Let the experiments begin, submariners and shipbuilders. Even as shipyards lay keels for a bulked-up “block” of its Virginia-class nuclear-powered attack submarine (SSN), the U.S. Navy espies an entirely new class of attack boat designated SSN(X). The SSN(X) program is slated to debut in 2031, a full twelve years from now. (The program gets underway in earnest in 2034, when the service expects to start buying two boats per year.) That timetable affords the submarine force a bit of leisure to ponder the nature of naval warfare, project the composition of future fleets, and postulate operational concepts and tactics whereby the silent service can navigate an undersea environment in flux.

In short, the experts have time to think about things first rather than vault straight into engineering.

Let’s put that leisure to use. Once submarine officers complete the intellectual legwork they can work alongside naval architects and weapons scientists to devise a sub design that empowers the U.S. Navy fleet as a whole to discharge the missions likely to be entrusted to it. But the design process must not stop with drafting impressive blueprints. The leadership and shipbuilders must treat the fresh design as a hypothesis. A ship design is nothing more than a nifty idea until shipwrights beat the idea into steel and crews take it to sea. Just like in chemistry or physics lab, the team should test their hypothesis in the field, unearth its faults and quirks, refine it, and keep testing until it yields satisfactory results.

Only then should it be pronounced fit for serial production.

That the scientific method should govern naval development sounds self-evident. Evidently, it wasn’t around the turn of the century. Back then a conceit seems to have bewitched the Pentagon leadership. It went something like this: designers could draw up plans for a new platform, engineers could pile on untried weapons, sensors, or other gear, and the platform could go into mass production before proving out in field trials. That’s like a car maker dreaming up a concept car, declaring the prototype ready for sale, and rushing it into production . . . before taking it on the track for a test drive.

Imagine the wreckage to the bottom line of any firm that invented and marketed products the Pentagon way. High-seas operations are the arbiter of what does and doesn’t work in marine technology, just as road testing vindicates a supercar design—or exposes its shortcomings.

The turn-of-the-century procurement philosophy underwent field trials of its own. It gave us littoral combat ships without working sensors and weapon systems but with unworkable crewing and operational schemes; stealth destroyers sporting advanced guns meant to fire ammunition that’s too expensive to buy; and aircraft carriers without operational weapons elevators to arm their warplanes. At best the fin de siècle approach causes delays and costs money as operators discover problems and engineers improvise fixes and retrofit them into vessels already displacing water. At worst, a shipbuilding program could fail altogether—rending a hole in the navy’s fleet design.

Subtract a ship type from the fleet and you risk grave operational or even strategic repercussions. All the more reason to experiment with SSN(X) prototypes long before 2031—and deposit that misbegotten legacy in history’s dustbin.

So much for the jeremiad. The nonpartisan Congressional Budget Office (CBO) has been on a tear at the navy’s expense of late, for instance by issuing an estimate that the SSN(X) will cost $5.5 billion per hull, as opposed to the $3.4 billion forecast in the naval establishment’s long-range shipbuilding plans. The CBO analysts explain that gaping 62 percent discrepancy as an artifact of a disagreement over hull size. Size matters. CBO says it  “assumed” the navy needs a bigger boat to get what it says it wants out of the new class in terms of armament, speed, and other attributes. A vessel boasting the tonnage of the older 9,100-ton Seawolf-class, not the less capacious 7,800-ton Virginia, would be necessary to furnish the internal volume to house these improvements.

Heavier means pricier. QED. Or has it? By all accounts, the Seawolf constitutes the state of the art in SSN technology. But the first boat in the class was commissioned a quarter-century ago, while the design dates to the 1980s. Technology and warfighting concepts have bounded ahead in the interim. Is it really safe to assume that the SSN(X) must necessarily be a Seawolf equivalent—an updated version of a thirty-year-old design?

Maybe, maybe not. Revisiting the fundamentals could help the silent service and the navy leadership answer the CBO’s challenge and vindicate their budgetary outlook. What sort of boat would do so? Rather than gainsay the legendary baseball sage Yogi Berra—who wisecracked that it’s tough to make predictions, especially about the future—let’s stipulate some principles to shape the inquiry shaping the SSN(X). One Principle to rule them all: do not content yourself with improving on existing boats, or even on long-cherished ideas about what a submarine is or how it ought to do business. Situate the new SSN in the strategic and operational context that will greet it when it takes to the depths in the 2030s.

Principle #1—break the mold. 

Over the past couple of years, the Naval War College convened two “Breaking the Mold” workshops aimed at shattering orthodox thinking about all things naval and maritime. I had the privilege of overseeing a cohort of Young Turks who hammered the mold with glee. They rejected measuring naval power by brute numbers of hulls in the water, the traditional gauge of who’s strong and who falls short. Instead, they fell back on a definition of military adequacy beloved of scribes from Clausewitz to Mahan to Corbett: the contender able to concentrate superior combat power at the decisive place and the decisive time is the likeliest victor and, therefore, is adequate unto its purposes.

Nothing in the strategic canon states that it takes a particular type of platform or payload to amass X + 1 units of combat power at a scene of action where the foe fields X units. Doubtless, the “ship” as traditionally understood will remain integral to the panoply of naval warfare. But sea power is undergoing a phase change with the inception of ultra-long-range precision weaponry, unmanned aerial, surface, and subsurface vehicles of many types, land forces able to strike out to sea, and on and on. Submarine warfare is not exempt from this incipient revolution.

Framers of the SSN(X) program should gaze darkly into the future, then, asking themselves whether the “sub” remains the sole bearer of undersea might or they should design the undersea fleet to accommodate new realities. SSNs that operate in concert with unmanned craft could be quite different from the lone wolves that now prowl the oceans and seas.

Principle #2—beware of breaking the mold. 

Now, there’s hidden peril to declaring a longstanding paradigm broken—namely that you might wrongly or prematurely abandon a model that remains valid but needs mending. Philosopher-scientist Thomas Kuhn declares that a “paradigm shift” takes place when so many “anomalies” accumulate—discrepancies the model cannot explain—that the old paradigm’s defenders can no longer explain them away or tweak the model to account for them. The reigning paradigm then collapses, yielding to another that explains reality better.

Plausible enough. But Kuhn implies, without quite saying so, that a twilight zone precedes the paradigm shift. That’s when anomalies have started to appear, but not in such numbers or so compellingly that they refute established orthodoxy beyond doubt. During that ambiguous phase, disputants might toss out a perfectly good paradigm that needs to be refreshed for new circumstances. Witness, for instance, the never-ending debate about whether the day of the aircraft carrier is coming to a close. Anomalies are legion, in the form of saturation missile barrages and the like; whether they’re fatal to carrier aviation remains in question. No naval engagement has put claims and counterclaims to the test—yet. Without a verdict of arms, who’s to say for certain whether these are false anomalies, manageable developments, or bona fide paradigm changers?

Discarding a valid paradigm would court danger just as clinging to an outmoded one would. Worse still, the healthy spirit of breaking the mold could merge with the U.S. Navy’s mostly healthy can-do spirit to produce an unhealthy technophilia—enthusiasm for a new paradigm that may never fulfill its promise. Here’s how that might work. Navy magnates have taken to calling the Trump administration’s proposed 355-ship force “The Navy the Nation Needs,” yet resource constraints appear to rule out more than modest expansion beyond today’s 290 hulls. How should can-do leaders fashion the navy the nation needs without the resources to do it?

Perhaps by indulging in technological fudge factors. Suppose the leadership broke the mold and redefined “the fleet” in terms of combat power rather than raw numbers. Decouple fighting power from numbers of ships and planes and it might be possible to declare that the U.S. Navy boasts 355 ships’ worth of combat power even though it never actually fields 355 ships. Proclaiming that unmanned vehicles and other newfangled gadgetry make up for that sixty-five-ship shortfall would embody the can-do spirit, letting the navy claim to have fulfilled its mandate—potentially without fulfilling it in reality.

The temptation could be fierce to fudge numbers and capabilities by, say, declaring that a large unmanned submersible constitutes an adequate substitute for an SSN, or a certain fraction of an SSN, in terms of combat power. Maybe it is, maybe not. Sweeping claims must be subjected to ruthless scrutiny, keeping service chieftains honest.

An old cartoon that circulates in mathematics departments illustrates the point. It shows a student scribbling equations on the chalkboard to his left and deriving the final result, an intricate formula, to his right. In between, though, he sketches a cloud bearing the words then a miracle occurs. “I think you should be more explicit here in Step Two,” advises his professor. Submarine fleet designers should dispense with orthodoxy by all means, but they must be explicit in Step Two. Otherwise, they might assume away unpleasant realities in their haste to embrace a wondrous, new, but potentially faulty paradigm. Reality bites—eventually.

New ideas about naval architecture must undergo the same rigorous critique as the old. Show your work!

Principle #3—tailor the sub to the setting. 

It’s natural to improve on what works. SSNs like the Virginia and Seawolf are proven designs, and they are Swiss Army knives, so SSN(X) designers will incline to improve on them by increments. Instead, the silent service ought to fit its new boat to U.S. maritime strategy and, in particular, to the operational environment. Happily, reading between the lines of official documents and early reporting on SSN(X), it appears that strategy is indeed impelling the process. The U.S. Navy has reoriented its endeavors from projecting power ashore to winning command of the sea. Which makes sense: after all, a fleet can’t project power ashore from the sea unless it controls the water space from which it will project power ashore. The leadership has seemingly gotten its priorities straight.

Let’s keep it that way. The fight for maritime command must take precedence—now as throughout naval history—and the undersea component must support the overall strategy. The attack in attack boat, that is, should refer to brawling with hostile subs and surface ships, not assailing land targets from an offshore sanctuary.

With regard to the setting, the nature of the operating medium—the sea—could be in flux owing to new anti-submarine warfare (ASW) technology and techniques. A few years ago, Center for Strategic and Budgetary Assessments analyst (and erstwhile submariner) Bryan Clark set tongues to wagging when he contended that new sensors and computer technology might render the oceans more or less transparent to sub hunters—and that the revolution could happen soon and abruptly. Yours truly riffed on Clark’s thesis on a number of occasions, speculating about how warfare in the depths would change if ASW forces could peer through the water with far more fidelity than today.

If subs could no longer rely on the operating medium to conceal their whereabouts, empowering them to strike and flee with impunity, then they would become prey as well as predators. I postulated that superempowered ASW wouldn’t put an end to undersea combat, but that this mode of battle would come to resemble air combat. How so? Sub designers and crews depend overwhelmingly on passive measures to hide from their foes. Silent-running machinery, sound-absorbent coatings on hulls, and other quieting measures damp a boat’s acoustic signature. The less sound a boat emits, the lower an enemy’s chances of detecting, tracking, and targeting it. If passive measures such as silencing no longer suffice, then subs may have to learn to employ active measures as well, much as how air forces now construct stealthy aircraft yet also deploy electromagnetic warfare—jamming, anti-radiation weaponry, and the like—to boost stealth jets’ prospects of evading detection in embattled airspace.

SSN(X) designers should entertain the likelihood that Clark’s thesis is correct in part or in whole, and that the class will have to cope with vastly more menacing surroundings during its service life than did its forebears. They should factor that worst-case analysis into the design, making it easier to adapt hardware and tactics should emerging trends make the unthinkable manifest. Studying aerial and surface warfare could help. Let’s hedge now lest events overtake the silent service later.

And lastly, it’s worth fabricating the SSN(X) for the theaters where national and military strategy say it’s destined to operate, namely the “marginal seas” that wash against the Eurasian “rimlands.” In other words, build the sub to command the waters Washington wants to command. Suppose, for example, U.S. military strategy envisions “tightening the chain”—closing Asia’s first island chain to maritime movement—to deter and apply pressure on China. American and allied submarines will spearhead such a strategy. What better platform than a torpedo- and missile-armed sub to lurk along the island chain, denying passage to Chinese surface and subsurface shipping?

That being the case, designers should ask themselves what capabilities a sub needs to execute such a mission and, as a corollary, what capabilities are superfluous for such a mission. They can consign extraneous capability to the cutting-room floor, helping the service meet the cost estimates it set forth for SSN(X) acquisitions. Maybe an island-chain submarine could be a smaller submarine—smaller than the Virginia class, let alone the Seawolf. Or maybe the new design could come in diesel-electric as well as nuclear-powered variants to hold down the expense. After all, a boat stationed in Japan or the Philippines need not remain underwater virtually forever, as a nuclear sub can, in order to patrol a strait. It can get by with shorter on-station times. Some navies install different propulsion plants in the same hull form. American yards could doubtless follow suit if the customer so directed. The question of whether the submarine force must remain all-nuclear is worth asking—and taking seriously.

Bottom line, we have time to think. Let’s unshackle ourselves from time-honored concepts, test out ideas old and new, discard the worst, and save and improve upon the best. Let’s enthrone the scientific method anew.

James Holmes is J. C. Wylie Chair of Maritime Strategy at the Naval War College and author of the forthcoming Brief Guide to Maritime Strategy. The views voiced here are his alone.