Michael Collins has provided comments on my article on national security intelligence ethics and offered a legal proceduralist conception in opposition to my teleological conception.[1],[2] Collins says that:

Just Intelligence Theory treats Intelligence Collection as ancillary to preventing or winning wars, thus it imports Just War Theory’s mentality. Miller’s examination, which identifies the makings of a workable theory in the necessity and proportionality analysis, maintains this war footing by focusing too much on the purposes of Intelligence Collection.[3]

… [please read below the rest of the article].

Image credit: Paul Frankenstein via Flickr / Creative Commons

Here I initially make a number of points in relation to Collins’ account of the content of my article since that account is incorrect in a number of important respects. Following that I discuss two other issues Collins raises.

Firstly, it is incorrect that I support Just Intelligence Theory or that my “examination maintains this war footing [of Just Intelligence Theory]”. Regarding Just Intelligence Theory, I say:

[W]hile so-called Just Intelligence Theory—comprising jus ad intelligentiam and jus in intelligentia—provides a useful starting point in the construction of a normative framework for national security intelligence activities, ultimately it is found to be wanting in a number of important respects. For instance, while war ought to be a last resort, intelligence collection and analysis ought to be a first resort. On the other hand, some constitutive principles of Just War Theory are, appropriately revised, applicable to national security intelligence activity, notwithstanding the essentially epistemic character of intelligence activity.[4]

Thus, sections 3 and 6 of my article discuss important respects in which the Just Intelligence Theory is to be found wanting. For instance, I argue in section 6 that “there is a normative principle governing espionage, in particular, that is not a constitutive principle of Just War Theory; this is a principle of reciprocity”.[5]

Now consider the claim that I “maintain this war footing [of Just Intelligence Theory]”. On the contrary, I say, for instance:

The targets of internal national security criminal intelligence activities such as terrorists are, presumably, criminals or suspected criminals. As such, they fall under what might be termed the law enforcement model under which law enforcement agencies such as the FBI conduct their activities, including their national security intelligence activities. However, the law enforcement model does not sit well with JWT [Just War Theory] nor, evidently, with JIT [Just Intelligence Theory].[6]

In fact, my view of national security intelligence is that it is considerably wider than military intelligence (and excludes some military intelligence) and also wider than criminal intelligence (but, of course, excludes most criminal intelligence). Accordingly, I say:

National security intelligence is sometimes collected, stored, analysed and disseminated, as actionable intelligence, by military organisations, sometimes by police organisations, but paradigmatically by intelligence agencies the institutional purpose of which is internal and/or external national security, e.g. the CIA, NSA, GCHQ, MI5, MI6, Mossad, RAW, ASIO, etc. Accordingly, what makes information or other data collected by these agencies national security intelligence is that these agencies collect, analyse and disseminate this information in the service of national security—national security being the primary institutional purpose of these agencies.[7]

More generally, in my article I do not offer a theory of national security intelligence. Rather I content myself with a brief and limited characterisation of national security intelligence as “intelligence pertaining to serious internal or external threats to the nation-state itself, or to one of its fundamental political, military or criminal justice institutions, and that these threats might emanate from state or non-state actors, e.g. terrorist groups”.[8] Since the nation-state also relies on economic (e.g. agribusinesses, banks) and epistemic, (e.g. universities, the news organisations) institutions, then, at least in principle, serious threats to these institutions might also come to be national security threats. However, I do make the point—apparently contested by Collins—[9] that the concept of national security should not become so permissive as to be equated with national interest since to do so would license all manner of morally illegitimate activities, and associated intelligence activities, in the name of national security, such as wars of conquest.[10] Moreover, I distinguish between institutions that conduct national security intelligence activity and, in relation to that activity, between the macro and the micro levels.[11] Importantly, and consistent with Collins’ own view,[12] and as I argue elsewhere in detail,[13] these institutions are primarily, but not exclusively, epistemic institutions. However, the burden of my article consists in, firstly, rejecting Just Intelligence Theory (as far as I am aware, the currently only worked out normative theory of national security intelligence on offer) and, secondly, offering analyses of four particular principles that I argue are relevant to the ethics of national security intelligence activity, namely (as the sub-title of my article makes clear), discrimination, necessity, proportionality and reciprocity. In doing so, however, one of my main points is that each of these principles, in its formulation and application in relation to national security intelligence activity, is very different to its formulation and application in Just War Theory, i.e. in kinetic military activity. For instance, Just War Theory’s jus in bello principle of discrimination prohibits deliberate targeting of innocent civilians but the counterpart principle of discrimination* in national security intelligence activity does not prohibit, for instance, surveillance of suspects or even, if necessary, of the innocent associates of suspects or offenders.[14]

Let me now turn to two other issues raised by Collins, namely, knowledge as an end-in-itself and institutional purpose. Collin’s views on these two issues are quite different from mine, although I do not address either of these issues in any detail in my article.

National Security Intelligence as an End-in-Itself 

Collins seemingly rejects my view, indeed the standard view, that national security intelligence is actionable in favour of the view that knowledge is an end-in-itself (“We should reframe our view of Intelligence Collection. Miller narrows his focus to actionable intelligence …”[15]). Contra Collins, I suggest that national security intelligence ought to be both an end-in-itself and actionable (which is, of course, not to say that every single item of intelligence is actionable on its own—actionable intelligence is, typically, a structured set of items). At any rate, Collins misunderstands the relationship between knowledge, including accurate national security intelligence, as an end-in-itself and actionable intelligence. Let me explain.[16]

Contrary to Collins view that national security intelligence is merely in end-in-itself and is not intended to be actionable, intelligence is in fact in the service of kinetic activity, e.g. making arrests, preventing cyber-attacks or bombing installations, and, therefore, needs to be actionable by the relevant decision-makers, e.g. police officers, combatants, politicians. Accordingly, intelligence officers need to be responsive to decision-makers but also, if their intelligence is ultimately to be beneficial, somewhat independent of decision-makers so that it is evidence-based and not vitiated by political interference. Moreover, some distinction needs to be maintained between intelligence as ‘raw data’ and intelligence as the epistemic product of some process of analysis and evaluation according to different criteria, including the likelihood that it is true, its importance (assuming it is true), and (relatedly) the reliability of its source. Hence the distinction between collectors and analysts, but also the need for ongoing cooperation between the occupants of these different intelligence roles.

In the light of the above-mentioned distinctions between collectors, analysts and decisions-makers operating in an adversarial context, it makes sense to introduce some notion of an intelligence cycle (of which there are many available models) involving not only a one-way circular process in which intelligence is directed to be collected, collected, analysed and acted on by decision-makers who in turn direct further intelligence to be collected—doing so in part because of the actions of antagonists (including in response to the actions consequent on the decisions of one’s own decision-makers). The process is not simply circular but also (at least ideally) two-way interactive at each of the points in the ‘circle’, e.g. between intelligence officers and decision-makers.

In addition, intelligence can be categorized in various ways according to its source, mode of communication, content, potential use and so on. For instance, regarding its potential use, intelligence can be categorized as strategic, tactical or operational. This latter set of distinctions can be seen in the light of the threefold distinction between the institutional, the macro and the micro levels. The institutional level refers to matters such as the purpose, structure, resources and culture of an intelligence agency, and its institutional relationships to, for instance, government or the military forces it serves. There is a further distinction between activities at the macro level (e.g. the design of a national intelligence strategy, the establishment of bulk databases by security agencies for national security purposes) and those at the micro level (e.g. the conduct of a specific operation utilising data from a bulk database).

Again, intelligence can also be categorized (inter alia, according to its source and mode of existence) as human intelligence (HUMINT), signals intelligence (SIGINT), social media intelligence (SOCMINT), open-source intelligence (OSINT), imagery intelligence (IMINT), geospatial intelligence (GEOINT) and so on. In recent times various forms of electronic intelligence have emerged as of great importance, notably metadata (phone or email data other than content, e.g. phone call number, time and duration of call) in the context of end-to-end encryption impeding access to content. Moreover, OSINT and SOCMINT are increasingly important intelligence sources. Thus intelligence agencies have increased their uses of data mining and analytics technologies, notably machine learning techniques and computer vision algorithms. More generally, the content of intelligence is multifarious. Content can include discrete items of information, e.g. name of a foreign agent, or larger fragments of epistemic material, e.g. a list of associates, the strategic plan of the enemy. The content might be a fact, a formula, a map, an image, an opinion, an ideological claim, an expressed emotion, a video clip, a narrative about a sequence of events, and so on.

Notwithstanding its multiplicity of forms, intelligence, and national security intelligence in particular, is an epistemic notion and, ideally it consists of knowledge, i.e. it is true or correct or accurate or probably true, or some such. Thus, intelligence officers aim at, or ought to aim at, knowledge. However, the fact that intelligence officers ought to aim at knowledge is entirely consistent with the above made claim that national security intelligence activity is ultimately undertaken in the service of kinetic action undertaken by other agencies. For knowledge is the fundamental proximate point of intelligence collection and analysis but, nevertheless, such knowledge, once acquired by intelligence officers, needs to be disseminated to others, notably decision-makers, and not ‘left in the head’ of the intelligence officer. In short, intelligence officers ought to have the acquisition of knowledge as their principal aim or end but, in addition, intelligence thus acquired needs to be actionable.

Indeed, the acquisition of knowledge is not only an end, but an end-in-itself, for intelligence officers; and this is so, notwithstanding the further requirement that the truths acquired be actionable. For the activities of intelligence collection and analysis are not related to knowledge merely as means to end, but also conceptually. Truth is not an external contingently connected end which some intelligence activities might be directed towards if the intelligence officers happened to have an interest in truth, rather than, say, an interest in falsity. Rather truth is internally connected to intelligence activity. Thus, aiming at truth is aiming at truth as an end-in-itself. This is consistent with also aiming at truth as a means to some other further end, such as winning a war. In other words, supposed intelligence activity which only aimed at truth as a means to some other end would not be genuine intelligence activity or would be defective qua intelligence activity, since for such a pseudo-intelligence officer truth would not be internal to his or her activity.

Such pseudo-intelligence officers would abandon truth-aiming if, for example, it turns out that the best means to the officer’s end is not after all truth, but rather falsity. Obviously, such pseudo-intelligence officers would be extremely dangerous since their intelligence would be very unreliable. For they are not simply officers who aim at (and more often than not acquire) the truth but who, nevertheless, often present false reports to their political masters (or other ‘clients’) knowing them to be false (or, more likely, to be somewhat misleading because unpalatable truths are omitted or downplayed). Rather these pseudo-intelligence officers do not aim at truth in the first place. That is, having little interest in the truth, they do not seek the truth and, as a result, do not themselves acquire knowledge; therefore, they do not have knowledge to pass on to their political masters.

Of course, in the real world such pseudo-intelligence officers are unlikely to exist in a pure form. However, in an intelligence agency lacking in independence and in which intelligence officers’ desire to please, or more likely, desire not to antagonise their political masters (e.g. some Soviet intelligence officers who served under Stalin) the commitment to the truth might well weaken, especially when one considers the inherent difficulties in acquiring accurate, significant national security intelligence from adversaries determined to maintain information security. As a consequence, such intelligence officers might initially have the practice of reporting what they know to be false or misleading on some occasions when it is politically or otherwise expedient to do so, but end up over time largely abandoning the practice of evidence-based truth-seeking in favour of selective data collection and skewed analyses in the service of personal, political or other non-epistemic agendas; that is, end up becoming something akin to pseudo-intelligence officers.

National Security Intelligence Agencies and Institutional Purpose

Collins rejects my teleological perspective in favour of a legal proceduralist one (“a theory of intelligence should start with the political theories constraining all government action. These take the form of neutral predetermined rules and procedures”[17]). While I concentrate on defending my teleological perspective here, I note in passing that in describing his proceduralist approach he constantly makes reference to institutional purposes, e.g. “Purpose can be part of the analysis”.[18] I suggest that this is, in fact, inevitable. This is because institutional purposes are a defining feature of institutions, normatively speaking. I also note that my teleological account of the normative foundations of institutions (detailed elsewhere[19]), including epistemic institutions, is framed in terms of institutional purposes, institutional structure and institutional culture. However, on this account institutional procedures are, or ought to, in large part derived from institutional purposes, e.g. the rights and duties of intelligence officers ought to derive in large part from the epistemic purposes of intelligence agencies. (There are, of course, human rights and other side constraints on intelligence activity.)

As we have just seen, whereas the primary institutional purpose of national security intelligence agencies is essentially epistemic, the realisation of this epistemic purpose serves a larger national security purpose only realizable by the kinetic activity of other institutions, e.g. the military. Accordingly, there is an institutional division of labour; the intelligence agency provides knowledge (or weaker epistemic goods) to the decision-makers, e.g. politicians, military leaders, who in turn act (or refrain from acting) on that knowledge. In order for this institutional division of labour to function successfully it is critical that the intelligence provided is reliable and, therefore, that the epistemic activity of the intelligence agencies is unduly influenced or otherwise undermined by the institutions which they serve, e.g. by their political masters.

Consistent with an appropriate level of responsiveness to their political masters’ national security intelligence demands, it is necessary that intelligence officers’ professional commitment to the epistemic purposes of their intelligence agencies override any personal loyalty they might have to their political masters; indeed, on occasion, they may need to speak unpalatable truths to power. However, it is also necessary that intelligence officers have an overriding professional commitment to the epistemic purposes of their intelligence agencies rather than seeking to realize the ultimate national security outcomes that might or might not flow from the decisions of the politicians, military leaders and other decision-makers who act on their intelligence. It is important that intelligence officers to not engage in institutional overreach.

National security intelligence activity is cooperative or joint in nature; indeed, it is a form of institutional activity. Accordingly, it is a species of joint epistemic action.[20] The activities of other security agencies are also forms of cooperative, institutional activity and, therefore, they are also species of joint action; however, they are predominantly species of joint kinetic activity.

As suggested above, and argued in detail elsewhere,[21] security agencies are, or ought to be, established to realise collective ends which are collective goods, namely security (to which the relevant citizens have joint rights), and inevitably do so via joint action (including the multi-layered structures of joint action constitutive of organisations). Intelligence agencies are no exception. However, the joint action which they perform is distinctive in that it is essentially joint epistemic action (at least, in so far as the intelligence agencies in question do not engage in so-called covert action, such as sabotage, targeted killing and other kinetic activity).

Importantly, joint epistemic action, as is the case with epistemic action more generally, while it is a necessary condition for kinetic action is not a sufficient condition. Rather, roughly speaking, it stands to kinetic action as beliefs stand to action (other than to mental actions, such as judgements), more generally, i.e. it is mediated by affective and, especially, conative (as opposed to cognitive) states, such as intentions, ends and the like. Hence, an intelligence report that Saddam Hussein is building WMDs does not in and of itself cause a kinetic response, e.g. war; rather the kinetic response depends on a decision to act (or not) based in part on the intelligence report but also and in part on some goal or end, e.g. to prevent Hussein from possessing an arsenal of WMDs, i.e. knowledge does not in and of itself generate kinetic action.

As mentioned above, there is a threefold distinction between the institutional, the macro and the micro levels. This threefold distinction brings with it a corresponding three-fold distinction between institutional purposes, macro-level purposes and micro-level purposes. Moreover, these three levels of purpose are nested and interactive, e.g. micro-level purposes can derive from macro-level purposes but, perhaps once acted upon, can also inform macro-level purposes. Further, the content of institutional purposes is relatively general whereas that of macro-level purposes more specific and that of micro-level purposes highly specific. Finally, as was mentioned above in relation to intelligence officers and decision-makers, the ongoing interactive process between these three levels and, in particular, between the formulation and re-formulation of purposes is not simply circular but also two-way or three-way interactive e.g. between strategic and tactical intelligence purposes.

The general point to be stressed here is that contrary to Collins’ gloss on my teleological account that “addressing Intelligence Collection only in relation to a pre-known purpose closes off preparatory intelligence gathering”,[22] on my account the content of the purposes of intelligence agencies, and of individual intelligence officers, range from general to specific, and are dynamic and interactive, both among intelligence officers themselves and in their interaction with decision-makers. There is no such closing off except in so far as there is the general institutional purpose of national security intelligence collection, analysis, and dissemination to decision-makers.

[1] Collins, Michael T. 2021. “Civil Service Intelligence Ethics: A Reply to Miller’s ‘Rethinking the Just Intelligence Theory of National Security Intelligence Collection and Analysis’.” Social Epistemology Review and Reply Collective 10 (9): 51-60.

[2] Miller, Seumas. 2021. “Rethinking the Just Intelligence Theory of National Security Intelligence Collection and Analysis: The Principles of Discrimination, Necessity, Proportionality and Reciprocity.” Social Epistemology 35 (3): 211-231.

[3] Collins 2021, 51.

[4] Miller 2021, 211.

[5] Miller 2021, 212.

[6] Miller 2021, 217.

[7] Miller 2021, 213-214.

[8] Miller 2021, 214.

[9] Collins 2021, 54.

[10] Collins 2021.

[11] Collins 2021, section 5.

[12] Collins 2021, 55.

[13] Miller, Seumas. Forthcoming. “National Security Intelligence Activity: A Philosophical Analysis” Intelligence and National Security.

[14] Miller 2021, 218.

[15] Collins 2021, 52.

[16] This material is drawn from Miller, Seumas. 2010. The Moral Foundations of Social Institutions: A Philosophical Study. New York: Cambridge University Press.

[17] Collins 2021, 52.

[18] Collins 2021, 53.

[19] Miller 2010.

[20] Miller, Seumas. 2015. “Joint Epistemic Action and Collective Responsibility” Social Epistemology 29 (3): 280-302.

[21] Miller 2010, 66-76; Miller, Seumas. 2016. “Police Officers, Regular Soldiers, and Normative Institutional Analysis.” Chap. 3 in Shooting to Kill: The Ethics of Police and Military Use of Lethal Force. Oxford: Oxford University Press.

[22] Collins 2021, 53.

Categories: Critical Replies

Tags: epistemic action, epistemic institutions, intelligence agencies, intelligence collection and analysis, joint epistemic action, Michael T. Collins, national security, national security analysis, national security intelligence ethics, Seumas Miller