Golden Dome’s Hedge: Lessons from Failed Military Programs

The Pentagon describes Golden Dome for America as a “system of systems,” an integrated air and missile defense architecture that combines space-based sensors, orbital interceptors, ground systems, and a command network designed to defeat any aerial attack. Eighteen months after its founding executive order, the program broke ground on its first operational site. Its architecture remains so closely held that the Congressional Budget Office cannot fully price it, and cost estimates span from $252 billion to $3.6 trillion. The program also carries significant risk of failure, depending on how many tightly coupled, interdependent systems it attempts to integrate.

One of the last times the Department of Defense organized a warfighting concept around a tightly coupled architecture was the Army’s Future Combat Systems program. The Army spent roughly $18 billion over six years without fielding a single vehicle, making it the most expensive canceled program in Army history. It failed because it made an unhedgeable bet on unproven, interdependent systems with no standalone value that couldn’t be abandoned once disproven.

Golden Dome is not an unhedged program. It is a layered portfolio with gated competitions, severable contracts, and layers that hold independent value, structured on paper to avoid the failure mode that doomed the Future Combat Systems program. Its hedges are still under pressure: lump sum funding outside annual congressional review, a final integration gate after the money’s spent, and a delivery date set by an administration term rather than technical readiness. The solution is to assign an independent authority, outside the program, empowered to fail its final gate.

The Future Combat Systems Concept

After the Army struggled to deploy Task Force Hawk to Albania in 1999, it concluded that the force was too heavy to arrive in time but too light to win once it was there. The proposed solution consisted of eighteen new interdependent systems, including light vehicles, ground robots, drones, sensors, and missiles, designed to function as a complete set and connected via a hypothetical battlefield network.

The wager was simple. Because a vehicle light enough to be carried aboard a C-130 transport aircraft could not survive a tank main-gun round, it could not be hit by one. Therefore, seeing and shooting first supplanted armor. The C-130 limit was a requirement rather than a tradeoff and changed everything that flowed from it. This logic echoes Wayne Hughes’s maxim of “attacking effectively first,” but only half of it. Hughes paired striking first with staying power, the capacity to absorb hits and keep fighting, but warned that forces unable to absorb hits are one exchange away from defeat.

Acquisition programs are concept-driven by design. What made the Future Combat Systems extraordinary was what the concept was asked to do — replace armored protection with a nascent property of a networked force that could not be tested until the force existed. The concept was self-reinforcing, requiring immature technologies to be viable and worth building. If networked protection underperformed, the force would not become proportionally less capable. Instead, vehicles would become disproportionately more vulnerable.

The Army traded a verifiable means of protection for an unverifiable one, then constrained vehicle weights to accommodate an aircraft’s load limits. When experiences in Iraq forced the issue, the Army mandated V-shaped hulls on the manned vehicles. But fitting inside a C-130 penalized every pound of armor or added vertical height, reducing capabilities elsewhere while still not addressing the original survivability shortfalls. The Army was unable to convince civilian leadership that the modifications were enough. The result was coupled conceptual and technical risk. Future Combat Systems represented brittle holism: No single system held value outside the whole, and every interdependent system was a critical vulnerability.

The Promise and Peril of Interdependence

In a loosely connected portfolio, a failed sensor is a failed sensor. In a tightly coupled architecture, it’s a failed operating concept, as tight coupling makes cascading, catastrophic failures nearly inevitable in complex interdependent systems. Government Accountability Office auditors recognized this and identified integration itself as the Future Combat System program’s dominant risk factor.

The Future Combat Systems’ network was the clearest case of possible failure. The program depended on the Joint Tactical Radio System, a separate program outside the Army, with costs outside the Systems’ budget. When they struggled to field the radio, the Army could neither replace the network nor field the force without it.

Golden Dome’s command-and-control layer presents the same risk. While its sensors and ground-based interceptors are mature, success depends on a new, yet-to-be-fielded network to marry them. This network is likely the Joint Fires Network, an ambitious effort that fuses disparate data across multiple transport layers with complex software and artificial intelligence. The program is already under stress, with fielding and training delays drawing Inspector General scrutiny. If the Joint Fires Network layer fails, shooters will lack timely sensor data across layers that would otherwise function independently. The questions Future Combat Systems poses are whether that layer is appreciated as the lynchpin it appears to be, and whether any alternative exists if it slips. No public communication suggests one does.

Why the Bet Could Not Be Hedged

Three structural properties of Future Combat Systems made the bet impossible to hedge. First, testing it was uneconomical. Validating the concept required the entire architecture to exist at once: the network, sensors, ground vehicles, and drones. Evidence of failure arrived only after the Army committed significant resources. A portfolio structure would have allowed the Army to learn from a failed sensor program without risking the entire armored force.

Second, the program could not be de-scoped. Because of tightly coupled interdependence, the program could not trade scope for affordability. Software proved the most complex and inflexible part of the program, as dropping a system did not remove code so much as force a rewrite for every component that depended on it.

Finally, the Army believed transformation could not wait. The expedited schedule treated unfinished inventions and systems as on-demand deliverables. Auditors continuously reported that most critical technologies were immature at launch and remained so years later. Moving fast surfaced problems after commitment, the most expensive time for correction.

This lesson in particular lands awkwardly today. The Pentagon has committed the department to higher risk thresholds, making accelerated production and rapid procurement the default. Much of that urgency is warranted, but there are tradeoffs between speed and validation.

The important distinction is between decisions that are cheap to reverse and decisions that are not. Fielding a drone quickly and finding it lacking costs a production run. Committing a force architecture to an integration concept quickly and finding it wanting could cost a decade. While RAND’s retrospective attributed the Future Combat Systems’ failure to execution rather than concept, the program’s structure made good execution unverifiable until it was too late.

Why Nobody Could Fold

The warnings about the Future Combat Systems were neither secret nor occasional. Government auditors, congressional researchers, and budget analysts published similar findings for half a decade, yet the program survived each review. Auditors judged the business case inadequate as early as 2006, and the same concerns persisted through 2008.

Several mechanisms kept the Future Combat Systems program alive. The program was the Army’s budget centerpiece and tied to its future identity, making the two difficult to separate. Further, sunk-cost fallacy and the Army’s advertisement that subcontracting work spanned more than 100 congressional districts also made canceling difficult. Ending the program required an outsider with the political will and no emotional stake in the outcome. Office of the Secretary of Defense oversight came late, and the independent review teams convened along the way lacked the independence and authority to force change. Secretary of Defense Robert Gates eventually canceled it in April 2009, amid two wars that fundamentally discredited the program.

This was the institutional consequence of an unhedgeable bet. When a concept becomes the service’s theory of itself, failure means conceding the theory, so the available responses narrow to continuing or canceling outright. A hedged program can be trimmed by its managers, but an unhedged one can only be stopped by someone who does not answer to them. The lesson for Golden Dome is that the authority to fail its final gate should rest with someone who owns neither the program, its schedule, nor the money.

The Golden Dome Test

Golden Dome offers a test of whether the department has learned the lesson of interdependence. It derives value from integration, but unlike the Future Combat Systems program, its individual systems have standalone value. THAAD and Patriot provide theater defense regardless of Golden Dome, and the missile-warning and tracking satellites are capabilities the Space Force is already investing in. Many of the individual technologies have been demonstrated independently. Yet industry concedes that assembling them into a working operational system is the hard part, and that assembly runs through a joint command-and-control network the program’s director has made a top priority.

Despite the interdependency risks, its current structure reflects awareness and discipline. The interceptor effort is a gated prize competition among twelve companies, moving from component testing through prototype and on-orbit demonstration, and the program states it will not make production requests until capabilities are proven.

However, two properties resist hedging, and they fail in different ways. The command-and-control network is the coupling risk, and it is Golden Dome’s version of the Future Combat Systems’ network: the program depends on but does not own the Joint Fires Network, and no viable alternative exists.

The space-based interceptor layer is a different problem. Continuous boost-phase coverage requires a minimum constellation of about 1,000 specialized satellites, so a partial constellation is not partial coverage but a gap. That number, like the C-130 weight limit, is fixed outside the program’s discretion and changes everything downstream. This property made the Future Combat Systems program impossible to de-scope, because reducing it did not produce a smaller version of the capability. Unlike the terminal and midcourse layers, which retain value if the boost-phase layer never materializes, the boost-phase mission has no reduced form.

This is why the program director’s testimony that space-based interceptors would be dropped if cost and scalability targets are not met matters more than it may appear. It is a fold criterion set in advance, an acknowledgment that a layer with no reduced form should be abandoned entirely rather than fielded partially. Whether that acknowledgment survives contact with a constellation under contract is the question.

That leaves two critical vulnerabilities: a network whose failure cascades across every dependent layer and a constellation that cannot be partially delivered. Both are tested at the final integration gate, when the money is already spent, and the hedges are thinnest.

A missile-defense advocate will likely object that the analogy proves too much. Future Combat Systems rested on unproven battlefield technologies, while missile-defense layers are demonstrated. That objection is valid and should be conceded. Fielding more Patriot and THAAD capacity is cheap to reverse, and slowing that down is not defensible. But the constellation size and the integration architecture are not, and those two decisions should not be made before they are validated.

Keep the Bet Foldable

The appropriate question for Golden Dome is whether the structured hedges hold under pressure. The deadline is set by an administration’s term rather than technical readiness, a forcing function similar to the C-130 weight limit. Nearly all of the program’s funding arrived as a lump-sum reconciliation outside the annual budget review where cost growth would ordinarily draw scrutiny. And the final integration gate comes after the money is spent. Will performance gates remain honest when political pressures mount against the demonstration schedule, and will the final integration gate be allowed to fail with production consequences?

Two refinements would make those hedges real. First, gate criteria should include cost and performance thresholds committed before the gate, not negotiated at it. A program can pass technical demonstration while unit costs drift to where the affordable fleet falls below the coverage threshold, passing its gates but still failing to deliver the capability. The Future Combat Systems program’s overarching cost estimates grew sharply without triggering the Nunn-McCurdy statutory tripwire, but the radio its network depended on did. Because the Pentagon measures cost overruns by individual program baselines rather than structural dependencies, it obscures the true systemic failure of an interdependent system of systems. Even when the wire tripped, it rarely stopped anything: All but four of the thirty-seven breaches between 2007 and 2016 were certified rather than acted on by curtailing the program.

Second, assign authority over the final integration gate now, and place it outside the program. Early gates test technical risk component by component, but interdependence means the concept-level claim is tested only at the last one, where sunk cost, politics, and momentum of earlier successes press hardest. The Future Combat Systems record shows that late oversight and weak, dependent review teams do not work. The remedy is a standing non-advocate review, chartered before the final gate, with cost and performance thresholds fixed in advance, and empowered to fail the gate outright. The Director of Operational Test and Evaluation and the Director of Cost Assessment and Program Evaluation already produce the independent judgments such a review needs, but neither holds the authority to act on them.

In 2007, the Army paid for a theory of survival while fighting a war that demanded more protection than the concept could carry, with no way to change course short of cancellation. Golden Dome has gates, severable contracts, and an outside architecture authority — more structure than the Future Combat Systems program ever had. What it lacks is a named independent authority allowed to fold the hand. Until it does, its hedges are only as strong as the program’s willingness to admit failure, and Future Combat Systems proved that hedge does not exist.

Image: U.S. Army via Wikimedia Commons.

Source link