soviet army and intelligence officers would periodically congratulate themselves about finally seeing victory coming closer, only to be disabused in the following spring.

Taliban activity in late 2004 and early 2005 had indeed declined as usual and coalition sources in Afghanistan repeated the Soviets' mistake by presenting this fact as a sign of incipient victory. It would be wiser to wait until late spring before assessing whether the October 2004 elections have effectively weakened the Taliban.

In the case of Iraq, it is not clear how claiming to have inflicted 15,000 casualties on the Iraqi insurgents in 2004 alone10 can be consistent with estimating the number of insurgents at 20,000. No insurgency could tolerate such a casualty rate, even if the casualty total includes losses inflicted in the Shi'a insurgents of Muqtada al-Sadr, who were no longer militarily active by the end of the year.

Estimating the size of an insurgency is made all the more difficult by the fact that most people involved in it are part-time guerillas only; Iraq is no exception. Because the fighters often melt away among the population after every action, tracking them down is a particularly difficult task for the coalition and its Iraqi government allies. The real problem, however, is that producing a realistic estimate of the strength of an insurgency can be politically unacceptable.

This might explain why Shahwani felt the need in his Al-Sharq Al Awsat interview to stress that his earlier estimate of 200,000 only referred to "passive supporters" and "sympathisers". In fact, opinion polls commissioned by the coalition itself reveal that one-third of the Sunni population in Iraq supported the insurgency in 2003, a percentage that might well have increased in the interim11. That would correspond to about 400,000 adult male Iraqi Sunni Arabs supporting the insurgency - about twice as many as Shahwani estimated.

Whatever the actual number of active combatants and of losses inflicted on them, there seems to be a pool of potential recruits large enough to maintain the insurgency in Iraq for a while, unless it can be defeated politically. Washington and Baghdad hope that the January elections will take the steam out of the resistance but such an outcome is far from assured. The Sunni Arab minority does not seem at all pacified by the outcome of the parliamentary elections; a political deal with the Sunnis could prove to be incompatible with maintaining good relations with the Kurds and Shi'a.

In any case, military victory over the insurgents for the US-dominated coalition and the Iraqi government remains a distant objective. It is plausible that most of the insurgents that have been killed, wounded or captured belonged to the 'part-timers' category, corresponding therefore to a much lower attrition rate than coalition sources imply.

Dr Antonio Giustozzi is research fellow at the London School of Economics and Political Science and is running a project on Afghan warlordism

1 AFP, 3 January 2004.

2 Al-Sharq Al Awsat, 5 January 2005.

3 A Cordesman, The Developing Iraqi insurgency: Status at End-2004 (Washington, CSIS, 2004, p2).

4 Washington Post, 6 February 2005; International Herald Tribune, 23 October 2004.

5 A Cordesman, op. cit.; Jeffrey White, Todd Orenstein and Max Sicherman, Gaining Ground-Resistance in Iraq since the Transition (Part II): Effects and Implications, PolicyWatch #901, Washington Institute for Near East Policy, 16 September 2004.

6 The Guardian, 13 November 2003.

7 Los Angeles Times, 21 February 2004.

8 These figures and those below are from A Giustozzi, War, politics and society in Afghanistan: 1978-1992 (Georgetown University Press, 2000, p279.

9 AFP, 'Estimates of the size of the Taliban range from 2,000 to 10,000', 10 January 2005.

10 Washington Post, 6 February 2005.

11 A. Cordesman, op. cit. (p1)

Explore our related content