PUBL0267 - Blog

Mirror, Mirror at War: When Intelligence Agencies Mistake Themselves for the Enemy

Western intelligence agencies systematically attribute adversary vulnerabilities they simultaneously share, driven by three structural incentives: narrative pressure, accountability asymmetry, and confirmation cascades, with five recent cases proving the pattern.

By Frederic Lemieux, PhD. & Shay Hershkovitz, PhD.

The Briefing That Ages Badly

There is a particular species of intelligence assessment that sounds authoritative when delivered, ages poorly under scrutiny, and is almost never subjected to retrospective accountability. It goes something like this: our adversary is running out of X. Their logistics are collapsing. Their industrial base cannot sustain the tempo. Their morale is broken. Give it a few months.

Since February 2022, Western governments, defense officials, and analytical institutions have delivered dozens of such assessments about Russia, the Houthis, Iran, and Hezbollah. Some proved correct. Many did not. And in a striking number of cases, the vulnerability attributed to the adversary turned out to describe a problem the assessing party was simultaneously confronting on its own side of the ledger – often at the very moment the claim was being made. Both of us spent years producing the kind of assessments described here. The pattern below is not a theoretical concern: it is operational.

“This is mirror imaging: the cognitive error of projecting one’s own logic, constraints, and vulnerabilities onto an adversary.”

Intelligence professionals have warned against it for decades. Robert Jervis, whose foundational work on political psychology and misperception [1], remains the field’s essential text, identified the root problem: states perceive adversaries through their own institutional, economic, and normative filters, and assuming adversaries share your constraints is among the most persistent sources of strategic error. The argument here is that this error is not primarily cognitive: it is structural. The incentive architecture of Western intelligence production systematically rewards finding the adversary’s vulnerabilities and penalizes finding your own.

What follows examines five cases from recent conflict theaters (Ukraine, Yemen, Lebanon, and Iran) where the adversary narrative and the mirror reality diverged in ways that mattered strategically. A sixth case, from Venezuela, illustrates a related but distinct failure: not mirroring an adversary’s vulnerabilities but mirroring an adversary’s conduct while underestimating how the international community would respond.

Together, they reveal a pattern consistent enough to warrant structural reform.

The Ammunition Case: The Most Operationally Damaging Mirror

In October 2023, NATO Military Committee Chair Admiral Rob Bauer told the Warsaw Security Forum something that deserved substantially more analytical attention than it received. “We now give away weapons systems to Ukraine, which is great, and ammunition – but not from full warehouses,” [2] he said. “We started to give away from half full or lower warehouses in Europe. And therefore, the bottom of the barrel is now visible.” He repeated the warning at the NATO-Industry Forum in Stockholm weeks later.

The timing was not incidental. At precisely the same period, Western officials were citing Russian ammunition shortages as evidence that the Kremlin’s war machine was approaching culmination. An internal U.S. estimate reported by the Washington Post in December 2023 [3] warned that Ukraine itself would face potentially catastrophic artillery ammunition shortages by late March 2024. National Security Adviser Jake Sullivan stated publicly in March 2024 “When Russian troops advance and its guns fire, Ukraine does not have enough ammunition to fire back.” [4]

Congressional Research Service reporting on U.S. security assistance to Ukraine [5] documented the pace and scale of ammunition drawdowns throughout this period. RAND analysis of NATO deterrence posture and readiness [6] had previously raised [7] stockpile sustainability concerns at alliance level.

Admiral Bauer was not issuing a classified warning to a restricted audience: he was making a public declaration to a major multinational security forum attended by defense officials from across the alliance.

The analytical failure is specific. Western intelligence produced and circulated an assessment that Russia was running dry during a period when NATO’s own stocks were running dry – and when evidence of that parallel problem was available to the same institutions producing the adversary assessment. The finished intelligence products of that period on Russian ammunition sustainability do not appear to have incorporated Admiral Bauer’s warning as a corrective or qualifying factor.

As Richards J. Heuer Jr. established in Psychology of Intelligence Analysis [8], once an analyst’s credibility becomes invested in a narrative, contradicting evidence is processed through that narrative rather than against it. The ammunition case is a textbook instance of Heuer’s framework at institutional scale: the contradicting evidence was public, specific, and senior. No one integrated it.

The Manpower Case: Symmetry Without Recognition

Throughout 2023, Western analysts pointed to Russia’s resort to convict recruitment as a signal of acute manpower desperation. The imagery was politically useful and analytically convenient. The parallel dynamic unfolding on Ukraine’s side of the line received substantially less attention in finished assessments.

By 2024, Ukraine’s mobilization crisis had become severe enough to generate a new entry in Ukrainian lexicography. “Busification“[9], the coercive street-level seizure of draft-age men and their forced transfer to mobilization centers, was recognized as Ukraine’s Word of the Year for 2024. Officers were documented raiding public venues [10], including a major concert in October 2024. Ukrainian courts filed more than 89,000 desertion and abandonment proceedings [11] in 2024 (3.5 times the 2023 figure) with independent estimates suggesting more than 20 percent of Ukraine’s million-strong army had deserted over four years of war.

ACLED conflict data [12] for the 2023 – 2024 period document the sustained intensity of fighting along the eastern front, providing quantitative context for these manpower figures: the desertion and mobilization numbers reflect combat losses and institutional strain under conditions of continuous high-tempo engagement, not administrative attrition in a static environment. Chatham House analysis of Russian mobilization constraints [13] throughout the same period documented comparable strains on the Russian side. Yet the symmetry between the two sides’ mobilization pressures was rarely represented as such in Western analytical products.

Atran documented repeated U.S. capability assessment failures following this pattern: analysts underestimated Ukrainian will to fight before February 2022, overestimated Iraqi Army capabilities in 2014, and underestimated Viet Cong sustainability. His finding, that material metrics cannot reliably predict will to fight, yet remain the default assessment framework, applies directly here. The intelligence community assessed Russian manpower through the lens of convict recruitment and concluded exhaustion. It did not systematically apply the same metric to Ukraine’s mobilization crisis as it was simultaneously developing.

The Drone Case: When Doctrine Meets a $50,000 Drone

The Pentagon spent decades optimizing for precision overmatch: expensive platforms, expensive munitions, and supply chains calibrated for high-end peer competitors. That doctrine met its cost-asymmetry reckoning in the Red Sea.

William LaPlante, Undersecretary of Defense for Acquisition and Sustainment, testified before a Senate Appropriations subcommittee on May 15, 2024 [14]: “If we’re shooting down a $50,000 one-way drone with a $3 million missile, that’s not a good cost equation.” C4ISRNet coverage from April 2024 [15] had made the exchange-ratio problem publicly visible weeks before formal testimony. Research on counter-drone defense [16] and magazine-depth challenges [17] had previously identified the structural vulnerability of high-cost interceptor inventories against low-cost saturation threats.

The Houthis had made this precise argument in their public statements throughout the Red Sea campaign. Skepticism toward adversarial public claims is analytically warranted; the analytical obligation is to assess whether the underlying exchange-ratio data supports the claim independently of its source. The data did. The failure here was doctrinal as well as cognitive: the United States had spent years assessing adversary capabilities through the lens of its own acquisition culture, assuming that the cost constraints inherent to its own procurement model represented terminal constraints for adversaries operating at a fraction of the price. That assumption inverted when applied to the Red Sea’s actual exchange ratios.

The Iran Nuclear Case: Mirror Imaging as Foreclosure

Of the cases documented here, the Iran nuclear diplomacy episode may be the most consequential, because it illustrates mirror imaging in its most dangerous form: not a forecasting error, but an interpretive structure that forecloses alternatives before they can be tested.

On February 28, 2026, Oman’s Foreign Minister Badr bin Hamad Al Busaidi announced publicly that a diplomatic breakthrough was within reach. Iran had agreed in principle to zero stockpiling of enriched uranium, full IAEA verification access, and conversion of existing stockpiles to fuel under international supervision. “Peace deal is within our reach,” Al Busaidi stated [18], with follow-on talks scheduled within 48 hours. CBS News [19] and The Hill [20] reported the announcement the same day.

Coordinated U.S.-Israeli strikes targeting Iran’s missile arsenal, naval assets, and nuclear infrastructure commenced the same day. Carnegie Endowment for International Peace [21] had tracked both the diplomatic channels and the parallel enrichment activities in the preceding period, providing the factual baseline against which the competing assessments of Iranian intent could, in principle, have been evaluated.

The official U.S. assessment justifying the operation held that Iran had been negotiating in bad faith and that the diplomatic track served as cover for continued weapons development. Whether that assessment was accurate is a question the available record cannot resolve: the diplomatic alternative was foreclosed before it could be tested. What the record does establish is the proximity (to the hour) between the Omani foreign minister’s public announcement and the commencement of strikes. Heuer’s framework describes this failure mode precisely: once a hypothesis achieves operational status, contradicting evidence tends to be reweighted to confirm the prevailing assessment rather than examined on its merits. The Oman announcement was public, verifiable, and temporally coincident with the strikes. Whether it received serious treatment as contradicting evidence, and what the sequence was between the intelligence assessment and the operational decision, are the kinds of questions that retrospective review mechanisms are designed to answer. They are also precisely the questions existing mechanisms rarely ask.

The Venezuela Case: Mirroring Conduct, Not Just Vulnerability

The Venezuela case adds a dimension the preceding cases do not. The United States did not mirror an adversary’s vulnerabilities; it mirrored an adversary’s conduct and the available record suggests the breadth of the diplomatic response was not anticipated at the scale it materialized.

On January 3, 2026, U.S. forces executed Operation Absolute Resolve, a large-scale military operation culminating in the capture of Venezuelan President Nicolás Maduro and his transfer to U.S. custody. The Stimson Center [22] and CSIS [23] documented the scope and regional implications in detail.

Within weeks, Brazil, Colombia, Mexico, and Chile had issued diplomatic protests invoking arguments structurally identical to those the United States deployed against Russia’s 2022 invasion: sovereignty, the prohibition on the use of force under the UN Charter, the illegality of regime change without Security Council authorization. The UN Secretary-General called the operation “a dangerous precedent.” Atlantic Council analyses [24] of Latin American responses documented the breadth and coherence of the regional reaction. The language of the protests was familiar because it was, in substantial part, the language the United States had authored and applied against Russia four years earlier.

The analytical failure here is distinct from the preceding cases. It is not Heuer’s confirmation bias operating on intelligence products about adversary material capabilities; it is what may be termed strategic ethnocentrism operating on diplomatic assessment. The assumption that the normative frameworks the United States deploys against adversaries carry asymmetric political force, that sovereignty and non-intervention arguments register differently when invoked by Washington than when invoked against it. The regional response demonstrated that assumption was not supported. The mirror here was normative rather than material, but it was a mirror.

Why the Incentive Structure Produces This Outcome

Mirror imaging, whether cognitive or normative, persists not because analysts are incompetent. It persists because the incentive structure rewards it. Three mechanisms operate consistently across the cases above.

Narrative coherence pressure. In wartime conditions, governments require confident assessments to sustain domestic and allied support. “The adversary is running out of X” is a useful narrative. “We are also running out of X” is not. Institutional pressure does not typically suppress the second finding outright. It creates conditions in which the second finding is not sought, or, having been found, is categorized as a logistics or procurement problem rather than an intelligence assessment. Admiral Bauer’s warning existed in the public record. Its analytical implications for Russian sustainability assessments were not incorporated.

Accountability asymmetry. Assessments about adversary vulnerabilities are rarely subjected to systematic retrospective review. The leaked U.S. estimate warning of Ukraine’s “potentially catastrophic” ammunition crisis received close scrutiny once published. The adversary-vulnerability assessments that accompanied it did not. The institutional cost of producing adversary-vulnerability claims that prove incorrect is low. As a 2009 review of the CIA’s internal postmortem on the Iraq WMD assessment documented [25], institutional self-examination tends to produce conclusions that stop short of implicating the assessment process itself.

Confirmation cascades. When multiple agencies, think tanks, and allied services draw on overlapping source streams, assessments about adversary weakness achieve apparent consensus rapidly. That consensus becomes harder to challenge not because the underlying evidence has improved but because the social cost of dissent has risen. Chesterfield and colleagues [26] documented in 2025 how “cultural capture” in large organizations produces precisely this dynamic: those with divergent views exit, dissenters face institutional pressure, and the consensus hardens independent of evidentiary improvement. The intelligence community is not a regulatory agency, but the structural parallel is uncomfortable.

Mirror imaging is the diagnosis. These three mechanisms are the institutional channels through which it persists despite being well-documented in the intelligence studies literature. Reducing cognitive bias through analytical training and structured techniques has shown limited effect at institutional scale. Changing the incentive structure is harder and more consequential.

A reasonable objection holds that both sides simply faced shared wartime constraints on industrial capacity – and that is partially true. But it does not explain why the same institutions assessing Russian supply never incorporated Admiral Bauer’s public warning as a corrective factor into their finished products.

What Should Change

Three reforms address these root causes directly. The Director of National Intelligence should amend the intelligence community’s governing analytic tradecraft standards to require that any finished assessment projecting a significant adversary vulnerability include a mandatory parallel assessment of whether the same vulnerability exists on the United States or allied side. The National Intelligence Council’s red cell function has the institutional standing to do this; what it lacks is permanent tasking. Red cells are currently activated at senior leadership discretion. That needs to change.

The FY2027 Intelligence Authorization Act should direct the Intelligence Community Inspector General to maintain a rolling database of major adversary-vulnerability assessments and their empirical outcomes, reviewed annually and reported to the intelligence committees in a classified annex. The 9/11 Commission [27] and the Robb-Silberman Commission [28] both identified this gap; neither built a permanent mechanism. The FY2027 authorization cycle is the vehicle.

Finally, the Office of the Director of National Intelligence should issue a formal policy memorandum designating which analytical products serve strategic communication functions (sustaining allied resolve, justifying resource requests) and which serve decision-support functions, with separate review chains for each. When the two collapse into the same document, the communication requirement shapes the analytical content, and mirror imaging stops being a failure mode and becomes a design feature.

The Mirror Does Not Lie

None of this is to suggest that the adversaries documented here were not facing genuine vulnerabilities. Russia’s economy has deteriorated under the weight of sustained wartime expenditure. Iranian military capacity is significantly degraded by the American and Israeli strikes. The Houthis’ capabilities, while resilient, were not unlimited.

The point is that intelligence assessments which systematically locate vulnerabilities in adversaries while overlooking equivalent vulnerabilities at home, or which assume that shared normative frameworks carry asymmetric political force, produce not just analytical errors but strategic miscalculations. They contribute to ammunition procurement decisions made on the assumption that the adversary will exhaust its supply first. They contribute to diplomatic decisions premised on the assumption that the adversary’s negotiating position lacks genuine content. They contribute to military operations planned on the assumption that international reaction will be manageable.

“The mirror at war is not a neutral instrument. It reflects what institutions are structured to see. “

The discipline of intelligence analysis exists precisely to interrupt that structural preference. Five cases of cognitive mirror imaging and one case of normative mirror imaging, across four conflict theaters, over four years, each with verifiable and publicly reported sourcing, suggest the interruption is not occurring with sufficient consistency or consequence.

The next briefing that tells you the adversary is running out of something would be well-served by a second slide: and here is what we checked on our own side.

REFERENCES

[1] Jervis, Robert. Perception and Misperception in International Politics. New ed. Princeton: Princeton University Press, 2017.

[2] Lendon, Brad, Radina Gigova, Fred Pleitgen, and Kostyantin Gak. “Western Ammo Stocks at ‘Bottom of the Barrel’ as Ukraine War Drags On, NATO Official Warns.” CNN, October 4, 2023. https://www.cnn.com/2023/10/04/europe/uk-nato-ukraine-war-ammunit.

[3] O’Grady, Siobhán, David L. Stern, and Kostiantyn Khudov. “With Western Aid Stalled, Ukrainian Troops Run Low on Artillery Shells.” Washington Post, December 22, 2023. https://www.washingtonpost.com/world/2023/12/22/ukraine-ammunitio.

[4] Kube, Courtney, Carol E. Lee, and Monica Alba. “White House to Provide $300 Million Weapons Package for Ukraine.” NBC News, March 12, 2024. https://www.nbcnews.com/politics/white-house-provide-300-million-w.

[5] Arabia, Christina L., Andrew S. Bowen, and Cory Welt. “U.S. Security Assistance to Ukraine.” CRS In Focus IF12040. Washington, DC: Congressional Research Service, May 22, 2024. https://www.congress.gov/crs-product/IF12040.

[6] Dee, Stuart, James Black, and Lucia Retter. “Resourcing the Ramp-Up: NATO and the Challenge of a Coherent Industrial Response to Russia’s War in Ukraine.” Commentary. Santa Monica, CA: RAND Corporation, July 25, 2024. https://www.rand.org/pubs/commentary/2024/07/resourcing-the-ramp-u.

[7] Shalom, Menny. “The Primes Aren’t the Real Bottleneck in U.S. Weapons Production.” War on the Rocks, January 26, 2026. https://warontherocks.com/2026/01/the-primes-arent-the-real-bottleneck.

[8] Heuer, Richards J., Jr. Psychology of Intelligence Analysis. Washington, DC: Center for the Study of Intelligence, Central Intelligence Agency, 1999. https://www.cia.gov/resources/csi/books-monographs/psychology-of-int.

[9] Dubynyansky, Mikhail. “‘Mogilyzation’ VS ‘Busification.'” Ukrainska Pravda, September 21, 2024. https://www.pravda.com.ua/articles/2024/09/21/7476081/.

[10] Hewson, Jack. “Lacking Manpower, Ukraine Resorts to Harsh Means to Force Draft Dodgers into Combat.” PBS NewsHour, November 27, 2024. https://www.pbs.org/newshour/show/lacking-manpower-ukraine-resorts.

[11] Proud, Ian. “Ukraine’s ‘Busification’ — Forced Conscription — Is Tip of the Iceberg.” Responsible Statecraft, November 4, 2025. https://responsiblestatecraft.org/ukraine-recruitment-army/.

[12] Armed Conflict Location and Event Data Project (ACLED). “Ukraine Conflict Monitor: Russia-Ukraine War Map.” Accessed April 10, 2026. https://acleddata.com/monitor/ukraine-conflict-monitor.

[13] Mathieu Boulègue and Justin Bronk, “Assessing Russian Plans for Military Regeneration: Modernization and Reconstitution Challenges for Moscow’s War Machine,” Research Paper, Chatham House, July 9, 2024, https://www.chathamhouse.org/2024/07/assessing-russian-plans-military-regeneration

[14] William A. LaPlante, “Department of Defense Acquisition Programs,” statement before the U.S. Senate Committee on Appropriations, Subcommittee on Defense, May 15, 2024.

[15] Colin Demarest, “Drone-Killing Costs Must Come Down, Says Pentagon’s Chief Weapons Buyer,” C4ISRNET, April 25, 2024, https://www.c4isrnet.com/unmanned/uas/2024/04/25/drone-killing-costs-must-come-down-says-pentagons-chief-weapons-buyer/.

[16] James Black, “David vs. Goliath: Cost Asymmetry in Warfare,” Commentary, RAND Corporation, March 2025, https://www.rand.org/pubs/commentary/2025/03/david-vs-goliath-cost-asymmetry-in-warfare.html.

[17] Jerry McGinn, “Magazine Breadth — Not Just Depth — Is Key to Munitions Industrial Base Resilience,” War on the Rocks, March 12, 2026, https://warontherocks.com/2026/03/magazine-breadth-not-just-depth-is-key-to-munitions-industrial-base-resilience/.

[18] Al Jazeera Staff and AFP, “Peace ‘within Reach’ as Iran Agrees No Nuclear Material Stockpile: Oman FM,” Al Jazeera, February 28, 2026, https://www.aljazeera.com/news/2026/2/28/peace-within-reach-as-iran-agrees-no-nuclear-material-stockpile-oman-fm.

[19] Margaret Brennan and Joe Walsh, “U.S.-Iran Deal Is ‘within Our Reach,’ Omani Mediator Says,” CBS News, February 27, 2026, https://www.cbsnews.com/news/us-iran-deal-within-our-reach-oman-mediator-says/.

[20] Laura Kelly, “Iran Agreed to ‘Zero Stockpiling’ of Nuclear Material in US Talks: Omani Foreign Minister,” The Hill, February 27, 2026, https://thehill.com/policy/international/5759623-iran-nuclear-deal-stockpiling-omani-foreign-minister/.

[21] Nicole Grajewski, “U.S. Aims in Iran Extend Beyond Nuclear Issues,” Diwan (blog), Carnegie Endowment for International Peace, February 25, 2026, https://carnegieendowment.org/middle-east/diwan/2026/02/us-aims-with-iran-extend-beyond-nuclear-issues.

[22] Benjamin N. Gedan, “What Will ‘Operation Absolute Resolve’ Mean for Venezuela and Latin America?” Stimson Center, January 4, 2026, https://www.stimson.org/2026/what-will-operation-absolute-resolve-mean-for-venezuela-and-latin-america/.

[23] Henry Ziemer, “The Geopolitics of Maduro’s Capture: What Does Operation Absolute Resolve Mean for Russia?” Critical Questions, Center for Strategic and International Studies (CSIS), January 20, 2026, https://www.csis.org/analysis/geopolitics-maduros-capture-what-does-operation-absolute-resolve-mean-russia.

[24] Atlantic Council, “Latin America,” Atlantic Council, accessed April 10, 2026, https://www.atlanticcouncil.org/region/latin-america/.

[25] James P. Pfiffner, “The Contemporary Presidency: Decision Making in the Bush White House,” Presidential Studies Quarterly 39, no. 2 (June 2009): 363–384, https://doi.org/10.1111/j.1741-5705.2009.03679.x.

[26] Alexandra M. Chesterfield, Tom W. Reader, and Alex Gillespie, “Cultural Capture Among Regulators: A Systematic Review,” Regulation & Governance, Early View, first published June 8, 2025, https://doi.org/10.1111/rego.70040.

[27] National Commission on Terrorist Attacks Upon the United States, The 9/11 Commission Report: Final Report of the National Commission on Terrorist Attacks Upon the United States (Washington, DC: U.S. Government Printing Office, 2004), https://www.9-11commission.gov/report/911Report.pdf.

[28] Ellen Laipson, “The Robb-Silberman Report, Intelligence, and Nonproliferation,” Arms Control Today 35, no. 5 (June 2005), https://www.armscontrol.org/act/2005-06/features/robb-silberman-report-intelligence-nonproliferation.

Author

DISCLOSURE

All written content of this article on this site is the exclusive copyright and property of Strategy International (SI) Ltd and the author who has written to It.

To note, the opinions stated do not necessarily reflect the official policies of Strategy International.

No prior use in part or in its complete form, written, words, maps, charts or statistical, numerical information can be made, unless there is a written prior request and consent by the author and Strategy International and its legal representative.

All requests should be directed at info@strategyinternational.org

error: Content is protected !