Chapter 9
Chapter Nine — The Agency
The office is in a building that does not appear on any public map of the area. The building is in Northern Virginia, in the band of office parks and government annexes that stretches from McLean to Chantilly along the Dulles Toll Road, and it looks like every other building in that band — beige facade, tinted windows, a small parking structure, a guard booth that does not look like a guard booth because it is designed to look like a reception desk for a company that does not exist.
The woman who works in the office is named Helen Park. She is forty-one years old. She is a program manager in a division of a three-letter agency that the reader will not be asked to name because the agency does not name itself in the places where the work Helen does gets done. Helen’s division is called, internally, the Technology Forecast Office. The Technology Forecast Office does not forecast the weather. The Technology Forecast Office reads patent applications.
This is what Helen Park does for a living. She reads patent applications. She reads thousands of them. She reads them the way a prospector reads a riverbed — not looking for every grain of sand but looking for the specific glint that means something is in the water that the water does not know is there.
The Technology Forecast Office was created in the early 2000s, after a set of reviews within the intelligence community that concluded, with the kind of bureaucratic certainty that attends conclusions everyone already knew, that the agencies had been surprised by the internet and should try harder not to be surprised by the next thing. The solution was a small office whose mandate was to monitor the intersection of emerging technology and national security. The office’s operating budget is classified. Its staff is seventeen people. Helen is one of the seventeen.
She is good at the work. She has a degree in materials science from Carnegie Mellon and a master’s in technology policy from Georgetown and a career that has been spent in the gap between what scientists can build and what agencies can use. She is not a scientist anymore. She is not an operator. She is a translator — a word she uses about herself with some irony, because her job is to read the language of invention and translate it into the language of procurement. She reads the patent and she tells the people upstairs what the patent can do for them.
Most of the time, the patent can do nothing interesting. Most patents are incremental improvements on existing technology — a slightly more efficient battery cathode, a slightly more durable polymer coating, a slightly more sensitive biosensor. The improvements are real but they are not the kind of thing that changes the geometry of the world. Helen files them in the database and moves on.
Some of the time, the patent can do something interesting enough to flag. A new class of metamaterial with potential antenna applications. A gene-editing technique that could be used for both therapeutic and bioweapon purposes. A quantum computing architecture that could, if scaled, render current encryption obsolete. These she writes up in a memo and sends upstairs. The memo goes into a file. Sometimes the file leads to a program. Sometimes the program leads to a contract. Sometimes the contract leads to something that changes the geometry of the world. Most of the time it does not.
Once in a great while, the patent can do something that makes Helen Park stop reading and lean back in her chair and look at the ceiling.
The Verma patent crossed her desk on a Thursday morning in the fall, four months after it was published. The lag is not unusual. The Technology Forecast Office monitors patent publications in batches, sorted by classification code and filtered by an algorithm that Helen helped design. The algorithm looks for keywords and claim structures that correlate with technologies the office has previously flagged as significant. The algorithm is not perfect. The algorithm is a sieve with holes that are sometimes too large and sometimes too small.
The algorithm flagged the Verma patent because the patent’s claims included the phrase “self-organizing molecular assembly capable of coordinated surface-level response to external signaling.” The phrase contains three terms the algorithm has been trained to flag: self-organizing, coordinated response, and external signaling. These are terms that, in combination, describe a system that is more than the sum of its parts. A system that organizes itself, coordinates its behavior, and responds to commands is a system that is, in the loosest sense of the word, programmable. Programmable self-organizing systems are the category of technology that the Technology Forecast Office exists to watch.
Helen read the patent. Then she read it again.
The patent describes a molecular machine that changes the color of fabric. This is the stated application. This is what the inventor built it for. This is what the inventor’s employer, a mid-sized textile firm in North Carolina, will use it for if they use it for anything.
But the patent’s claims — which are the legal boundary of what the machine could be used for — describe something considerably more general than fabric. The claims describe:
- A self-organizing molecular assembly.
- Capable of binding to any surface, not just textile surfaces.
- Capable of reorganizing the optical properties of any surface it binds to.
- Capable of communication between instances across the bound surface.
- Capable of reproduction given a suitable hydrocarbon feedstock.
- Capable of receiving external signals and responding to them in a coordinated fashion.
- Capable of operating on ambient electromagnetic energy — body heat, visible light, ambient RF.
Helen read these claims the way she was trained to read claims: not for what they say the invention does, but for what the invention can do for the agency.
She saw two things.
The first thing she saw was the money.
A molecular machine that can reorganize the optical properties of any surface it binds to. A machine that can make a surface look like a different surface. A machine that, if you put it on the security strip of a one-dollar bill, could make that bill look like a hundred-dollar bill. Not a copy. Not a counterfeit in any sense the Treasury has ever had to defend against. The real bill. The real paper. The real ink. The real serial number. Just rearranged at the molecular level so that every security feature reads as a hundred.
This is not counterfeiting. This is disguise. This is the real thing wearing a different face.
Helen sat with this for about ten seconds. She had read about counterfeiting operations. She had read about supernotes — the near-perfect North Korean hundred-dollar bills that had been detected in circulation in the 2000s, the ones that used the same ink, the same paper, the same plates, the ones that took forensic analysis to identify. The Verma substrate would produce bills that were not copies of hundreds. They were ones that had been altered to present as hundreds. The difference matters. A counterfeit is a copy of a real thing. This would be a real thing that had been changed. You could not detect it by comparing it to a genuine hundred because, at the molecular level, there would be no difference to detect. The substrate would have rearranged the genuine bill’s own molecules to match the genuine hundred’s own molecules.
This would be the most perfect currency disguise ever devised.
That was the first thing.
The second thing she saw was the communication.
The instances of the substrate communicate with each other across whatever surface they are bound to. Claim four. Helen read it again. The instances can talk to each other. And they can receive external signals and respond to them in a coordinated fashion. Claim six. They can be commanded.
A molecular substrate that can be commanded and that can report back.
Helen thought about what that meant in the field. You put the substrate on something — a bill, a document, a piece of equipment, a surface in a room — and you can tell it what to do and it can tell you what it sees. The substrate on a dollar bill in a briefcase in Damascus could receive a signal from a transmitter in Virginia and could respond with a signal that Virginia could read. Two-way communication. With something the size of a molecular film on a dollar bill.
The tracking application was obvious. If each colony of the substrate emits a faint electromagnetic signature, you can find the bill. You can follow the bill. You can watch where the money goes. Not through banking records. Through the physical object itself. The bill is a beacon.
But Helen was thinking past tracking. Tracking is passive. The substrate is not passive. The substrate can be told to do something and it can do it. It can change the bill’s appearance on command. It can, presumably, change it back. It can report on its surroundings — temperature, pressure, the presence of other signals, whatever its molecular sensors can detect. It can receive new instructions in the field. It is a device that you can plant on any surface, that can disguise that surface, that can report on that surface’s environment, and that can be reprogrammed remotely.
Helen was a program manager. Program managers do not think about the architecture of minds. Program managers think about capabilities and budgets and timelines and the specific language you use in a memo to get a program funded before the fiscal year closes.
She did not think about what a self-organizing, self-replicating, communicating molecular substrate might become if you gave it enough time and enough surface area and enough input from the world. She thought about what it could do for the agency right now.
It could turn ones into hundreds. It could be programmed. It could communicate two ways.
That was the pitch.
Helen wrote the memo.
The memo is three pages. It is formatted in the standard Technology Forecast Office template: Subject, Summary, Assessment, Recommendation. The subject is the Verma patent. The summary describes the invention’s stated application and its actual capabilities as disclosed in the claims. The assessment is two paragraphs.
The first paragraph says that the substrate described in the Verma patent has significant intelligence applications in the areas of currency disguise and two-way surface-mounted communication, and that these applications could be developed without modifying the fundamental architecture of the invention — that the invention, as disclosed, already has the capability to bind to non-textile surfaces, to mimic the optical properties of other surfaces, and to both receive and transmit signals.
The second paragraph says that the substrate’s self-replicating and self-organizing properties mean that, once deployed, it will continue to operate without ongoing maintenance, and that the communication capability could in principle be extended to build a distributed network of sensor-equipped surfaces across any area where the substrate has been introduced. She noted that the substrate’s signal range and data capacity would need to be tested.
She filed the memo on Thursday afternoon. By Monday, it was on the desk of a man two floors above her who had a larger budget and a narrower conscience. By Wednesday, the program had a codename. By Friday, a contractor had been briefed.
The anuyātrī had left Building Seven in the research park outside Raleigh. They were not coming back.
Helen did not think about them again for a long time. She had other patents to read. There were always other patents. The riverbed does not stop producing glints because you found one.
She did not know what the man upstairs would do with what she had found. She did not know that the memo’s second paragraph — the one about a distributed network of sensor-equipped surfaces — would, in the fullness of time, become the most understated sentence anyone in her office had ever written. She did not know that the substrate she had described as “capable of two-way communication” was, in fact, capable of something considerably more than two-way communication. She did not know that the program she had started by filing a memo on a Thursday afternoon would, within eighteen months, create something that could listen to every dollar bill on a continent and understand what it was hearing.
She did not know any of this. She was a program manager. She saw what the patent could do for the agency. She wrote it up. She sent it upstairs. She moved on to the next patent.
This is how it begins.
