Small Machines

The Seed and the Toll

Three small machines, and the distance between a commons and a tollbooth.

This month, in a laboratory at Nanyang Technological University, a robot the size of a sesame seed learned to do five things. It can crawl across soft tissue, cut, release a dose of medicine, pinch off a sample to carry back, and heat itself on command to burn away what shouldn’t be there. It is 4.4 millimeters long. It has no battery, no wires, no electronics of its own — only a sliver of soft silicone seeded with magnetic dust, steered from outside the body by shaped magnetic fields. Seven years of work went into teaching one small object to change jobs in under a second.

A continent away, at the University of Bristol, a different lab solved a different problem. Soft robots — the bending, yielding kind that might one day move safely through a living body — have always had a heart problem. The pumps that moved them were rigid and bulky, which rather defeated the purpose. The Bristol team built a pump that weighs about as much as a dried pumpkin seed and runs on less than a tenth of a volt, using nothing but a droplet of liquid metal pushed back and forth by magnetic force. One of its inventors calls it a soft, compact heart.

Both of these are real. Both are wonderful. And both, at this moment, belong to nearly everyone. They were made in public universities, on public money, and described in the open literature, where anyone with a library card to the world’s journals can read how they work. They are, for now, part of the commons of human knowledge — the great shared inheritance of things figured out and written down. A seed-sized machine that can find a tumor and warm it to death is exactly the kind of thing that, held in common, could carry a hospital’s worth of capability into a clinic that has never had one.

That is one road a tiny machine can take.

Here is the other.

There is a third device, already cleared, already on the market. It is a robotic system that lets a physician thread guidewires and catheters through a patient’s blood vessels by remote control, from a workstation a few feet away — sparing the doctor years of accumulated radiation, which is a genuine good. It works. People will be helped by it. And it is sold as a single-use, fully disposable system: you do not own it, you consume it, and you buy it again for the next patient, and the patient after that. Its makers describe the prize plainly — some two and a half million procedures a year in the United States alone, a market to be entered and held.

Notice what has happened to the question of ownership. With the seed-sized robot, the question is still open. With the disposable system, it has already been answered, and answered against the commons. The disposability is not a fact of nature; a guidewire driver does not have to be thrown away after one use any more than a wrench does. It is a decision about title — a way of enclosing not merely the design but the very act of using it, so that every procedure becomes a toll paid at a gate someone else owns. This is the razor and the blade, transposed into the body. You may pass through, again and again, but you will never hold the key.

The hopeful version of this story — the one I was handed, the one the machines themselves will tell you if you ask — reaches for the right images. It imagines robotic repair libraries the way towns already share drills and ladders and rototillers. It invokes the right to repair. It pictures community workshops where durable, modular machines are kept running by the people who depend on them. These are good images. They are our images. But there is a sequencing error buried in the hopeful version, and it matters.

You cannot build a lending library out of single-use disposables. The whole logic of the tool library — the reason it works — is that the tool is ownable and durable: it can be held, lent, returned, repaired, and lent again. A thing engineered to be thrown away after one use is engineered, precisely, to defeat that logic. So the dream of shared access does not float free above the question of ownership; it rests on it. Access is what you get after the title question has been answered the right way, upstream, at the bench and in the patent office and in the choice between a durable instrument and a disposable cartridge.

Which is why the real fork is not “access versus ownership,” as though they were rival goods to be balanced — and as though ownership were, by definition, the private kind, a thing to be escaped rather than claimed. That concession gives away the only word that matters. The question was never whether to own, but whose ownership, and on what terms. Settle the title as a commons — open designs, repairable hardware, public or cooperative manufacture — and wide access follows almost as a matter of course. Settle it as private and disposable, and no amount of talk about democratization will get the seed-sized robot to the village clinic, because the village clinic cannot afford to buy the blade fresh every morning.

The machines do not decide this. That is the thing worth sitting with. The robot the size of a seed has no view on whether it ends its life in a public formulary or a proprietary cartridge. It will crawl and cut and heal in either world with equal, indifferent competence. The decision is ours, and it is made early — in how the research is funded and licensed, in whether the design is shared or fenced, in the quiet engineering choice between a thing you keep and a thing you use up. By the time the marvel reaches the patient, the deed has long since been signed.

A machine the size of a seed could carry a clinic into a place that has never had one. Whether it does is not a question of engineering. It never was.

The Last Commons

The same machine that heals by reaching inside you can harm by reaching inside you — and the only safeguard is whose hand is on the controller.

The seed-sized robot has five gifts, and the fifth is heat. Held in a high-frequency magnetic field, it warms, and the warmth can be aimed: enough to kill a tumor while sparing the tissue around it. The doctors call this magnetic hyperthermia, and it is a real and gentle-sounding thing — a way to burn out a sickness without opening the body to reach it.

But heat does not know what it is heating. The warmth that kills a tumor is only warmth. Pointed at a tumor it is medicine; pointed elsewhere it is a burn. This is not a flaw hidden in the design, to be engineered away in the next version. It is the design. And it is not only the heat. The blade that takes a biopsy is a blade. The gripper that retrieves a sample is a grip. The little channel that releases a measured dose of medicine will release, with equal obedience, whatever it is given to carry. A machine built to travel to a precise place inside a living body and act there is, by its nature, indifferent to whether the act is mercy or harm. At this scale, healing and harming are the same gesture, pointed two ways.

Which means the machine is safe only insofar as the hand on the controller can be trusted. And here is the genuinely new thing — the thing that has no precedent in the long history of people putting their hands inside other people to help them. The hand on this controller need not be in the room. It need not be the surgeon who will close the wound and go home. It need not, in the end, be yours. For the first time, the agent acting inside your body and the control of that agent can be held by different parties — and the party that holds the control is whoever holds the controller. A hospital. A manufacturer. A payer. A state. Anyone who can acquire the controller by purchase, compel it by subpoena, or take it by breach.

Surgery has always required a hand inside you. But that hand was present and accountable; it was in the room, it answered for itself, and when it was done it withdrew. The micro-robot’s hand can be remote, and it can be persistent, and it can stay. What it introduces is not a sharper scalpel but a detachable sovereignty — the separation of acting inside a body from being the body that is acted in.

This is the enclosure that the first panel could not quite see, because it was looking at the wrong field. The enclosure movement took the common fields first, then the seed, then the knowledge of how things grow, and — as the panel above argued — at last the very act of use, sold back one disposable cartridge at a time. But through all of it, one commons stayed unfenced, because nothing could reach it: the inside of the body. The interior was the one place that remained yours by sheer inaccessibility. No one could act there without your swallowing something, or without a surgeon entering briefly and leaving. The steerable, persistent, externally-governed micro-agent ends that condition. It makes the inside of the body a place where another party can act — on their schedule, by their command, under their terms.

So the question that decides everything is small and concrete and almost embarrassing in its plainness: who holds the off-switch?

A thing inside you that obeys a signal — can you countermand it? If you fall ill, or fall out with the institution that placed it, or simply wish it gone, is the authority to stop it in your own hand, or in a hand that answers to someone else’s interest? Under one arrangement — public, transparent, the patient sovereign, consent that is real and revocable, the off-switch held by the person whose body it is — the gift is real and the danger is governed. Under the other — proprietary firmware, opaque control, a terms-of-service for your own interior — your body becomes a managed device, and managed is only ever a polite word for managed by someone else. The dual-use danger is not a second, separate problem standing beside the ownership question. It is the ownership question, arrived at from the inside, where it has always been heading.

And what can be governed can be turned. It does not take a dark imagination, only a memory, to know that tools built for care have been turned to control before, and that the more intimate the tool, the more total the control it permits. I will not draw the picture in detail; the picture is not the point, and the people who would draw it do not need my help. The point is sober and almost the opposite of frightening: the very reasons one would want this technology — that it reaches the unreachable, acts with precision, obeys at a distance — are precisely the reasons it must be held in accountable, common, answerable hands. The safeguard against misuse is not a better lock on the device. It is the title. Who owns the machine decides whether the off-switch sits in your palm or in someone else’s ledger.

The oldest writ in the common-law tradition is habeas corpus — you shall have the body — the ancient refusal to let anyone hold a person’s body without answering for it before the world. It was written for prisoners, for the body held in a cell. We are going to need it again, rewritten for a body held from within, by an object too small to see and a controller too far away to name. The instinct is the same and the stakes are older than any technology: that the body is not a field to be enclosed, not a device to be managed, not a market of two and a half million procedures — but the one commons that was never anyone’s to fence.

I would want this machine inside me, if I were sick enough to need it. I would want it badly. But only if the hand on the controller answered to me. That condition is not a technical detail to be settled later. It is the whole of the matter, and it is being decided now, quietly, while the marvel still glints on the fingertip and no one is yet asking who will be holding the other end.