Civil Service Intelligence Ethics: A Reply to Miller’s “Rethinking the Just Intelligence Theory of National Security Intelligence Collection and Analysis”, Michael T. Collins

Intelligence Collection—like tax collection, courts, Law enforcement, environmental regulation, etc.—is an exercise of government power.[1] Like other exercises of authority, Intelligence Collection must work within constraints (at least in countries bound by the rule of Law). Intelligence Collection is also—alongside scholarship, journalism, and Law—an epistemic, or truth-seeking institution, and its ethical considerations should reflect that knowledge is determined first and then brings about action … [please read below the rest of the article].

“And Ye Shall Know the Truth and the Truth Shall Make You Free” — John 8:32, as inscribed on the wall of the Old Headquarters Building of the U.S. Central Intelligence Agency.


Article Citation:

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.

🔹 The PDF of the article gives specific page numbers.

Seumas Miller recently examined Just Intelligence Theory as a means of setting ethical restraints on that power. (2021). This piece is an intriguing starting point and I agree with Miller that Just Intelligence Theory—derived from Just War Theory—does not adequately address the ethical dilemmas facing intelligence collectors (though it may show promise for evaluating discrete operations that mix collection with kinetic action). This realm of political philosophy is relatively new and requires deep examination.

I write here to suggest that the issue should be reframed. 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.

I suggest that the operative starting premise is that Intelligence Collection is an epistemic institution within civil governments. Rather than importing ideas from the exigencies of war, a theory of intelligence ethics should start with the political theories constraining all government action. These take the form of neutral predetermined rules and procedures decided in advance of the factual circumstances that dictate the purpose of any given government endeavor.

I offer, as example, the Law. Both Law an Intelligence Collection are epistemic exercises that can lead to non-epistemic consequences vis-à-vis the government actions facilitated or directed as a result of an epistemic finding. Law’s aspirational requirement of “neutral principles” in epistemic determinations and adherence to a tapestry of rules and processes independent of individual controversies, Can serve as an imperfect example of a process-based approach constraining government epistemology. A true intelligence ethos will, however, take its own form, as demonstrated by the stark differences between criminal and national security intelligence.

In short, a lawyer suggests an Intelligence Collection ethical framework modeled after process and procedure. Procedures inherently mean more work for lawyers. How do I justify this rent-seeking?

Knowledge as an End In Itself

We should reframe our view of Intelligence Collection. Miller narrows his focus to actionable intelligence—i.e., intelligence that facilitates informed action or inaction—(2021, 214). This narrowing omits the many rotations of the intelligence cycle that produces the background information leading to the point were a purposive collection could be contemplated.

First, the big picture is necessary for any intelligence to be actionable (see e.g., Tenet 2009, 107–09; discussing the impact of decreased capabilities across the entire intelligence community in relation to specific desired operations). An imperfect but helpful analogy describes Intelligence Collection as finding pieces of a jigsaw puzzle of the whole world (see Vandepeer 2014, 46). You are never sure the pieces are from the same set, they don’t quite fit, and the underlying image seems to move. Sometimes you can tell with relatively few pieces that a part of the puzzle is nothing but blue sky and may be saved for later. Other portions are really interesting and you will choose to start looking specifically for pieces that fit.

The pieces of the puzzle are scattered; found by way of the “INTs”: signals intelligence (SIGINT), intercepting electronic communication and breaking encryption; imagery intelligence (IMINT), images collected from satellites an aircraft; human intelligence (HUMINT), that collected from human contacts; measurement and signatures analysis (MASINT), a highly technical refinement of sigint and imint combined with spectral imagery to identify objects; and open source information (OSINT) analysis of information available to the general public (Erwin 2013, 15–17). These various sources provide raw information in the intelligence cycle—a revolving wheel of requests for information, leading to planning and implementing collection operations, leading to processing and exploiting information, leading to analysis and production of findings, leading to dissemination, leading to requests for more information and so on.

FBI Intelligence Cycle Graphic:

Second, addressing Intelligence Collection only in relation to a pre-known purpose closes off necessary preparatory intelligence gathering (acquiring information that would be necessary for future activities) and limits cross-purpose information sharing. And though purposes can be divided into distinct categories—e.g., criminal intelligence (“investigation, arrest and the prosecution of criminals”) (Miller 2021, 213), military intelligence (“winning wars, deterring military aggressors”) (213), and market intelligence (“gaining commercial advantage in a context market-based competition”) (213), these distinctions put the cart (or more accurately carriage) before the horse. The Intelligence Collection may lead to the stated purpose, but purpose drives rather than creates the Intelligence Collection itself. A driver must care for his horse on all journeys, but pushing the horse harder when time is of the essence is acceptable.

Moreover, a purpose-limited definition of intelligence presupposes that the decision-maker already has enough information to direct collection solely at a purpose. Purpose can be a part of the analysis (some practitioners suggest that individual agencies create mission specific codes of conduct; see Bailey and Galich 2012, 81–82)—a weight on the scales one way or another—but starting from purpose breaks the scales (for reasons discussed below).

A theory of ethics must address the whole cycle. When filling in the puzzle pieces of a particular situation, the collection has to start somewhere. Some puzzle piece must be found indicating that that portion of the whole may be worth concentrating on. And, in portions of the cycle, the purpose is merely finding more information. An ethical theory that accounts for the last step, but cannot account for the steps that got there will not help practitioners making initial investigative moves.

Purpose is one element underlying an eventual theory of intelligence ethics. And it is better to frame Intelligence Collection as having many purposes. This can address the common characteristics of Intelligence Collection—clarifying the puzzle picture.

Reaching the same conclusion from a different route, the amorphous “national interest” or “prosperity” definition of national security that Miller rejected as overly permissive (2021, 214), could take the place of purpose in this analysis. Given the broad range of goals covered in that definition, “national security purpose” would be a stand-in for an evolving variable in the calculous: N, where N represents the given moral weight to a specific manifestation of national security interests. Ex post, courts routinely way a variety of “compelling/important government interests” against the incidental or intentional infringement of rights; the degrees of scrutiny change with regard to the right at issue (see e.g., O’Brien, 391 U.S. 367; weighing the government interest in managing the selective service system by prohibiting destruction of draft cards against the implicated speech of draft card burning). And ex ante, a stated or manifest purpose is but one element of a proposed collection activity.

So, the analogy from war is limiting. Both war fighting and Intelligence Collection are both “activities that, nevertheless, stand in (roughly speaking) the relationship of knowledge to action[.]” Miller (2021, 216). But the end of war achieves a fixed and definite goal—cessation of conflict. War’s means are necessary evils (killing and maiming and breaking things). This end/means combination becomes good in the sense that it mitigates bad acts. Because, Intelligence Collection’s end is either not fixed, or fixed to an amorphous and permissive end, the collection itself must be the basis for an ethical theory.

Completing the Moral Balancing Test

Purpose breaks the balancing test, but its damage vanishes when collection independent of purpose is the starting point. A balancing grounded in the purpose of an activity is necessarily unanswerable until a result is reached. Miller (2021, 216). But the radical uncertainty would remain even after the result is known.

If the purpose is preventative of war, then only a failure can be weighed against the reality of the war that occurred. Success is judged against the imagined likely or worst-case scenario. As Giangiuseppe Pili notes (2021, 2–3), the counterfactual thumb on the scale renders a purpose-based analysis unworkable. And this problem is compounded by the existential threats possible in worst case scenarios. If the analysis is radically incomplete when the result of an Intelligence Collection operation is unknown, it necessarily remains radically incomplete if the scope of an averted catastrophe is also unknown.

From the other direction, this objection does not account for the general value of reduced uncertainty. It is not the case that an unknown value makes a calculation radically incomplete. Miller (2021, 216). Rather an unknown value, attached to the various known values from previous Intelligence Collections provides the bounds of moral value. Consider Miller’s (2021) dirty bomb example:

An intelligence unit is seeking to find out whether a terrorist group has the capability of producing a radiological ‘dirty’ bomb (a weapon of mass destruction or WMD) and, therefore, ‘turns’ a disaffected member of the terrorist group (Informant) and tasks him to get this information; an extremely dangerous job which puts the life of Informant at serious risk. Informant discovers that the terrorist group has no such capability and has no plans to develop it. Unfortunately, however, informant’s Intelligence Collection activities have raised suspicions and shortly thereafter he is murdered by the terrorist group (216).

The result is unknown, but the possible results are predictable. Either:

(a) there is a bomb;
(b) there are the makings of a bomb;
(c) there is no bomb, or;
(d) the results are inconclusive, necessitating further information.

(a) and (b) have obvious epistemic and moral value; there is or will be a bomb that authorities may address. (c)’s value is less obvious, but falsifying a hypothesis with certainty allows redeployment of resources and avoids potential panic from an unnecessary evacuation. Still valuable, (d) shows the need for further information. All have different levels of value, but the possibilities set upper and lower bounds to how much the information is worth the risk. Also important is the previous Intelligence Collection that led to the dirty bomb hypothesis.

If purpose makes analysis radically incomplete with successful results because the cost of failure remains unknowable, but a range of value-determinative results can be controlled for before an operation, then knowledge itself—of the result or possible results—dictates the Intelligence Collection’s moral value. Moreover, a situation allowing this type of calculous arises out of a foundation of existing intelligence.

A government must have intelligence already to get to this point, and knowledge dictates the moral valuation of a collection activity. This also suggests that the functional acquisition of knowledge is the best groundwork for a moral framework.

Intelligence Collection, unlike war, is not something governments do when the need arises. Intelligence Collection is simply something governments do.

Epistemic Institutions

As such, Intelligence is an epistemic institution with an ongoing epistemic mission. Jonathan Rauch identifies four key sectors of the “reality-based community”—scholarship, government (including intelligence), journalism, and Law—that principally seek truth (116–19). In seeking truth, these institutions are bound by formal and informal rules of conduct—the titular constitution. Truth requires an institutional rather than personal approach because, as David Hume stated: “Reason is, and ought only to be slave to the passions, and can never pretend to any other office than to sever and obey them.” (1739–40, Bk. 2, Pt. 3, § 5; see Rauch 2021, 33; for a psychological analysis of this phenomenon see Haidt 2012, 25). Using reason to seek truth requires institutional norms that encourage disconfirmation of passionate ideas. Rauch (2021, 121). In all institutions some concept of ethics sets bounds on the epistemological mission. Intelligence Collection can draw on this body of knowledge.

Toward that purpose, Intelligence Collection has much in common with Law. This is unsurprising because each exercises government power.[2] In both institutions, the epistemic mission is cabined by strict adherence to the constraints limiting the powers of legitimate sovereigns in relation to their citizens and to those of foreign nations. And each can pursue their missions through means that are acceptable when governments, but not ordinary citizens act. Manget (2006, 429­–34). A citizen may really want Ms. Smith to tell what she saw on the day of a car accident; only a court may compel her to testify.

Both proscribe methods of obtaining information that violate higher ideals (for Law see U.S. Const. amends. IV–VI; for intelligence gathering see Executive Order 13491). Both have jurisdictional boundaries limiting areas of investigation—e.g., federal district courts are limited in the matters they may entertain, see 28 U.S.C. §§ 1330–69, and the CIA only operates outside the United States, 50 U.S.C. § 3036(b). Each seeks veracity. Federal courts follow rules respecting a morally neutral prearranged determination about the veracity of certain information; see e.g., Fed. R. Evid. 801 (prohibiting hearsay); Fed. R. Evid. 803 (listing exceptions) while human intelligence collectors have many pre-approved techniques for identifying deceitful sources (see FM 2-22.3; 2006 ¶ 9-26). Together, these systems weave together moral and epistemological considerations while setting limits on the institutions’ authority to entertain questions.

These rules, as well as their counterparts in scholarship (i.e., use of the scientific method or requiring conflict of interest statements) and journalism (see e.g., New York Times 2019) model the great Madisonian compromise in American political theory. Rauch (2021, 97). These institutions are comprised of people, and people on a whole may charitably be described as not “angels” (Federalist No. 51, Madison). James Madison’s insight was that institutions may be obliged to control themselves if the governing rules allow competing interests to serve as a check on the institution as a whole (Federalist No. 51, Madison).

This competitive check is obvious in Law where, ideally, counsel for one party must assume that any ethical, factual, procedural, or legal lapse on his part will be caught and exploited by the opposing party. Scholarship is also rife with attempts to debunk currently prevailing theories; the scientific method is premised on testing hypotheses under the scrutiny of peer review. Intelligence Collection is more co-operative, but as both an epistemic institution and one wielding government power, a useful starting point may be to look to the structuralist forces that compel honesty and adherence to norms in Law.

Structuralist Approaches to Government Functions

Intelligence Collection is a routine government function (albeit a rather interesting one). Its moral weight is tied in part to that of the state. Pili already examined Intelligence Collection in light of the possibility of a state’s legitimacy slippage (2021, 3–4). But as Madison warns, it is necessary to constrain legitimate, but imperfect, states (Federalist No. 51, Madison). Just Intelligence Theory (specifically necessity and proportionality requirements), starting from the micro level and working up to the macro level, identifies workable principles in individualized collection situations and scales up. Miller (2021, 224–25). But if Intelligence Collection is viewed as an epistemic institution that acts with the authority of government, a better approach may be to start with a theory of government action and work down. Toward this end, the example of neutral principals as a top-down epistemic and ethical constraint may prove a more inclusive and applicable theory.

Law faces two layers of epistemic questions: those of Law and of fact. For the former, arguments are settled by courts which have the last say on the meaning of Law (Marbury, 5 U.S. at 177; Wechsler 1959, 6–7). Legal decisions are made by people with biases and flaws. The “legal process school” holds that court decisions, regardless of the presence of value judgments, must “rest[] on reasons with respect to all the issues in the case, reasons that in their generality and their neutrality transcend any immediate result that is involved.” Wechsler (1959, 19). A competitive check exists when critics of a decision challenge its reasoning. This is rooted in the Judiciary’s unique place within the Madisonian schema—the Legislature controls the public money, the Executive wields the power to act, but, as Alexander Hamilton explained:

The judiciary, on the contrary, has no influence over either the sword or the purse; no direction either of the strength or of the wealth of the society; and can take no active resolution whatever. It may truly be said to have neither force nor will, but merely judgment; and must ultimately depend upon the aid of the executive arm even for the efficacy of its judgments (Federalist No. 78).

An epistemic institution, the Judiciary must convince others to act. Much like Intelligence Collection is an epistemic predicate to further actions.

More than merely explaining a method of interpretation, this comports with broader “Rule of Law” principles (i.e., the theory that rules governing individuals are generally applicable and that government discretion should be limited, see Aristotle, Politics, Book III, ch. xi). Neutral, predictable, rules of interpretation set limits on the discretion available to magistrates and allow parties to predict how the Law will apply to their case and persuade courts that their interpretation is in line with these principles. Though many argue vehemently about what these principles should be in big interpretive cases, the concept has more teeth in everyday rules of evidence and procedure that govern Law’s second epistemic realm—questions of fact.

In rule-bound fact-finding, rules of evidence and of discovery and of how facts may be presented choreograph the fact-finding process. It is imperative to both the legitimacy of outcomes and to the efficient functioning of the court system that the rules governing what is hearsay, or how cross-examination is to be conducted, or the burden of proof applied to certain claims, remain constant or ad least knowable from case to case. These are not moral judgments. Indeed, the rules of evidence differ state to state. And entirely different systems function perfectly well in different countries (there is no moral reason that an anglosphere-type, adversarial system is better than the inquisitorial system in France). What matters is that the rules governing the epistemic exercise are knowable and explainable, and somewhat predictable, to the point that violations are challengeable.

For Intelligence Collection, a pre-conceived, macro-level, procedural framework of ethical decision-making may be more conducive than a micro-level balancing test such as Just Intelligence Theory. First, as government actors, Intelligence Collection is already amenable to rules-based approaches. Second, pre-deciding certain issues lesson’s the impact of hypothetical worst-case scenarios (Intelligence Collections version of the personal biases of lawyers). And third, rules apply equally whether the purpose of a collection activity is directed at a specific event or merely exploratory.

The Criminal Intelligence Process

Though I suggest a process-based approach to intelligence ethics, criminal intelligence—while facially similar—is a poor comparator. Outside of a few overlapping areas, Intelligence Collection is a fundamentally present and future-looking exercise while court processes almost universally look to the past (Manget 2006, 415–16).[3]

In most cases, criminal intelligence is gathered after the fact and used to determine guilt. If a crime has been committed, the damage is already done. The accused is presumed innocent until proven guilty. And improperly obtained evidence can be excluded from the trial making it more difficult to prove the accused guilty—the “exclusionary rule” in the warrant context. (see generally Mapp, 367 U.S. 643). While this limits a trial’s epistemic capabilities by excluding otherwise relevant evidence, it preserves the broader moral value of forbidding unreasonable searches. Ex post, the legal process imposes a consequence of inability to punish against the violation of rules governing investigations. The state’s job is harder, society as a whole may be in some vaguely increased danger, and the accused reaps a small benefit. The direct consequences lay with the parties involved: the state and the accused, who is at this point innocent.

But ex anti, where Intelligence Collections live, something like the exclusionary rule would be unworkable and morally questionable. An intelligence briefer should not omit damning evidence of an imminent attack simply because it was improperly obtained. The direct consequences are borne by the preventable victims.[4] And the perpetrator, instead of escaping punishment, is harmed because he was allowed to commit such an atrocity (possibly leading to a far greater punishment than he would have otherwise earned). Indeed, this distinction is present in the doctrine of exigent circumstances—exceptions to the general search warrant requirement when quick action is needed, e.g., responding to an ongoing crime or emergency (see generally Brigham City, 547 U.S. 398).

❧ ❧ ❧

Intelligence Collection is a fact of international relations. It has long since moved past its starting point as an ancillary activity to war. Pili (2021, 2). Just Intelligence Theory is a flawed approach but serves as an example of why the ethics of Intelligence Collection, separate from its purposes, deserves further exploration. I think viewing Intelligence Collection through the theories constraining governments generally, rather than just during wartime, would be a beneficial next step in this process. The Law provides an example of how this constraining feature appears in other government contexts.

Author Information:

Michael T. Collins,, is an attorney in Washington, D.C. The views expressed here do not represent those of any employers, affiliates, or clients past, present, or future.


Bailey, Christopher E. and Susan M. Galich. 2012. “Codes of Ethics: The Intelligence Community.” International Journal of Intelligence Ethics 3 (2): 77–99.

Carlin, John P. and Garrett M. Graff. 2018. Dawn of the Code War: America’s Battle Against Russia, China, and the rising Global Cyber Threat. New York: PublicAffairs.

Erwin, Marshall Curtis. 2013. Intelligence Issues for Congress. Washington: Congressional Research Service.

Haidt, Jonathan. 2012. The Righteous Mind: Why Good People are Divided by Politics and Religion. New York: Pantheon Books.

Hamilton, Alexander, John Jay, and James Madison. The Federalist Papers. New Haven: Avalon Project. https://avalon. (per legal conventions cited in text as Federalist No. XX author).

Hume, David. (1739–40) 2011. A Treatise on Human Nature. Reprint in David Hume: The Essential Philosophical Works. Hertfordshire: Wordsworth Editions.

Manget, Fred F. 2006. “Intelligence and the Criminal Law System.” Stanford Law and Policy Review 17 (2) 415–36.

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.

New York Times. 2019. “Ethical Journalism: A Handbook of Values and Practices for the News and Editorial Departments.” Accessed September 15, 2021.

Pili, Giangiuseppe. 2021. “The Missing Dimension—Intelligence and Social Epistemology: A Reply to Miller’s ‘Rethinking the Just Intelligence Theory of National Security Intelligence Collection and Analysis’.” Social Epistemology Review and Reply Collective 10 (7): 1–9.

Rauch, Jonathan. 2021. The Constitution of Knowledge: A Defense of Truth. Washington: Brookings.

Tenet, George. 2009. At the Center of the Storm: The CIA During America’s Time of Crisis. New York: HarperCollins.

Wechsler, Herbert. 1959. “Toward Neutral Principles of Constitutional Law.” Harvard Law Review. 73 (1): 1–35.

Vandepeer, Charles. 2014. Applied Thinking for Intelligence Analysis: A Guide for Practitioners. Canberra: Air Power Development Centre.

Legal Authorities


US Constitution. https://www.

Department of the Army. 2006. Field Manuel 2-22.3: Human Intelligence Collector Operations. Fort Belvoir, VA: Army Publishing Directorate. (cited as FM 2-22.3).

Executive Order 13491, Ensuring Lawful Interrogation. (2009) 2020. Reprinted in U.S. Office of the Director of National Intelligence, Office of General Counsel. Intelligence Community Legal Reference Book. Washington: Government Printing Office. 771–75. Legal Reference Book 2020.pdf

Federal Rules of Evidence. https://www.

United States Code. https://www. (Title 18 contains criminal Law while most Law governing Intelligence Collection is found in Title 50. Title 5, governing civilian government action, and Title 10, governing military actions, may also be of note to those interested in the underlying legal framework.).


Brigham City v. Stuart, 547 U.S. 398 (2006):

Marbury v. Madison, 5 U.S. (1 Cranch) 137 (1803):

Mapp v. Ohio, 367 U.S. 643 (1961):

Shelley v. Kraemer, 334 U.S. 1 (1948):

United States v. O’Brien, 391 U.S. 367 (1968):

[1] Following Miller’s (2021, 215) lead, this discussion will confine itself to “national security intelligence collection, analysis and dissemination.” Contra Miller’s scope (212–213), these activities are viewed more to the full extant.

[2] Court decisions, even in civil suits, carrying the government’s imprimatur its accompanying force. Courts, therefore, will not, for instance, enforce racially restrictive covenants, Shelley, 334 U.S. 1.

[3] As President George W. Bush asked FBI Director Robert Mueller after being briefed on the criminal investigation into the September 11th attacks: “What are you doing to stop the next attack?” (Recounted in Carlin and Graff, 2018, 138).

[4] Certain areas of criminal law are more proactive, but the general rules are backward facing.

Categories: Critical Replies

Tags: , , , , , , , ,

Leave a Reply