The Dispersion Paradox: When Sound Tactics Become a Death Sentence

December 2025. Russian casualties reached approximately 1,200 personnel per day — killed and wounded — according to the Ukrainian General Staff reporting. This wasn’t the result of Russian tactical incompetence. It came despite systematic Russian adaptation to drone warfare throughout 2025: radical dispersion to 4-6 soldier assault groups, spider hole tactics, thermal camouflage, nocturnal movement bounds, and extensive electronic warfare screens.

The Russian General Staff implemented precisely the adaptations that Western tactical manuals would endorse. Every force protection principle suggested that dispersion would reduce vulnerability. Instead, casualties stayed catastrophically high: Ukrainian reporting put Russian losses at approximately 1,200 personnel daily in late December 2025, with drones causing an estimated 70% of casualties. UK Defence Intelligence confirmed rates exceeding 1,000 daily throughout late 2025. The mathematics of force protection appeared to have inverted: dispersed assault teams that should have been harder to detect and engage were dying in record numbers.

This pattern suggests the dispersion paradox: in environments of artificial intelligence (AI)-enabled persistent surveillance, the doctrinal benefit of spreading forces may invert — isolation and extended exposure can create more lethal targeting opportunities than concentration. If true, established Western force-protection assumptions require rapid empirical testing and doctrinal revision. The alternative is discovering the answer through operational casualties rather than controlled experimentation.

What Russian Forces Did Right

Russian tactical evolution through 2024-2025 followed sound doctrinal logic. Early Ukrainian first person view (FPV) drone success punished concentration mercilessly. Russian forces adapted systematically: assault groups shrank from platoon strength to 4-6 soldiers, units pre-positioned concealed fighting positions (“spider holes”) for rapid disappearance, movement shifted almost exclusively to short nocturnal bounds of 200-400 meters, and thermal camouflage nets became standard equipment down to squad level. These deliberate, doctrinal adaptations should have reduced vulnerability.

The timing of Ukraine’s drone evolution coincided with Russia’s completed tactical adaptation. Throughout 2025, Ukrainian forces routinely deployed autonomous systems across the 800-mile front line. Ukrainian forces tested drone swarms in over 100 operations involving 8-25 drones each during 2025, with AI handling navigation and targeting to reduce human operator requirements. By October 2025, Ukrainian forces recorded 25,000 confirmed Russian combat deaths from drone strikes alone, the highest monthly toll of the war at that point. Casualties remained at these elevated levels through December.

The critical capability wasn’t just improved hit rates—though autonomous systems achieve 80-90% accuracy compared to 20-50% for manual operators. The capability that matters is persistent surveillance. The paradox centers specifically on long-endurance ISR drones maintaining overlapping coverage — distinct from short-duration FPV strikes that exhaust batteries quickly or loitering munitions with fixed patrol times. Ukrainian operators manage dozens of simultaneous video feeds during major operations, which is impossible with purely manual control.

This creates an asymmetry that dispersion may magnify rather than mitigate. The pattern is consistent with a troubling hypothesis: that persistent surveillance exploits dispersion differently than conventional weapons systems.

The Mechanism: Dense but Fleeting vs. Sparse but Persistent

Massed formations present “dense but fleeting” signatures. A company-sized assault creates an unmistakable thermal bloom and movement signature but compresses the engagement window —often just 10-20 minutes to detect, decide, and engage. A concentrated force moving quickly can exploit coverage gaps.

Dispersed assault groups present fundamentally different targeting opportunities. A six-man team infiltrating through urban rubble might be detected a dozen times over two hours: brief thermal glimpses, momentary street crossings, signatures at spider hole entrances. No single detection presents a clear shot for human operators struggling to manage multiple feeds and manual targeting.

But AI-enabled persistent surveillance doesn’t need a single clear shot. It aggregates. The autonomous network tracks patterns of life, predicts routes, and pre-positions for the next exposure. Multiple brief detections that overwhelm human operators become a targeting solution for machine learning systems designed precisely for pattern recognition and predictive tracking. The dispersed team moving over hours provides not fewer engagement opportunities but potentially more — extended exposure during prolonged infiltration.

Dispersion also eliminates mutual support at the tactical level. A company under drone attack can suppress threats through volume of fire and maneuver elements to cover each other’s movement. Electronic warfare systems can create protective bubbles. Air defense can engage drones approaching friendly formations. A six-man team isolated 400 meters from the nearest friendly element has limited suppression capability against aerial threats and virtually no maneuver options when engaged. Detection increasingly equals death because there’s no support element to provide covering fire or electronic protection.

The July 2025 capture of Russian positions near Borova demonstrates this vulnerability. Ukrainian forces conducted what military analysts termed “the world’s first successful assault entirely using air and ground drones,” capturing Russian-held positions without deploying infantry. Russian soldiers, isolated in their dispersed positions with no mutual support, surrendered via cardboard signs to drones rather than face systematic elimination. Ukrainian infantry entered and secured the positions without resistance.

This wasn’t a failure of the Russian soldiers’ will. It was a failure of tactical geometry: dispersed defensive positions that lacked the mutual support to counter autonomous systems with persistent coverage.

Why This Matters Now

The casualty pattern from October through December 2025 matters because it occurred after Russian forces had fully implemented dispersion tactics. If dispersion provided proportional protection, casualties should have stabilized or declined as Russian forces completed their tactical adaptation. Instead, Ukrainian forces reported sustained high drone-kill totals precisely when Russian dispersion should have been most effective — a pattern that raises the possibility of a fundamental tactical vulnerability.

This pattern doesn’t prove the dispersion paradox — the data isn’t granular enough. We lack casualty breakdowns by formation size, detailed after-action reports comparing survival rates of different unit configurations, or controlled observations of similar operations with different dispersion patterns. The sustained high casualties from October through December could reflect other factors, such as improved Ukrainian drone production, more aggressive Russian offensive operations, seasonal conditions affecting concealment, or cumulative degradation of Russian force quality. The correlation between completed dispersion adaptation and elevated casualties is suggestive, not definitive.

However, the pattern is troubling enough to demand immediate investigation. If autonomous systems with persistent surveillance genuinely exploit dispersion better than they exploit concentration, NATO operational concepts are built on a lethally wrong assumption.

Current Western doctrine for large-scale combat operations treats dispersion as proportionally protective: spreading your forces wider, reducing your signature, and you proportionally reduce vulnerability to precision fires. Brigade combat teams are designed to operate dispersed over 25-kilometer frontages specifically to reduce vulnerability to sensor-to-shooter networks.

That assumption may have worked against systems with human operators and limited coverage. Picture a brigade combat team conducting dispersed operations in a high-intensity conflict: battalion task forces spread across 20 kilometers, company teams moving independently with degraded communications, platoons conducting infiltration toward objectives using terrain masking and careful movement. Textbook execution of the Multi-Domain Operations doctrine.

Now add adversary AI-enabled persistent surveillance at scale—autonomous drone networks maintaining overlapping coverage, aggregating brief detections into targeting solutions, tracking patterns of life for hours. If the dispersion paradox holds, those dispersed platoons aren’t protected by spacing. They’re isolated targets with extended exposure and no mutual support. The doctrinal assumption that dispersion equals survivability becomes the mechanism that kills them.

The PLA is aggressively fielding autonomous systems without Western institutional constraints on AI warfare. If adversaries deploy comparable capabilities before NATO revises doctrine, the consequences will be measured in casualties, not theoretical debates.

What Western Forces Must Do

Western militaries cannot wait for the next conflict to validate or refute this pattern. Four immediate actions are required:

• Test dispersion thresholds against AI-enabled opposition. Current force-on-force exercises test dispersion against conventional threats—artillery, direct fire, and early-generation UAVs. None replicates AI-enabled persistent surveillance. Training centers should establish red teams equipped with autonomous drone networks and task them with one mission: attrit blue forces executing doctrinal dispersion. Measure actual survivability rather than assume linear protection. Key metrics include survivability by unit size during multi-hour movements, detection-to-engagement conversion rates, optimal movement tempos, and maximum mutual-support radius for effective combined-arms response. Ukrainian data from late 2025 provides strong indicators that these patterns exist. Western forces need empirical validation under controlled conditions.
• Revise movement doctrine for persistent surveillance environments. If extended infiltration creates more vulnerability than rapid movement, Western tactical mobility concepts require adjustment. Current doctrine emphasizes terrain exploitation and careful movement to avoid detection. But if AI systems aggregate brief detections into targeting solutions, rapid movement through detection zones may create less vulnerability than slow tactical movement, extending exposure time. Movement doctrine revision cannot wait for deliberate institutional processes. Training centers should implement immediate guidance: in AI-contested environments, prioritize speed over stealth, maintain tighter mutual support even within dispersed formations, and accept reduced frontage coverage to ensure no element operates beyond the immediate support range of other units.
• Challenge core assumptions in war gaming. War colleges and staff exercises should red-team Multi-Domain Operations specifically: what happens when adversaries field AI-enabled persistent surveillance at scale? Do dispersed formations survive first contact? Can small units operate independently when isolation creates extended exposure? Brigade and battalion commanders need clear guidance now: under what conditions does doctrinal dispersion create acceptable risk versus unacceptable vulnerability? What indicators suggest persistent surveillance coverage that makes dispersion lethal?
• Evaluate countermeasures and their operational costs. Distributed electronic warfare, mobile jamming, decoys, and counter-UAS systems could blunt autonomous surveillance advantages—but each carries costs in weight, power, mobility, and signature. Testing must determine whether countermeasures restore dispersion’s value or create different vulnerabilities.

The Adaptation Race

The Russian General Staff implemented a tactically sound dispersion doctrine. The timing failure came from adaptation speed: by the time Russian forces fully implemented dispersion tactics, Ukrainian autonomous systems had evolved to exploit exactly those tactics. The adaptation cycle is compressed to months rather than years.

Western militaries face the same race. Ukrainian forces took 18 months to overcome institutional resistance to AI autonomy despite an existential threat and immediate battlefield feedback. Western forces operating peacetime training cycles and alliance consensus requirements face steeper adaptation curves. By the time Western doctrine incorporates lessons from Ukraine’s 2025 experience, autonomous systems will have evolved further.

The dispersion paradox may not be real—the sustained high casualties from October through December 2025 could have alternative explanations, and the mechanism might not hold at scale or with proper countermeasures. But the potential consequences of being wrong are severe enough that Western forces cannot assume the paradox is false. The hypothesis deserves rigorous empirical testing before the next major conflict validates or refutes it through operational outcomes. Program offices and operational commands face a choice: treat this as a force protection question requiring immediate investigation, or wait for battlefield experience to provide the answer.

Russian casualties in December 2025 suggest the costs of waiting. Western forces can learn from Ukrainian data and Russian experience now, or discover these lessons through their own casualties later. The adaptation clock is running. Doctrine needs to run faster.

The post The Dispersion Paradox: When Sound Tactics Become a Death Sentence appeared first on Small Wars Journal by Arizona State University.