States of Ignorance The Unmaking and Remaking of Death Tolls

For the sake for comprehension, the timeline below lists the major events associated with producing estimates of Iraqi civilian deaths: March 2003 US-led invasion of Iraq October 2004 Fi

States of Ignorance:

The Unmaking and Remaking of Death Tolls

Pre-formatted version of Rappert, B 2012 ‘States of Ignorance: The Unmaking and Remaking of Death

Tolls' Economy and Society 41(1): 42-63.

Brian Rappert

University of Exeter

ABSTRACT

This article considers the complications and tensions associated with knowing about

ignorance In particular it attends to how the social analysis of ignorance hazards being associated with its production In does so through questioning how the UK government contended the number of civilian deaths stemming from the 2003 Iraq invasion could not

‘reliably’ be known The twists and turns of official public statements are interpreted against back region government and civil service deliberations obtained under the British Freedom of Information Act Far from settling what took place, however, this material intensified the problems with analysts attributing and characterizing strategies for

manufacturing ignorance From an examination of the choices, contingencies, and challenges in the way actors and analysts depict ignorance, this article then considers future possibilities for inquiry whereby social analysts can question their ignorance while questioning claims to ignorance

Key words: Partial knowledge; armed violence; pragmatism; Freedom of Information;

Iraq

Author’s Comments: My thanks to Linsey McGoey, Richard Moyes, three anonymous

reviewers, and the participants of the "Strategic Unknowns” conference at the Sạd Business School for their advice and assistance in the production of this article

As elaborated in the Introduction to this special issue, recent years have witnessed

renewed concern across many fields to the complex and intertwined relation between knowledge and ignorance The growing literature into the place of ambiguities, absences,undecidables, and uncertainties has examined the variants of ignorance (Smithson 1993), what they mean for the diverting, displacing, and denying of responsibility (McGoey 2007; Kyriakoudes 2006), and how claims to not know are argued (Stocking and Holstein1993; Proctor and Schiebinger 2008) As part of this, unknowns have not simply been orientated to as ‘knowledge gaps’ that need to be filled Instead, relations of ignorance allow for the highly productive negotiation of identities, boundaries, and hierarchies Also, what is ‘unknown’ is not simply the result of a lack of efforts to establish

knowledge Rather ignorance can be deliberately manufactured

Attention to the knowledge-ignorance inter-connection has run along side related

developments in thinking about the relation between transparency and concealment Herein, openness is not a condition that can be taken for granted, but rather a negotiated accomplishment rendered through mediating mechanisms that simultaneously produce non-disclosure (Vattimo 1992; Power 1999; Christensen & Langer 2005)

While recent studies have elaborated the often nuanced relevance of ignorance for

organizational life and social interactions, arguably they have been much less attentive to the meta-level question of how ignorance is produced through the analysis of ignorance Notions such as ‘symbolic violence’ (Bourdieu and Passeron 1977) and ‘ontological gerrymandering’ (Woolgar and Pawluch 1985) have been proposed in the past to depict what social analyses do through selectively leaving out certain details Yet, the rich empirical and conceptual studies to date of ignorance have generally sought to produce

knowledge about ignorance, rather than attend to how their work might be bound up with

its reproduction

This article takes as its central concern the tensions of knowing about ignorance

Borrowing a long standing distinction in social constructivism (Bloor 1976), it examines the intertwined issues of the place of ignorance in the conduct of social ‘actors’ as well as

in the claims making of ‘analysts’ of social actors Attending to the latter is of vital importance in the study of strategic unknowns That ‘ignorances’ or ‘negative

knowledges’ are being placed under the label of ‘strategic’ speaks to the manner in which they are often regarded as products of intended or purposeful action Combining this starting concern with calculation with the added focus on the negative uses of ignorance offers a tempting project to analysts: unmasking how strategic unknowns are the

outcomes of individual and institutional duplicity The efforts undertaken by the tobacco industry to evade a casual link between smoking and cancer is the archetypal example in this respect A danger with such a pursuit is that it projects a coherence in action and competence to actors which might well be unwarranted

The two and twined aspects of the production of ignorance by actors and analysts are considered in relation to attempts to gauge the human costs of conflict Among the many

aspects of the tolls of war, this article considers one: the assembly of civilian fatality figures This is done specifically in relation to the British government’s claimed

understanding about the number of deaths stemming from the 2003 Iraq invasion In giving an account of Iraqi deaths, this article queries the state of ignorance at two levels:

1) The claims made government actors regarding the (im)possibility of producing fatality estimates;

2) The meta-epistemological level of how analysts can know about the ignorance production strategies of actors

The ‘Enacting Ignorances’ section provides a chronology of deliberations by government officials, the examination of which provides many grounds for contending that ignorance about fatalities was deliberately manufactured to deflect political criticism In doing so, agoal of this section is also to ask what is at stake in how analysts orientate to claims aboutthe production of unknowns Problems are identified with concluding officials sought to manufacture ignorance In the ‘Attributing Ignorance’ section these are expanded upon

and complemented with additional considerations to call into question how analysts

substantiate claims about strategic unknowns The final section offers a rethinking of research directions and questions attuned to the sensitivities of ignorance within the study

of ignorance

Freedom of Information

Much the argument of this article is advanced through comparing official statements and other publicly available material against documents obtained through under the 2005 UK Freedom of Information (FoI) Act This Act was justified as part of attempts to open the traditionally closed off British political system to public scrutiny And yet, as argued elsewhere (Wasserstein 2001) and as will be evident in this case, FoI responses are characterized by limitations and vagaries that mark a highly managed form of disclosure The plainly fractured understandings enabled by FoI responses will be used to bring to the fore some of the more commonplace difficulties of knowing about ignorance

More specifically, the article draws on three sets of overlapping requests made between 2008-2010 by the author, Richard Moyes (Landmine Action) and another (for more details1) In total, some 48 emails, letters, and other documents were obtained

For the sake for comprehension, the timeline below lists the major events associated with producing estimates of Iraqi civilian deaths:

March 2003 US-led invasion of Iraq

October 2004 First Lancet survey published

November 2004 Parliamentary statements on first Lancet study

December 2005 Iraqi government and parliamentary elections

October 2006 Second Lancet survey released

October 2006 Parliamentary statement on second Lancet study

January 2008 New England Journal of Medicine study published

Enacting Ignorances

There are no reliable figures for Iraqi civilian deaths since March 2003 The Iraqi

Ministry of Health has informed us that the number of civilians killed in security incidents is 1,203 and 3,992 wounded dating from when statistics began on 5 April 2004 However they reflect only hospital admissions and may not be

comprehensive It is not possible to break these down into how they were killed orwho may have been responsible It includes casualties caused by terrorist action

By 17 November 2004, however, official statements suggested not only that reliable

figures did not exist, but that it was not possible to derive such figures The then Foreign

Secretary Jack Straw (2004) argued that ‘In many cases it would be impossible to make a reliably accurate assessment either of the civilian casualties resulting from any particular attacks or of the overall civilian casualties of a conflict This is particularly true in the conditions that exist in Iraq.’

As a way into understanding how claims of ignorance figured in debates about Iraqi dead,this sub-section considers the background leading to the above contention of the British Foreign Secretary in November 2004

Based on the FoI responses the trigger for government and civil service public

declaration of the impossibility of civilian deaths was a survey in the medical journal The

Lancet released online in late October 2004 (Roberts et al 2004).2 This study, lead by a group at Johns Hopkins University, employed cluster statistical sampling techniques Using a baseline morality rate, the authors estimated 98,000 more Iraqis died than would have in the absence of the war (with a 95% confidence interval estimation range from 8,000 to 194,000) These numbers contrasted with those derived from other methods being used at the time.3 Notably, the “Iraq Body Count” relies on English language news accounts as well as other substantiated reports Over roughly the same time frame, it counted between 14,284 and 16,419 non-combatants deaths from military or paramilitary violence

Certainly it is possible to cite evidence from the FoI responses obtained that suggest officials were deliberately working to manufacture ignorance (along the lines of Proctor 2008) in the face of the comparative large and media prominent estimates advanced by

the Lancet study.4 One instance is a passage in a RESTRICTED security level letter fromthe Chief Economist of government ministry dated 8 November 2004 The name of this advisor and his ministry were redacted from the released versions – though presumably

he was from the Foreign and Commonwealth Office (FCO)5 After discussing various

aspects of the validity of the 2004 Lancet study, this Chief Economist commented:

It might also be possible, as Gerald Russell has suggested, to try and validate the study’s pre-invasion estimate of mortality by checking it against unpublished [Iraqi Ministry of Health] health figures But there is (a) no certainty at this stage that this kind of work would invalidate the Lancet findings, or (b) any guarantee that if it does produce a difference answer, that the rejection of the Lancet

findings would be conclusive

This passage suggests particular findings were sought (i.e those that ran against the high

Lancet findings) Likewise, inter-ministry email correspondence on 9 November 2004 to

formulate the official statement by the UK Foreign Secretary suggests officials tried to encourage public statements that would raise doubts about the estimation of 98,000 deaths In the exchanges, one official (name and ministry redacted) cited an October poll

by the International Republican Institute (IRI 2004) It indicated that 22 percent of some two thousand Iraqis responded in the affirmative when asked: ‘in the past year and a half, has your household been directly affected by violence in terms of deaths, handicap or significant monetary loss’ Another individual (again name and ministry redacted) responded to the citation of this result: ‘The IRI survey seems to me to harm our

argument rather than help, but it is certainly useful to know.’ Consistent with the intent to

selectively disregard evidence which did not undermine the Lancet study, it has not been

possible to find any mention of this IRI finding in subsequent public government

statements about Iraqi dead

Although grounds exist for maintaining that ministry staff sought to deliberately foster

ambiguity surrounding the reliability of the Lancet study, it is difficult to assess and

determine whether staff consciously intended to produce ignorance surrounding the study’s comparatively high estimations One complication is knowing what those under scrutiny knew In this case, it is uncertain to what extent the contentions about the lack of reliable figures stemmed from the intention of officials to create doubt, or from their own lack of knowledge about the possibility of estimations Save for fairly narrow

methodological interventions by technical advisors, the 2004 deliberations obtained underthe FoI action were not well informed by the history of attempts to calculate deaths in war Repeated uncertainty and confusion was expressed by individuals about the basic statistical matters under discussion

Perhaps a more fundamental trouble in moving from text to strategy is the standing of language However otherwise diverse, many of the strands of what are called Discourse Analysis today treat accounts of the world as managed descriptions given in and for particular interactional settings (Lynch and Bogen 1996; Edwards 1997; Alvesson 2002; Blommaert 2005) When language is approached as a form of social action between a sender and an audience – rather than just a means of general representation – extracting out some stable, definitive meanings from words becomes problematic.6 In this regard,

the quotations above relating to a desire to find grounds against the Lancet estimations

need to be made sense of as part of ongoing internal exchanges between individuals These are laden with mutual expectations (e.g., about literalness), taken for granted understandings, varying levels of trust, organizational idioms, unspoken presumptions,

etc (Arminen 2000).7 As such – and as in the case of the 2009 leaked emails about the

‘tricked’ manipulation of global warming data by researchers at the University of East Anglia – arguments that some are trying to dupe others can be queried for the way

meaning gets attributed ‘out of context’

Furthermore, as Gilbert and Mulkay (1984: 2) argue, attributing definitive meaning to statements presumes that ‘the analyst can reconcile his version of events with all the multiple and divergent versions generated by the actors themselves’ In other words, social actors can produce contrasting accounts; sometimes differentiated along the forms

of communication (e.g., formal writing, informal collegial banter, interviews, emails, etc).Those who take it as their role to decipher what is really meant must find a way of sortingthis diversity Often this is done by establishing ‘linguistic consistency,’ as Gilbert and Mulkay suggest: what is taken as literally descriptive is that which is in line with what analysts judged to be the overall gist of the material available As a result, some

statements are taken at face value while others are disregarded Moving on from my

aforementioned impression of the overall drive towards doubting the Lancet death

estimations to characterize the intentional ignorance production strategies afoot would likewise require sidelining certain statements made by officials; such as those that

referred to a ‘genuine judgment’ that no reliable methodology was possible and that the

Lancet study was ‘straight from the department of guesswork.’8

If these difficulties stemming from treating language-as-social-action pose challenges for analysis in general, then these are all the more acutely experienced where information restrictions are imposed Most of the names, many of the positions, and some

organizational affiliations were redacted from the FoI released documentation.9 Such conditions of partial disclosure frustrate government openness leading to comprehension

One possible response to the concerns about the interpretation of such ‘back region’ deliberations would be to place more analytical investment in statements made in front regions – such as the one given by the Foreign Secretary Jack Straw in relation to the

Lancet article As these set out informed authorized positions for the public by named

individuals, they might be treated as offering relatively fixed points for analysis

And yet, trying to extract a stable reading is arguably problematic because of the

equivocal meaning of key points Consider the 17 November 2004 statement by Jack Straw to the House of Commons So, with regard to the central question of the

possibility of deriving figures for civilian casualties, Straw (2004) maintained both 1) that:

In many cases it would be impossible to make a reliably accurate assessment either of the civilian casualties resulting from any particular attacks or of the overall civilian casualties of a conflict This is particularly true in the conditions that exist in Iraq

and 2) that the hospital reports complied by the Iraqi Ministry of Health suggesting that there were 3,853 civilian fatalities from the military or terrorist action between 5 April

2004 and 5 October 2004 were the ‘most reliable available’ figures.

Arguably the meaning of ‘reliable’ was ambiguous It was both impossible to achieve reliable figures, but possible to specify the most reliable ones The slipperiness of this term makes it difficult to establish inconsistencies, even with other official statements For instance, on 17 November 2004, Baroness Symons read to the House of Lords the same assessment as given by Jack Straw to the House of Commons Yet, in response to questioning in the House of Lords, earlier on 7 June 2004 she also had said:

[Iraqi Ministry of Health] statistics are not reliable, as Iraqis often bury their deceased relatives without official notification/registration This has been

particularly true during periods of heightened conflict The MoH does not

therefore have accurate figures for civilian deaths or their causes for the past year

(Symons 2004b)

Determining whether this statement is consistent with the aforementioned ones she made

on 17 November or 24 June 2004 would seem to hinge on the meaning given to the notions of ‘accurate (enough)’, ‘comprehensive (enough)’ and ‘reliable (enough)’

To complicate matters further, it appears that some of those considering official responsesrecognized a need for front and back region strategic management of what was said An internal FCO letter sent to 10 Downing Street on 14 October 2004 stated:

The US have, like ourselves, stuck to the line that there are no comprehensive figures for civilians casualties and do not comment on suggested figures The Embassy in Washington has asked for the US’s official estimate of civilian

casualties in Iraq We still await the responses from the State Department and Department of Defense

In sum, if we produce a figure that differs from the Iraqi government figures, we will have to defend it – and the way it was arrived at – before parliament and the media.WWWWW_WWWWW_WWWWWWW_WWWW_WWWWW_WWWW_WWWW We recommend that for the moment we continue to put our public emphasis on specific atrocities against civilians, such as the mass killing of Iraqi children in Baghdad on 30 September, and their attempts to thwart our efforts to stand up independent Iraqi security forces

This passage would seem to indicate this FCO official (a) recognized the possibility that the US and the UK could derive civilian causality estimates; (b) believed the US had already produced one; (c) was unaware of the estimation; and (d) thought it was best to maintain a public line that it was not possible to derive a comprehensive figure With regard to (d), it was recognized that producing figures might require defending them To adhere to the stated recommendation to not comment on suggested figures would have the impact of reducing the need for detailed defense The passage suggests not treating government statements at face value because at least some individuals saw the need for the sort of calculative control moves of the kind Goffman (1970) deemed as ‘expression games’

While ministers Straw and Symons made seemingly tension-ridden claims about the overall possibility of producing reliable figures in their addresses to parliament

statements, in relation to the 2004 Lancet report the specific challenges offered by them

on 17 November centered on the limitations of the report’s data that were inputted into the statistical methodology Doubts were raised in relation to the said small sample size, the overall ‘limited precision’ of the data that was noted by the study’s authors, and the possibility of accurately attributing who was responsible for violence and who was a

civilian The contrast between the Lancet study with the Iraqi MoH numbers was also

cited as raising grounds for doubt Thus, while it was impossible to achieve ‘reliable’ figures for the UK, the most ‘non-reliable’ one could be identified

In choosing to question the Lancet study at the level of the data, it was possible for

ministers to disregard its findings without questioning the statistical clustering method employed – something technical advisors warned against in the FoI documentation obtained.10 Working at the level of data at least opened the prospect of reliable figures in

the future – if the data could be improved And yet, this also sat uneasily against the

aforementioned overall contention that reliable figures were impossible in the case of Iraq

2006

According to the FoI released material, another round of inter-ministry deliberation aboutIraqi civilian deaths began in the autumn of 2006 with the publication of a second survey

by Johns Hopkins University in The Lancet (Burnham et al 2006) on 12 October That

work estimated 654,965 more Iraqis died up to that date than would have in the absence

of the war (with a 95% confidence interval estimated range between 392,979 to 942,636).With this second study, it is possible to ask how UK claims developed from that in 2004 Immediate British government commentary on the 2006 study was highly critical and wide ranging On the day of its publication, the Prime Minister’s official spokesman was reported to have commented that:

The problem with this is that they are using an extrapolation technique from a relatively small sample, from an area of Iraq which isn't representative of the country as a whole We have questioned that technique right from the beginning and we continue to do so The Lancet figure is an order of magnitude higher than any other figure; it is not one we believe to be anywhere near accurate (Tempest 2006)

Instead of accepting the estimations in the Lancet, the figures from the Iraqi MoH were

said to be the only accepted ones

On 19 October 2006 though the Parliamentary Under-Secretary of State, Foreign and Commonwealth Office, Lord Triesman (2006), gave the following prepared answer to theHouse of Lords (see as well Ingram [2006]):

My Lords, every civilian death is a tragedy and must be of concern in Iraq, as

elsewhere However, we continue to believe that there are no comprehensive or reliable figures for deaths since 2003 Estimates vary according to the method of

collection The figure of 655,000 given in the recent Lancet survey is significantly

higher than other estimates, including those provided by the Iraqi Government

We believe that the Iraqi Government are best placed to monitor deaths among their own civilians

This statement carried on with certain tracks of contention in 2004: no reliable figures

were available, estimates varied, and the Lancet study was out of line with others

The statement differed though in that while the quality of the data was a central concern

in 2004 official parliamentary statements, reference to it was not included within such statements in 2006

Through the FoI releases, it is possible to speculate about the causes and consequences for this omission Similar to 2004, this material from 2006 indicates civil servant

advisors (from Department for International Development [DFID] and the Ministry of

Defence [MoD]) recommended against criticizing the Lancet study’s methodological

design (as seemingly done in the statement above by the Prime Minister’s official

spokesman).11 In the case of the DFID advisor, grounds were given for why the 2006

Lancet study might have underestimated mortality.12 In addition, the advisors noted improvements to the data obtained; including the larger sample sizes, sampling

techniques, and the use of enhanced death certificate verification When these advisers’ appraisals about methodological soundness later got out to the press, they were seized upon to indicate government skullduggery (Horton 2007)

By not noting the improvements made to the data, the official statement by Lord

Triesman failed to acknowledge how the 2006 Lancet study redressed 2004 ministerial

criticisms The very brevity of the statement13 though arguably frustrates efforts to assessthe grounds offered for Lord Triesman’s evaluation

As another point of contrast between 2004 and 2006, instead of the (qualified)

endorsement of Iraqi Ministry of Health figures, in 2006 Lord Triesman contended that

‘Estimates vary according to the method of collection’ This shift to methodological pluralism, the quality of the data in 2006 was diminished as a relevant consideration sincethe data was not the (only) source for the divergent deaths figures Yet, adopting this stance brought critical questioning during a debate in the House of Lords:

Lord Marsh: My Lords, does the Minister agree that the methodology of this study was unique in the way in which it was pursued? It is difficult to see how theGovernment can take the line, “The study was done in a way which is well

known, and it was done very well, but we don't think that it is worth very much”.Lord Triesman: My Lords, that is not the view that I have put at all I said that there are different methods which have arrived at very different figures and that

those methods also are legitimate The way in which data are extrapolated from samples to a general outcome is a matter of deep concern and merits considerable study rather than the denunciation of one method compared with another.

In pointing to the ‘deep concern’, the response by Lord Triesman appears to open up statistical methodologies to questioning in a manner that were arguably not subject to in

2004 official parliamentary statements that took data as their concern; though again what

is being discussed might well be regarded as unclear

Just what Lord Triesman’s statements should be taken to mean overall for the standing of death estimates is difficult to comprehend The monitoring of deaths was deemed a responsibility of the Iraqi government, and therefore, presumably, doable at some level

As such, ignorance was framed in a way that further knowledge was attainable (as in Ravetz 1987) And yet, the suggestions of methodological pluralism undermined the prospects of sorting out said equally legitimate tallies In this thinking it might be said,

following Norris (1994: 290), that the dead became ‘objections of deconstruction, figures

impossible to verify and locate and therefore incapable of serving any intellectual

operation other than that of the impossibility of determining their reality.’

What does seem certain though is that in practice the need for ‘considerable study’ mentioned in Lord Triesman’s reply was not part of an announcement that the UK would

be commissioning such work

As with previously cited FoI material, the 2007 material raises basic concerns about the ignorance of British officials; in particular in relation to error and uncertainty So in relation to the former, for instance, the 7 December two-page checklist (titled ‘Analysis

of Iraqi Civilian Death Tolls’) included the following evaluations about the 2004 Lancet

However, in the internal letters obtained by the author, while certain critical points were raised (mainly relating to data), the MOD’s Chief Scientific Adviser judged that the

‘design of the study is robust’ and a ‘Chief Economist’ noted above (presumably from theForeign Office) concluded the statistical methodology ‘appears sound’.14

In relation to concerns about uncertainty, for instance, the checklist document suggested the basis for some Iraq government figures was unknown to the UK For both the Iraqi Ministry of Interior figures and joint Iraqi Ministry of Interior, Health, and Defence figures, it was written that no details were provided about how they were put together.15

So despite the UK proposing the Iraq government was best placed to monitor deaths, the former did not know how the latter was doing so

And yet, while it seems appropriate to label these as instances of error and uncertainty, such characterizations may be ill-judged Especially given what seems to be the lack of knowledge displayed across the civil service regarding Iraqi deaths, the uncertainty aboutthe Iraqi government methods for compiling death figures might well stem from

confusion

2008

Post-2007, public commentary and debate about Iraq civilian deaths continued For

instance, at the start of 2008, the New England Journal of Medicine published an

interview survey conducted by various Iraqi organizations in collaboration with the

World Health Organization (Iraq Family Health Survey Study Group 2008) From March

2003 to June 2006, it estimated 151,000 violent Iraqi deaths – including combatants and civilians

This study does not appear to have lead to further activities in UK ministries No

documents were released under the FoI Act relevant to this study Little public attention was given to the overall topic either A House of Commons parliamentary question in February 2008 about Iraqi deaths (Rennie 2008) brought a reference back to a 2007 statement by the Secretary of State for Foreign and Commonwealth Affairs indicating:

The Government do not collate figures for civilian casualties in Iraq The

Government of Iraq is best placed to monitor the numbers of Iraqi civilian

casualties, but we continue to believe that there are no comprehensive or reliable figures for deaths since March 2003 as estimates vary according to the method of collection (Howells 2007)

As such, the government did not endorse the findings in the New England Journal of

Medicine but instead said estimates vary

As in previous years, the British government statements in 2007 and 2008 identify the Iraqi government as institution that should be monitoring deaths As in previous years, there is no suggestion that this government might be divided about the desirability for or