Mesiac: máj 2026

A new ICE brief by our colleague Miro Sedlák raises a provocative but timely question: has @NATO’s hard‑won predictability become too easy to read? Drawing lessons from recent exercises, the piece shows how standardisation and openness – once stabilising strengths – can now generate operational blind spots. The piece is an important contribution to a growing debate on deterrence, adaptation, and how alliances stay credible in a changing warfighting reality.

The Alliance Legibility Trap: Why NATO’s Predictability Has Become a Strategic Vulnerability

Hedgehog 2025 was read across allied capitals as a drone story. The deeper warning is about NATO’s planning culture, and the trap closes first on the exposed flank.

Institute for Central Europe — Policy Brief | May 2026

 

Executive Summary

In May 2025, more than sixteen thousand Estonian and allied troops trained across Estonia under the codename Hedgehog. According to a Wall Street Journal report by Jillian Kay Melchior, drawing on participant accounts, a team of roughly ten Ukrainian drone operators acting as the opposing force simulated the destruction of seventeen armoured vehicles and conducted around thirty additional strikes in half a day.

Two NATO battalions were effectively eliminated in the scenario. One commander summarised the result for Melchior in a single unprintable word (Wall Street Journal).

A separate press story about Joint Viking 2025 in Norway, claiming Finnish reservists had been asked to ease up on demoralised American trainees, was publicly denied by Finland’s Jaeger Brigade. But the story continued to circulate across allied capitals through early 2026 because the underlying anxiety was real.

Both episodes have been read as warnings about drones. The deeper warning is about NATO’s own planning culture. Two decades of alliance investment in interoperability, doctrinal codification and exercise transparency have produced a force whose templates adversaries can study at low cost. Russian intelligence services have the means and the motive to do precisely that, and the open-source tools to make it cheap.

Central European states on the exposed flank face this problem most acutely, because the certification incentives that condition alliance integration and European Defence Industrial Programme (EDIP) funding reward exactly the predictability that has become a vulnerability.

This brief argues that defence ministries in Bratislava, Prague, Warsaw, Bucharest and the Baltic capitals can act now, at the national level, to build operational illegibility into their planning without weakening the interoperability that makes collective defence possible.

 

Two Exercises, One Diagnosis

Hedgehog 2025 ran from 5 to 23 May. More than sixteen thousand troops took part, with units from twelve NATO countries deployed alongside the Estonian Defence Forces. The exercise rehearsed Estonia’s regional defence plan and brought Challenger 2 tanks and Apache attack helicopters into the order of battle.

Among the participants were around ten Ukrainian drone operators from the Nemesis and Rarog unmanned-systems formations, embedded inside a roughly hundred-person composite opposing force led by Aivar Hanniotti, the Estonian Defence League’s coordinator for unmanned systems. The Ukrainian team brought the Delta cloud battle-management system. Hanniotti’s unit deployed more than thirty drones across an area smaller than four square miles (United24).

What happened during the live phase reached the public only in February 2026, through a Wall Street Journal piece by Jillian Kay Melchior and through follow-up interviews given by the Ukrainian operators themselves.

According to that reporting, in a single half-day engagement the Ukrainian team within Hanniotti’s opposing force simulated the destruction of seventeen armoured vehicles and conducted around thirty additional strikes. Two NATO battalions were effectively wiped from the scenario. The targeted formation included a British brigade and an Estonian division attempting a mechanised attack (19FortyFive).

According to one participant, the NATO battle group was “just walking around, not using any kind of disguise, parking tents and armoured vehicles” (Euromaidan Press). The verdict from one unnamed commander, also reported by Melchior, was a single unprintable word.

These details come from press reporting and participant interviews, not from a published Estonian or NATO after-action review. A reasonable objection is also that exercises like Hedgehog are designed to expose weaknesses, and that the opposing force (OPFOR) is meant to win at least some of the engagements.

That is true. The diagnostic value of Hedgehog is that it worked as intended. The harder issue is that the same data set is now available to anyone watching, and for some services watching is professional work.

The reflex reading of all this was that NATO has a drone gap. That would be incomplete. The deeper finding is that NATO units behaved as if drones were not part of the operating picture at all, because their planning assumptions were inherited from a different war.

Convoys moved in daylight without dispersion. Tents and armoured vehicles were parked openly. The templates they followed were ones an adversary equipped with cheap aerial reconnaissance and an artificial-intelligence fusion layer could read like a printed schedule. West Point’s Modern War Institute has been warning for years that Western land forces are preparing for the wrong future war (Modern War Institute). Hedgehog was that warning arriving in operational form.

Joint Viking 2025 in northern Norway entered the same conversation through a different door. The Times of London reported in early 2026 that Finnish reservists, playing the OPFOR, had performed so well against American trainees that exercise organisers had asked them to ease up. Finland’s Jaeger Brigade publicly denied the account, telling The War Zone that its troops had operated on the same side as US forces and were not running OPFOR activities at all (The War Zone).

The denial is on the record. What survived the denial, and what mattered politically, was the broader sense across allied capitals that something was off in how Western conventional units were being prepared for a war that no longer rewards old habits. The story circulated because it was plausible, and the plausibility itself is the diagnostic point.

Hedgehog and the Joint Viking controversy converge on the same place. NATO formations, where they were tested against opponents who refused to fight inside the alliance’s expected grammar, came off badly. The Ukrainians at Hedgehog used a kill chain measured in minutes. Whoever was actually playing OPFOR at Joint Viking used terrain knowledge and operational habits that no NATO doctrinal manual codifies. In both cases the smaller side was the one whose pattern the larger side could not read in advance.

This is not a drone gap and neither it is a force-quality gap. It is a legibility gap, and it is structural.

 

The Predictability Inheritance

How did NATO planning culture become so legible to its adversaries? The answer lies in features the alliance built deliberately, for sound reasons, over decades.

In the Cold War, interoperability was a confidence-building tool. Predictability inside the alliance reassured allies that no national contingent would freelance into a crisis. The same predictability, projected outward toward the Warsaw Pact, deterred miscalculation. Soviet planners knew where NATO formations would hold and how they would mobilise. That was the whole point. It bought stability across the inner-German border for forty years.

After 2014, a different dynamic took hold. Enlargement and the Russian seizure of Crimea drove a standardisation push that produced an explosion of doctrinal codification, certification regimes, and interoperability standards. New accession rounds meant new templates, rehearsed and published. Permanent Structured Cooperation (PESCO) projects produced documentation. NATO Centres of Excellence generated open analytical literature on tactics and procedures.

These artefacts serve real institutional purposes. They are also an open library for any adversary intelligence service willing to read.

A more precise vocabulary helps here, because three different things are usually fused in the popular complaint about NATO predictability and they should be kept apart. The first is institutional interoperability: shared standards, common radios, fuel types, command-and-control architectures. This layer is technical and largely beneficial. It is what allows a Polish brigade to share fuel with a Dutch one in a crisis.

The second is exercise and political transparency: publicly announced drills, observer programmes, press briefings, treaty obligations to declare large-scale activity. This layer is normative and also largely beneficial. It sustains democratic oversight and reassures allies that no one is freelancing.

The third is operational predictability. How units actually move on the ground. Where they camp. How they manoeuvre under stress. This is where the trap closes.

Standardisation does not force a battalion commander to park armoured vehicles in the open. The Hedgehog descriptions of unconcealed convoys in daylight are not directly the fault of any technical standardisation agreement.

But the institutional layers feed the operational layer in subtle ways. A force certified against published templates will reach for those templates under pressure. The exercises that train it are observed, so the procedures rehearsed are also the procedures revealed. And an alliance whose doctrinal output is freely available online gives any adversary intelligence service a head start on modelling how the next exercise will run.

NATO exercises themselves are intelligence collection events for any service that bothers to watch. They are publicly announced months in advance. Their scenarios are outlined in press releases. Their participating units are listed on alliance websites. Foreign attachés have observed them in person. Footage circulates on social media. Russia has been observing and almost certainly systematically analysing these exercises since at least 2014, and its services have had a decade to refine the pattern library.

The arrival of cheap commercial open-source intelligence (OSINT) and AI-driven pattern recognition has compressed the time between observation and exploitation. What once required a foreign intelligence service with satellites and large analyst teams is now accessible to anyone with an internet connection and a modest budget.

The Atlantic Council has described the Russian-Ukrainian war as a compute war, in which the side that fuses sensor data fastest wins (Atlantic Council). The same compute that helps Ukraine target Russian columns also helps adversary services target NATO templates.

Interoperability remains the alliance’s most important practical achievement and the precondition for any credible Article 5 response. What needs management is the operational layer that has grown on top of it without an explicit correction.

The alliance needs the common radios. The common schedules can vary. And even perfect tactical adaptation by NATO units would not fully close the legibility gap, because institutional transparency and standardisation give adversaries a baseline model of alliance behaviour that no amount of dispersion training can erase.

 

Two Wars Already Teaching the Lesson

In late February 2026, Tehran broke its own script. For years, Iranian retaliation had followed a calibrated pattern, with telegraphed escalation ladders and short conflict windows. The June 2025 Twelve-Day War ended on roughly those terms. Mona Yacoubian of the Center for Strategic and International Studies documented what changed when the second war began at the end of February.

By the first week of March, Yacoubian counted fourteen countries hit, Hezbollah reactivated as a second front in Lebanon, Iranian efforts to mine the Strait of Hormuz, and an Iranian ballistic missile that triggered NATO air defences in Turkey (CSIS). Tehran’s choice was deliberate. Yacoubian’s reading is direct: Tehran abandoned calibration because calibration had become Washington and Jerusalem’s targeting aid. Predictability was killing the regime’s deterrent.

A few weeks later, Ukrainian forces ran a counteroffensive on the Oleksandrivka axis at the southern junction of Dnipropetrovsk and Zaporizhzhia oblasts. By early April, Commander-in-Chief Oleksandr Syrskyi reported that the operation had recovered roughly 480 square kilometres and twelve settlements since late January (Ukrinform).

The scale was modest by twentieth-century standards. The timing was decisive. According to reporting on the Institute for the Study of War’s assessment, Ukrainian advances burned through Russian operational reserves and complicated Moscow’s planned spring-summer offensive (ISW). Syrskyi himself said Russia had to redeploy reserves from Pokrovsk and Ocheretyne. The disruption of Russian command and control mattered at least as much as Ukrainian firepower. Russian planners could not read the Ukrainian pattern in time to counter it.

The two cases involve very different actors, a state in one and a hybrid front-line force in the other. The principle that links them is the same. The side whose pattern the adversary cannot read is the side the adversary cannot plan against.

NATO has the rare advantage of absorbing this lesson before pressure arrives. Iran and Ukraine did not.

 

The Central European Stake

Slovakia, Czechia, Poland, Romania and the Baltic states are the geographic front line of any Russian probe against NATO. They are also the states most actively rebuilding their forces around alliance standards, because that is the precondition for integration into collective defence and for access to EDIP funding.

The certification incentive structure tends to reward legibility. A force that wants to plug into NATO command arrangements has to publish doctrine, demonstrate interoperability, undergo evaluation and document its order of battle in formats the alliance can read. Each of those steps is also a step toward becoming readable to adversary intelligence. The dynamic affects every ally to some degree. The exposed flank simply feels it first and most acutely.

The Slovak case is worth examining honestly. The recent national defence reform legislation framing citizen obligations is politically sound and gives clarity about what the state expects in a crisis. That is a real achievement. But the public legibility of the framework is also real. Anyone with a browser can map its structure and its planning timelines.

Polish and Czech force-structure documentation exists as a standing open-source product available to anyone who looks for it. Baltic exercise patterns have been observable, and almost certainly catalogued by Russian services, since at least 2014.

None of this is a criticism of the political logic that produced these arrangements. Interoperability and transparency serve democratic and alliance purposes that are not negotiable. Citizens in democracies have a right to know what their armed forces look like, and so do the allies they will fight alongside. The point is narrower: the operational cost of these features has not been priced into national defence planning, and it should be.

EDIP itself is part of the picture. The European Defence Industrial Programme is designed to scale common procurement and shared capabilities, which is necessary and overdue. It also indirectly nudges national programmes toward equipment standardisation that can erode the small variations adversaries currently cannot model. Both things are true at once, and neither is grounds for opposing EDIP. They are grounds for designing national complements to it (Institute for Central Europe).

On 9 April 2026, Estonia became the first NATO country to translate the Hedgehog experience into a procurement decision. Defence Minister Hanno Pevkur announced the suspension of a €500 million programme to buy new infantry fighting vehicles and the redirection of those funds to air defence, drones, and unmanned systems (ERR).

The decision was framed publicly as flexibility over inertia, citing lessons from Ukraine, market conditions, and the recommendation of Chief of Defence Major General Andrus Merilo. None of this means heavy armour has no place in a modern army; Pevkur was careful to say the existing CV90 fleet would be life-extended, not retired. What it does mean is that Tallinn is willing to act on the experiment it ran in its own forests, and that is itself analytically significant.

The Pevkur decision is a meaningful step in the direction this brief argues for. It is also necessarily incomplete. A reallocation from heavy armour to air defence and drones addresses the platform layer of the problem. It does not yet address the legibility layer. A force that has switched from CV90s to IRIS-T and Ukrainian-style strike drones is better equipped to fight a 2026 war. It is not yet harder for Russian intelligence services to read in advance.

The argument here is not whether to buy tanks or drones or air defence. It is whether any combination of them, integrated under doctrine that adversaries cannot model in advance, is harder to plan against than a force that follows the published template.

Major General Ilmars A. Lejins of NATO Multinational Division North recently put the platform half of this point more directly than this brief has: heavy metal plus drones plus AI equals one system, and the question is not who wins between platforms but who integrates new and old faster.

The observation is correct. The legibility layer is what comes after the integration is achieved. A fully integrated force whose template the adversary has already studied at low cost is still a force the adversary can plan against. A less integrated force whose patterns the adversary cannot read is not.

Finland is the implicit counter-example. A total defence culture with deep historical roots produces a kind of operational illegibility as a byproduct, because force structures shaped by long national tradition do not reduce neatly to a published template (EUobserver). Nobody is suggesting that anyone should import the Finnish model wholesale. The observation is about what kinds of force structure are harder for an adversary to read in advance.

Put plainly: force structures built on standardised templates are easier for an adversary to model than force structures shaped by long national defence traditions. And this asymmetry is the legibility trap in its most concrete form. Central European militaries built around NATO certification look like NATO certification documents. Forces shaped by deeper histories look like nothing in particular.

The point of this brief is not that Central Europe should drift away from NATO. We are part of NATO. We are interdependent inside an alliance whose collective defence guarantee remains the single most important security asset any of our states possess, and no amount of domestic political theatre suggesting otherwise changes that arithmetic.

The argument here is the opposite. Precisely because the alliance is indispensable, its exposed members have an interest in noticing where its strengths have started to work against it, and in taking national-level corrective action that does not require alliance-wide reform to begin.

 

How to Build Illegibility

First, build illegibility into national force design. Establish an explicit budget line, modest in scale and clearly identified in the national defence programme, for capabilities and procedures that introduce variation an adversary cannot model from open-source observation alone. The framing should be operational security and red-team innovation, kept inside the normal accounting structure and visible to oversight.

The activities funded would include irregular task organisation, classified national annexes to alliance standards (designed to remain interoperable at the technical level while varying operational application), equipment combinations exercised only in closed national drills, and composite battalions whose order of battle varies between exercises.

The objective is to remain interoperable for collective defence while denying adversaries a complete planning picture of what national forces will actually do under stress. Annual reporting to the relevant parliamentary defence committee in classified session keeps the line accountable inside democratic oversight.

Second, adopt selective opacity in national exercise design. Maintain full political visibility for NATO collective exercises like Steadfast Defender or Hedgehog itself, where allied confidence and deterrent signalling depend on being seen.

National-level exercises are a different category. They should be conducted with restricted observation. Scenarios should vary deliberately and experiment with non-template manoeuvre. After-action reports should not be published in their entirety. The boundaries here are democratic, not secretive. Parliamentary oversight committees and allied liaison officers retain full access, and Vienna Document obligations are honoured in full.

What changes is the routine publication of operational details from which adversary planners can build a model. Some allies on the exposed flank are already becoming more attentive to the trade-off between visibility and operational disclosure. Others should follow without waiting for an alliance-level signal.

Third, coordinate within Central and Eastern Europe, not only through NATO. A Bucharest Nine framework, with V4 participation where political conditions allow, would multiply the effect of national measures without requiring slow and politically expensive alliance-level reform.

Such coordination can begin small. A working group of defence planning staff from the most exposed members, meeting twice a year, comparing notes on what the adversary appears to be modelling and what each national force is doing to disrupt that modelling. A pilot off-template wargame, designed nationally and assessed regionally, would test the concept at low political cost.

 

Hedgehog 2025 and the Joint Viking controversy have been read across the alliance as wake-up calls about drones, and that reading is not wrong. The harder reading is the structural one, because it touches the institutional design of NATO itself.

NATO has spent twenty years becoming the most interoperable military alliance in history. Two decades of work on standards and on doctrine. That achievement is real, and it is worth defending.

But interoperability without operational illegibility is a trap. It produces an alliance whose strengths are visible to its adversaries before they are deployed against them, and an exercise calendar that doubles as an enemy training resource.

Central and Eastern European states will pay first if that trap closes around the alliance. They are also the states with the strongest analytical interest in arguing, inside the alliance and inside their own ministries, that the time to build operational illegibility into national defence design is now.

Hedgehog 2025 is still fresh. The lessons of a future confrontation with Russia, or with whichever adversary studies the next round of alliance exercises most carefully, are still avoidable. The work is national and the cost is modest. The window will not stay open indefinitely.