What sits in today’s open literature, dismissed, the way Ufimtsev’s math once sat in a Soviet drawer?
Epistemic label: SPECULATIVE. This piece argues from an analogy and reasons forward from open sources. The history is settled. The forecast is not.
I. The equation they gave away
In 1962, the Soviet state publishing house Soviet Radio released a slim monograph by a physicist named Pyotr Ufimtsev: Method of Edge Waves in the Physical Theory of Diffraction. Its subject was narrow and forbidding. Classical physical optics could already predict how a radar wave scattered off a smooth surface; Ufimtsev went after the harder remainder — the field diffracted specifically from the edges of a conducting body — and reduced it to computable correction terms he called elementary edge waves. The payoff, buried in the mathematics, was enormous: it meant the radar return of a shape could be predicted feature by feature, edge by edge, before anything was built.
The Soviet military classified almost none of it. The judgment of the day was that the work was elegant and useless. To apply those formulas to an actual airframe in 1962 demanded computation that simply did not exist on the Soviet side in any design-useful form, and a secrecy culture that prized prestige publication saw no reason to bury a paper its own evaluators considered an academic curiosity. So it circulated. In 1971 the U.S. Air Force’s Foreign Technology Division translated it in full — Technical Report AD 733203, stamped, with no apparent irony, “Approved for public release; distribution unlimited.” And there it sat, in Western technical libraries, largely unread.
What turned a dead paper into a strategic shock was not new physics. It was a man with a computer and the wit to recognize what he was looking at. In the mid-1970s, as DARPA and the Air Force pushed for a genuinely low-observable aircraft, a Lockheed Skunk Works engineer named Denys Overholser — hired a decade earlier precisely because he was comfortable with computers when most aerospace math was still done on slide rules — worked the Ufimtsev formulas into a program he called ECHO 1. It could compute the radar cross-section of a surface built from flat panels, or facets, and let designers minimize the return of every edge and junction before metal was cut. Ben Rich, who ran the Skunk Works, recalled in his memoir that the early faceted model came back roughly a thousand times less visible than anything they’d built; pressed for a comparison, Overholser put the predicted signature at the size of an eagle’s eyeball. Kelly Johnson, Rich’s legendary predecessor, looked at the ungainly shape and called it the “Hopeless Diamond.” It flew anyway, as the Have Blue demonstrator in 1977, and matured into the F-117A, operational by 1983 — the first combat aircraft whose outer shape was dictated by an equation rather than by aerodynamics.
That is the case. It is worth being precise about its structure, because the structure is the whole point. The paper was never the secret. The Soviets published it on purpose. The secret was the conjunction — Ufimtsev’s edge-wave mathematics joined to a computational capability that made it actionable. One side held the math and lacked the enabler. The other side supplied the enabler, recognized what the math unlocked, and classified the application immediately. Call this the Soviet Equation: a strategic surprise assembled from two openly available halves, where the decisive ingredient is not secrecy but recognition, and the side that loses is the side whose institutions cannot see across the gap between an interesting result and a deployable capability.
The man who saw across that gap died this past April, at 86, much of his life’s work still classified. His passing is a reasonable occasion to ask the uncomfortable forward-looking question. The Ufimtsev story is usually told as a tale of Soviet blindness. But blindness of exactly this kind is not a national trait. It is an institutional failure mode, and it is available to anyone — including us.
So: where, right now, is the West holding the role the Soviets held in 1962? Whose open papers are we reading as academic curiosities because we are measuring them against the wrong benchmark, in the wrong journals, waiting for the wrong enabler?
II. Three places the equation may be running
We assess these as areas where the dynamic may be operating. We could be wrong.
That caveat is not boilerplate. The honest difficulty of this exercise is that you cannot identify a Soviet-Equation blind spot from inside it; if the pattern were obvious it would not be a blind spot. What follows is therefore a forecast built from three observations: that the underlying literature is open and technically mature, that it is being read primarily by communities other than the defense establishment, and that the missing enabler in each case is plausibly closer than Western assessments assume. The three candidates are deliberately uneven in confidence, and I will say where the analogy strains.
Candidate 1 — China: the anti-stealth equation hiding in optics journals
If the original Ufimtsev case was about defeating radar by reshaping a target, the cleanest modern inversion is about seeing what shaping was meant to hide — and it is being published, in volume, by a single Chinese laboratory whose work is filed almost entirely in photonics rather than in radar or electronic-warfare venues.
At the National Key Laboratory of Microwave Photonics at Nanjing University of Aeronautics and Astronautics (NUAA), the group around Shilong Pan and Fangzheng Zhang has spent years moving the generation and processing of radar signals out of the electronic domain and into the optical one. The strategic logic is buried in an unglamorous bottleneck. Conventional radar resolution is capped by the analog-to-digital converters in the receiver, whose precision collapses as bandwidth rises. Photonics sidesteps that wall: it offers enormous instantaneous bandwidth, low frequency-dependent loss, and — a detail that should arrest any electronic-warfare planner — intrinsic immunity to electromagnetic interference. The published results are not modest. The group has demonstrated chip-scale photonic radar in Laser & Photonics Reviews, sub-centimeter and even millimeter-class range resolution in IEEE Transactions on Geoscience and Remote Sensing, three-dimensional photonic MIMO imaging with broadband digital beamforming in IET Radar, Sonar & Navigation, and inverse-synthetic-aperture imaging that survives active interference.
Read those metrics through a defense lens and a capability assembles itself. A sensor with sub-centimeter resolution does not merely detect an aircraft; it begins to resolve the individual scattering features — the edges, seams, and junctions — that low-observable design exists to suppress. A sensor immune to electromagnetic interference is, by construction, hard to jam. The Western reflex is to file all of this under benchtop physics: fragile, laser-dependent, decades from surviving the vibration and thermal load of a fighter pod. That reflex may even be correct about fighter pods — and entirely beside the point. The near-term application is not airborne. It is a ground-based or naval strategic air-defense node, where size, weight, and laser fragility matter far less, and where a jam-resistant, ultra-high-resolution sensor that bypasses the ADC ceiling is precisely the instrument you would build to find shaped, low-RCS targets.
The Ufimtsev parallel here is almost exact, and it is structural rather than rhetorical. The math is open and mature. The missing enabler is ruggedization and systems integration — the contemporary equivalent of 1970s compute. And the work is siloed in optics and photonics citation networks, read by materials scientists and laser engineers, largely outside the radar and EW literature where its implications would register. The paper is not secret. The conjunction is what we are not watching.
Candidate 2 — India: the compute-at-the-edge equation in materials science
The second candidate is the most physically beautiful and the one Western defense AI is least equipped to notice, because it is framed not as a weapon system but as a chemistry problem.
A growing body of work — much of it from Indian institutions, including device-physics groups associated with the IITs — concerns memristors and in-memory logic built on two-dimensional materials such as molybdenum disulfide (MoS₂). The relevant numbers are striking: switching events dissipating on the order of femtojoules, operating currents down in the sub-picoamp range, analog behavior suitable for running neural-network inference directly in the memory array rather than shuttling data to a separate processor. Work reported out of IIT Delhi’s nanoscale facility and validated across journals like npj 2D Materials and Applications shows wafer-scalable MoS₂ memristor arrays performing image classification at high accuracy, in a non-von-Neumann architecture that does the computation where the data already sits.
Here is the part that should change how a strategist reads the literature. In these devices the switching mechanism itself is a defect. The resistance state is modulated by the migration of sulfur vacancies — missing atoms — along the material. The imperfection is not a yield problem to be engineered away; it is the computational primitive. That single fact upends the benchmark the West uses to dismiss the work. Our defense-AI pipeline, primes and startups alike, is built on pristine, high-yield silicon — NVIDIA-class accelerators and the data-center economics that come with them. Against that benchmark, defect-mediated 2D logic looks commercially hopeless: it will never out-compute a GPU, and it makes a terrible server.
But a server is the wrong product. A loitering munition, a smart submunition, an autonomous micro-drone — none of these needs a ten-year processor lifespan or data-center throughput. It needs a few tens of minutes of local, unjammable, low-power inference for sensor fusion or terminal target recognition, in a package that doesn’t need a cloud link and doesn’t cook the airframe. Defect-tolerant, sub-milliwatt edge logic is, against that requirement, not inferior at all. It is closer to ideal. The West is measuring expendable-munition compute against the standard for durable data-center compute and concluding there is nothing here. That is the 1962 evaluation, almost word for word: theoretically interesting, operationally irrelevant. The honest counter-note: the gap between published device physics and a hardened, production-grade edge-inference chip remains real and non-trivial — larger, I think, than in the photonic-radar case. This is a candidate whose enabler has not yet fully arrived. But the direction of travel is unambiguous, and the literature is being read in the wrong building.
Candidate 3 — Iran: the mathematics of “good enough”
The third candidate is where I will be most explicit about the limits of the analogy, because it strains in an instructive way.
Iranian academic strength in a relevant domain is real and verifiable. Iran produces a substantial open literature in robust and fractional-order control of multi-agent systems — the mathematics of getting a distributed swarm to hold cohesion under uncertainty, disturbance, delay, and lost packets. The names recur in legitimate venues: work on adaptive robust control of fractional-order swarm systems under model uncertainty and external disturbance (Naderi Soorki and Tavazoei, IET Control Theory & Applications); event-triggered flocking control explicitly in the presence of cyber-attacks (Amirian and Shamaghdari, in a Proceedings of the Institution of Mechanical Engineers aerospace journal); fractional-order consensus under interval uncertainty from groups at K. N. Toosi and Tarbiat Modares. This is not fringe output. It is a coherent national competence in the control theory of degraded, contested, communication-starved coordination.
Set that beside what Iran has actually fielded. The Shahed-136 — foam, plywood, an inertial-plus-satellite navigation package, a unit cost variously estimated at $20,000–$50,000 — has, per CSIS, CFR, and specialist trade analysis, validated a doctrine of cost-imposition and attrition: saturate defenses with cheap, expendable, one-way munitions and force the defender to burn million-dollar interceptors until the exchange ratio becomes unsustainable. The drones are deliberately built to degrade gracefully rather than to perform exquisitely. The point is reached even when individual airframes are destroyed; the strategic function is interceptor depletion. That the U.S. then reverse-engineered the airframe into its own low-cost LUCAS munition only underscores how badly the capability was under-priced by Western standards of what counts as “military-grade.”
Now the caveat the disclaimer was written for. The clean Ufimtsev template — a specific dismissed open paper that a rival then weaponizes — fits Iran least well of the three. Iran’s drone program is not a forgotten monograph; it is an actively tracked weapons enterprise. And I could not verify, against tier-1 sources, the more specific claims sometimes made in this space — exact packet-loss tolerances, named institutional swarm-telemetry papers achieving cohesion at 80–90% link failure. Those may exist; I did not confirm them, and I will not assert them. What I can defend is the shape of the blind spot, not a single citation: the West has a structural bias toward data-rich, high-bandwidth, redundant, exquisitely encrypted architectures, and a corresponding tendency to read adversary work optimized for catastrophic degradation as merely inferior engineering. The Soviet Equation does not require Iran to be hiding a paper. It requires the West to be benchmarking the wrong variable — perfection of the link rather than viability of the mission under its loss. On that narrower claim, the evidence is suggestive. On the strong claim, we could be wrong, and I want that on the record.
III. The actionability gap, and what would tell us we are right
The thread running through all three candidates is the lesson the F-117 actually teaches, which is not “watch foreign papers.” Foreign papers are watched constantly. The lesson is that the value of open technical literature is never in the mathematics or the physics alone. It is in the conjunction of that work with a specific enabling capability that converts theory into a deployable system — and that institutions reliably misjudge how near that enabler is, because they evaluate the foreign work against their own benchmark of what a finished capability is supposed to look like.
In 1962 the Soviets had the math and lacked the compute, and concluded there was nothing there. The gap closed in roughly a decade, on the other side of the world, the moment compute caught up and one engineer recognized it. Run the same diagnostic on the three candidates and the enabler-gaps look uneven but closing: for photonic radar, the missing piece is ruggedization and integration, and a ground or naval node sidesteps most of what’s hard; for defect-tolerant edge logic, it is manufacturing maturity, and that gap is genuinely wider; for frugal attritable autonomy, the enabler arguably already arrived and was demonstrated over Ukraine and the Gulf before most Western assessments updated.
There is also a shared tell, and it is the most operational thing in this essay. In every case the work lives in the wrong citation network. Photonic radar is read by photonics; 2D-material logic by materials chemistry; degraded-environment swarm control by control theorists. The defense and EW communities that would recognize the implications are not, by and large, reading these journals — exactly as the FTD translation sat unread in 1971 not because it was hidden but because it was shelved in the wrong section. A Soviet-Equation blind spot is, concretely, a place where the people who could recognize a capability and the people who are reading the literature are different people, in different buildings, citing different papers.
What would tell us this forecast is right rather than merely plausible? Three markers, in rising order of significance. First, convergence of authorship — device physicists beginning to co-publish with systems or defense-adjacent groups, the signature of a field crossing from “interesting” to “applied.” Second, a quiet narrowing of the open literature — the most operationally specific results thinning out of the journals, the way a topic goes dark when someone with a clearance starts paying attention. Third, and unmistakably, a fielded demonstrator that the assessments did not see coming — the photonic-radar equivalent of the Hopeless Diamond, ungainly and dismissed right up until it works.
None of this is a claim that China has an operational anti-stealth grid, that India is building memristor munitions, or that Iran has solved contested swarm autonomy in a drawer somewhere. It is a claim about a recurring failure mode and three plausible instances of it, offered under a Speculative banner precisely because the whole value of the Ufimtsev case is hindsight — and the whole difficulty of applying it is that you have to make the call before the demonstrator flies. The Soviets had every page of the answer in 1962 and filed it as a curiosity. The least we can do, sixty-four years on, is read the journals we’ve decided are beneath our attention with the suspicion that the next equation is already lying in one of them, openly published, waiting for the enabler and the engineer.
Sources
History. Pyotr Ufimtsev, Method of Edge Waves in the Physical Theory of Diffraction (Soviet Radio, Moscow, 1962); U.S. Air Force Foreign Technology Division English translation, 1971, Technical Report AD 733203 (FTD-HC-23-259-71), DTIC. Ben Rich & Leo Janos, Skunk Works: A Personal Memoir (1994). John A. Tirpak, “Denys Overholser, Stealth Pioneer Whose Work Led to the F-117, Dies at 86,” Air & Space Forces Magazine, May 1, 2026. Wikipedia/primary-corroborated detail on Lockheed Have Blue, the F-117 program timeline (DARPA contract 1976; Have Blue first flight 1977; F-117 operational 1983), and ECHO 1.
China — photonic radar. Pan / Zhang group, National Key Laboratory of Microwave Photonics, NUAA: “Chip-Based Microwave-Photonic Radar for High-Resolution Imaging,” Laser & Photonics Reviews 14(10), 2020; microwave photonic imaging radar with sub-centimeter / millimeter-level resolution, IEEE Transactions on Geoscience and Remote Sensing, 2022; high-resolution 3D TDM-MIMO photonic radar with broadband digital beamforming, IET Radar, Sonar & Navigation, 2024; SPIE technical commentary on photonics overcoming ADC bandwidth limits and EMI immunity.
India — 2D-material edge logic. MoS₂ memristor / RRAM device work including IIT Delhi nanoscale-facility research (sub-pA currents, femtojoule switching, sulfur-vacancy-mediated resistive switching), Materials Today/ScienceDirect; wafer-scale solution-processed MoS₂ memristor arrays for in-memory computing and high-accuracy image classification, npj 2D Materials and Applications and related reviews.
Iran — robust swarm control and applied doctrine. M. Naderi Soorki & M. S. Tavazoei, “Adaptive robust control of fractional-order swarm systems…,” IET Control Theory & Applications, 2018; N. Amirian & S. Shamaghdari, “Event-triggered flocking control of multi-agent systems in presence of cyber attacks,” Proc. IMechE Part G: J. Aerospace Engineering, 2021; fractional-order robust consensus work from K. N. Toosi and Tarbiat Modares groups. Applied doctrine: “Unpacking Iran’s Drone Campaign in the Gulf,” CSIS, 2026; “The New Era of Drone Warfare Takes Root in Iran,” CFR, 2026; “Shahed Drone: Iran’s Attrition Weapon and the Cost-Exchange Crisis of 2026,” Quwa; reporting on U.S. LUCAS reverse-engineering of the Shahed-136.
Note on sourcing: All specific technical and historical claims above were checked against the cited tier-1 sources. Where the underlying draft research contained mismatched or unverifiable references — and in the Iran case, specific swarm-telemetry performance figures — those claims were either re-grounded in verifiable sources or explicitly flagged as unconfirmed in the text.
