Mass, reimagined

Mass — having more of something than your opponent — is one of the oldest ideas in warfare, and one of the most expensive. For the last several decades, Western defence procurement moved decisively the other way: fewer, more capable, more expensive platforms, each so valuable that losing one is a strategic event.

Small autonomous systems reopen the question, because they change what a single unit costs and therefore what it is reasonable to risk.

The exchange rate problem

The clearest illustration is defensive. If a threat costs very little to produce and the established means of countering it costs a great deal, the defender loses the exchange even when every single engagement succeeds. Win often enough at that rate and you are still bankrupt.

The same arithmetic applies offensively, and to reconnaissance. If answering a question about a location requires risking an expensive platform, there are questions you simply will not ask. Lower the cost of asking and the set of feasible questions expands dramatically.

Cheapness is not a compromise on capability. At sufficient scale, it is a different capability.

Why 'more drones' is not a swarm

It is tempting to conclude that the answer is simply quantity. It is not. A hundred platforms each requiring an operator is not mass; it is a staffing crisis. The value only materialises if the group can be directed as a single entity — which is a software problem, and a hard one.

A genuine swarm has properties that a collection does not:

• Tasking scales sub-linearly. One operator directs the group; workload does not grow with the number of members.
• The picture is shared. What one member observes informs what the others do, producing an understanding none of them had alone.
• Attrition degrades rather than stops. Losing members reduces capability proportionately instead of ending the task.
• Coordination survives poor communication. Members cannot assume a reliable link to each other, let alone to a central controller.

The two-tier idea

One architecture we find genuinely interesting is a two-tier arrangement: a larger carrier platform that provides range, endurance, compute and communications, releasing a group of very small platforms to do the close-in work.

The appeal is the division of labour. The small platforms carry almost nothing but sensing and autonomy, which is precisely why they can go low, close and into spaces a larger aircraft cannot. The carrier stays back, coordinates them, and relays what they find to the people in command. Neither tier could do the job alone.

We describe this at concept level deliberately. It is a research direction, not a product, and the engineering questions underneath it — coordination on almost no bandwidth, collision avoidance in tight volumes, cost per unit — are exactly the ones that decide whether the idea survives contact with reality.

The accountability question

Numbers raise a question that a single platform does not: what does meaningful human command look like when there are many systems rather than one? We do not think the answer is a person approving every action — that reintroduces the workload problem the architecture exists to solve.

Our position is that command operates at the level of the task and its limits. A person sets what the group is to achieve and the boundaries it must not cross, retains the ability to intervene or recall at any point, and remains responsible for the outcome. That principle does not become negotiable because the system got larger.

Where this sits

Mass through small autonomous systems is one of the more consequential shifts in the field, and it is early. Our work here is research: we describe the shape of the idea publicly and keep the engineering for the people who need it.