JADC2 Is a Data Contract, Not a Joystick

Every Joint Terminal Attack Controller (JTAC) builds a primary, alternate, contingency, and emergency (PACE) plan before stepping off. When the primary radio dies, you already know what you're reaching for next, and nobody has to call a meeting about it.

The Department of War (DoW) has no PACE plan for its own network. It has a vision statement.

That vision is Joint All-Domain Command and Control (JADC2). And we keep trying to build it as a joystick.

BLUF: JADC2 isn't a single pane of glass for senior leaders. It's a federated data contract: shared schemas, identity, lineage, and service-level agreements that let units fight as one without waiting on headquarters. Write the contract, protect the packets instead of the pipes, and run the whole network on a PACE plan that assumes the expensive parts get shot first.

The joystick fantasy

Most JADC2 briefings I've sat through eventually show the same picture: a senior leader in an operations center, every sensor and shooter on one screen, able to reach down into any fight on the planet.

That picture is the problem. Centralized command and decentralized execution are why American forces have out-fought larger militaries for generations. The person closest to the fight decides, inside the commander's intent. A network built so a flag officer (FOGO) on another continent can steer a squad isn't modernization. It's a regression with better graphics.

It's also a target. A force whose decisions all route through one headquarters has a center of gravity (COG) the enemy can find and hit. Micromanagement is the vulnerability our adversaries are counting on us to keep.

The sharpest proof came the night before D-Day. Cloud and flak scattered two American airborne divisions so badly that only about four percent of the 82nd's two regiments west of the Merderet River landed where they were meant to, and the 101st came down spread across an area twenty-five miles long and fifteen wide.^[1] It should have been a catastrophe. Instead it became a weapon. Men from both divisions intermixed in the dark, formed little bands of paratroopers (LBPs) around whatever objective was closest, and fought scores of small battles no German headquarters could map.^[2] To the defenders, every field or village might hold one American or a hundred, so they couldn't tell a raid from the main landing and lost hours treating the whole assault as a feint.^[1] The Seventh Army's own war diary logged an estimated 30,000 to 40,000 paratroopers against the roughly 13,000 actually dropped — a two- to three-fold overcount that froze its response while it hunted for a main effort that was everywhere and nowhere.^[3] The commander of the German 91st Division, Wilhelm Falley, was away at a war game in Rennes; speeding back to take control of his sector, he was shot dead in his own staff car by the very paratroopers his centralized headquarters couldn't find.^[4] That is how free people fight: from inside the chaos, faster than a hierarchy can re-form. A JADC2 that pushes decisions upward and waits on headquarters doesn't just forfeit that edge — it hurts the force it was built to help.

The man on the ground will always be the best ground commander. Our job is to feed him better data.

A contract, not a dashboard

What JADC2 actually needs is boring, and that's the point:

Schemas. Versioned definitions of what a track, a target, or a fuel state looks like, so every system reads it the same way.
Identity. Access tied to who you are, what mission you're on, and the health of your device, not to which network you're plugged into. That's zero trust architecture (ZTA) as doctrine, not as a product.
Lineage. Every piece of data carries where it came from and who touched it, so a shooter can trust a track they didn't generate.
Service-level agreements (SLAs). How fresh, how complete, how fast. Written down, measured, and owned.

The beating heart of JADC2 is a schema registry, not a slide deck. If a unit can't publish and subscribe to a mission thread at machine speed, it isn't in the fight. And if two systems can't prove they speak the same schema, no amount of integration money fixes it.

And "measured" has to mean numbers, not adjectives. A real data SLA grades the things a shooter actually leans on: freshness — the 95th-percentile age of a track, its "P95 age" — completeness against the target set, label fidelity (how often two analysts, or a model and an analyst, agree on what a thing is), lineage depth (how many hops back you can trace a datum), and time-to-decision (T2D): sensor event to human call, with the audit trail attached. Each one is a dial you read in real time, not a promise on a slide. And because they're numbers, you can pay on them — later in this series I'll show how outcome contract line item numbers (CLINs) release money when the metric moves, instead of when a milestone slips.

From where I've sat, the honest state of joint data is that each service's waveforms, formats, and even definitions of the same unit rarely line up, so every new system gets a bespoke connection to every other one. A contract replaces hundreds of custom handshakes with one standard everyone builds to. Later in this series I'll call those contracts application programming interface (API) fragmentary orders (FRAGOs): interface standards that are versioned and owned like any other order.

Protect the packet, not the copper

We still act as if security comes from the pipe: a classified circuit, a government network, a military satellite. Modern cryptography makes the pipe irrelevant. With the Commercial National Security Algorithm Suite 2.0 (CNSA 2.0) and zero trust, mission data can cross commercial and contested links, because each packet is validated, not the perimeter. Its post-quantum algorithms also blunt "harvest now, decrypt later" (HNDL) collection, where an adversary stores today's traffic to break once quantum computing matures.

Once the packet protects itself, you can use anyone's pipe. That's cheaper, and it's much harder to kill.

A PACE plan for the whole network

Layer	What it is
Primary	The best-performing, lowest-cost commercial internet service provider (ISP) available
Alternate	Commercial satellite internet
Contingency	Tactical radio, such as high frequency (HF)
Emergency	Couriers carrying signed media

This isn't just for JADC2. It works for the Department of Defense Information Network (DoDIN) as a whole, down to email on base.

The alternate slot is where people flinch, so here's the math. The military's protected and wideband satellite communications (SATCOM) fleets, from AEHF to MUOS to Wideband Global SATCOM, add up to a few dozen exquisite, enormously expensive satellites. To a space-capable adversary, that's a short target list. Starlink alone has more than 11,000 satellites in orbit.^[5] You can't shoot down a constellation that adds satellites faster than you can find them. The Space Force's own shift toward proliferated constellations is the Department quietly conceding the point.

The idea that military SATCOM is inherently more secure doesn't survive contact. Survivability comes from numbers. Security comes from the encryption on the packet.

The objections, answered

"You're betting the network on one company." No. A data contract makes the provider swappable, which is the whole reason to write one. Commercial satellite internet is the alternate, not the primary, and it should be bought from more than one constellation, with terms that guarantee service in a crisis. Vendor dependence is a real risk, and a contract plus a PACE plan is how you manage it. A single government constellation doesn't remove that risk. It just puts all of it in one exquisite basket that's very easy to target.

"Commercial networks aren't secure enough." The pipe was never what made traffic secure. The encryption is. If CNSA 2.0 and zero trust protect a packet on a government circuit, they protect it on a commercial one.

"Leaders need visibility." Visibility, yes. Control, no. A good data contract gives commanders at every echelon a better picture than they've ever had. It just doesn't pull the decision away from the person who can see the target.

The fix

Stop funding JADC2 as a system and start funding it as a contract. Publish versioned schemas and SLAs for the first few joint mission threads, enforce access by identity, and require every new program to prove it can publish and subscribe to them before it gets money. Rebuild the network around a PACE plan that protects packets and assumes the exquisite parts die first. None of it requires a new authority. It requires deciding that JADC2 is plumbing, not a joystick.

JADC2 isn't a way for senior leaders to relive their company-grade years. It's a promise that every unit's data will make sense to every other unit. In a software-defined force, data and timely updates are fuel and ammunition; JADC2 is the logistics system that moves them to the fight.

What you can do Monday

Unit commanders: Write a PACE plan for your unit's data, not just your radios, and test what happens when the primary drops.
PEOs / PAEs: Before your next milestone, show which joint schemas your system publishes to and subscribes from. If the answer is none, that's the finding.
Contracting officers: Put interface conformance and data rights in the contract: the government owns the data, and the system must export it in the published schema. Buy the commercial license, not the vendor's source — owning the data and the schema is what keeps the provider swappable and the market competitive.
DIU and the service innovation organizations: Make plugging into a published schema, not a bespoke integration, a condition of award for commercial vendors.
OSD (CDAO and the DoW CIO): Publish a versioned schema registry and graded data SLAs — P95 age, completeness, label fidelity, T2D — for the first three joint mission threads, in a repository, not a PDF.
Congress: No new authority needed. Ask for JADC2 progress reported as schemas published, mission threads meeting their SLAs, and the share of systems that conform, not dollars spent.