The NSA's Future Files: Grading the Spy Agency's Crystal Ball
Locked away for 15 years by a FOIA request, eight NSA technology forecasts are finally public — brain-network interfaces, terahertz x-ray vision, battery-free devices, the end of Moore's Law. What did the agency see coming, what did it miss, and what does the scorecard teach about prediction itself?
By MyAudioBooks.ai ·
Listen free: The NSA's Future Files: Grading the Spy Agency's Crystal Ball
Somewhere in the National Security Agency, there is an office whose job is to see the future — not the future of wars or elections, but the future of technology itself: what will be possible in ten years, in twenty, in thirty, and what it will mean for the agency that listens. Every intelligence service has such an office, because every intelligence service lives or dies by the same terror: that the world will invent something the agency did not see coming, and that the first hint of its existence will arrive too late to matter. The forecasts these offices produce are among the most closely held documents in government — not because they contain secrets exactly, but because they contain the map of what the secret-keepers are watching for, and a map of what a spy agency fears is itself intelligence. This year, after a Freedom of Information request that had been pending since two thousand eleven — fifteen years, longer than some of the technologies it asked about have existed — a set of these forecasts was finally released to the public: eight National Security Agency technology studies, written years ago, covering subjects from brain-network interfaces to the materials that would replace silicon. They are now readable by anyone. And because they were written then and we live now, they come with a gift no forecast ever has when it is written: a scorecard. We can check the spy agency's crystal ball against the world it was trying to see — what it got right, what it got wrong, and what the difference teaches about prediction itself.
At My Audio Books dot A I, you can create your own audiobooks from prompts, turn your documents into audio, all with one subscription, and store your items in your own personal library.
First, understand the document, because a technology forecast is a stranger artifact than it sounds. It is not a prediction in the fortune-teller's sense — the agencies that write them are usually careful to disown the word. It is a structured act of imagination: a disciplined attempt to answer, for some emerging field, the questions an intelligence service actually needs answered. How does the underlying science work, stated precisely enough that an analyst can follow it? What is the current state of the art, worldwide, including the laboratories the agency does not have access to? What is physically possible in principle, what is merely engineering, and what is fantasy? Who is working on it, funded by whom? And above all: what happens to us — to the mission, the collection, the adversaries — if it works? The finished study is part textbook, part threat assessment, and part letter to the future, written by people who know they will be graded by it. The eight studies in this release span the full strangeness of the early twenty-first century's technological frontier: the interfaces that would connect brains directly to networks; the terahertz frequencies that would see through clothing and walls; the harvesting of energy from the environment to power devices that never need batteries; and the materials sciences that would carry computing past the end of Moore's Law — the sixty-year doubling rhythm that the entire modern world was built on and that everyone, even then, knew was ending.
Now the scorecard, because that is what fifteen years of hindsight buys. On the brain-network interfaces, the forecast reads today like a summary of the present: direct connections between neural tissue and computers, first for medicine — restoring movement to the paralyzed, speech to the silent — and then, in the forecast's careful longer arc, for augmentation, communication, and control. The medical half of that future has arrived on schedule: brain-computer interfaces have now been implanted in human patients by multiple companies, paralyzed patients are moving cursors and robotic arms with their thoughts, and the first speech prostheses driven by neural signals have given voices back to people who lost them. The augmentation half remains, as the forecast suggested it would, contested and slow — but the trajectory it drew is the one the field is on, and the security questions it asked — what does collection mean when the signal source is a brain — read less like paranoia every year. On terahertz systems, the forecast tracked a technology that sees through barriers — imaging through clothes, packaging, and walls — and here the scorecard is more humbling for the agency: the physics worked, but the revolution did not arrive on time; terahertz imaging found its niches in security screening and industrial inspection rather than becoming the universal x-ray vision the optimists promised, a reminder that physical possibility and deployment reality are different countries. On energy harvesting, the forecast anticipated the trickle-power world — devices sipping energy from light, vibration, heat, and radio waves, never needing a battery change — and that world has quietly arrived at the level of sensors and small devices, the invisible infrastructure of the Internet of Things, while the grander visions of battery-free everything remain, as ever, ten years away. And on the post-Moore's-Law materials, the forecast asked the right question of the right century: what comes after the doubling stops? — and the answer the world actually delivered was stranger than any of the scenarios: not a new substrate but a new architecture, the age of specialized chips and massive parallelism, in which Moore's Law died and computing power kept doubling anyway, through paths the forecasters saw only in outline.
At My Audio Books dot A I, you can listen to this story and thousands of others that explore the hidden science and mechanics behind the headlines.
The strongest case for the forecasters — stated at full strength, because the record deserves it — is that horizon-scanning of this kind is one of the most underrated instruments of statecraft, and these documents show it working as designed. The purpose of a technology forecast is not to be right; it is to make the future legible early enough to prepare for it, and by that standard the scorecard is strong: the fields the agency flagged a decade-plus ago are, without exception, the fields that matter today — neural interfaces, through-barrier sensing, battery-free devices, and the end of the silicon roadmap are the actual frontiers of twenty twenty-six, not the false trails that consumed the era's hype cycles. The studies' technical cores are sober where public futurism was drunk: the same years that produced breathless magazine covers about teleportation and the singularity produced, inside the agency, careful maps of what the physics actually allowed and who was actually doing it. And the document that asks the right question is worth more than the document that guesses the right answer — the post-Moore study, which missed the architectural turn but named the wall the industry would hit, served the mission better than a hundred correct trivia predictions would have.
And the strongest case against the crystal ball — stated with the sympathy the forecasters have earned — is that the scorecard's pattern is the oldest one in prediction: right about the science, wrong about the world. Every technology in the release arrived later, slower, and more sideways than the forecasts' framings implied — not because the physics was misread but because deployment is not physics; it is economics, regulation, accident, and fashion, the four horsemen that mock every roadmap ever written. The forecasts also share the institutional blind spots of their authors: they watch the technologies a signals agency fears — sensing, computing, communication — and the decade's actual earthquakes, the mRNA platform and the transformer architecture, were biological and statistical rather than electrophysical, outside the frame entirely. And there is a subtler failure built into the exercise itself: a forecast that is read by the people it forecasts about changes the thing it forecasts — the agency's watching alters what the watched build, and the document that maps the future becomes one of the forces shaping it, in ways no scorecard can disentangle.
Three developments would sharpen what this release ultimately teaches, and each is observable. First, the next tranche — the next slice of a larger release, since these eight studies come from a bigger forecasting program: and further releases — the FOIA pipeline that produced these did not close — will show whether the program's hits and misses follow the same pattern across more fields and more years. Second, the current forecasts' eventual unsealing: the studies being written inside the agency today — about the technologies of our moment — will be judged by the same test in fifteen years, and the continuity or evolution of the program's blind spots will tell us whether institutions learn to see better or merely see differently. Third, the fields themselves: if neural interfaces continue the trajectory the oldest forecast drew — from medicine toward augmentation and control — the release will stand as the document that saw the century's deepest technology coming; if the field stalls where it is, the forecast becomes instead the record of a plausible future that physics allowed and history declined.
It is worth saying what this article has not claimed. It has not claimed the forecasts were secret prophecies; they were disciplined staff work, and the article says so. It has not claimed the agency saw everything coming; the blind spots are documented here at length, including the two largest technologies of the era. It has not claimed any current NSA capability from these documents; they are studies of public science, not operational records, and the article treats them as such. And it has not claimed prediction is useless; the strongest case for the craft is stated here in full. The claim is narrower and more durable: the people charged with seeing the future saw more of it than the public did, missed it in instructive ways, and left us the receipts — and the receipts are now public, fifteen years after someone first asked for them.
Which returns to the office, and the terror that keeps it in business. Somewhere in the same agency today, another analyst is finishing another forecast about another technology that does not quite exist yet — writing the textbook and the threat assessment and the letter to the future, knowing the grade arrives in fifteen years, in a world they are trying to imagine from a desk. The eight studies now public say the desk is a better vantage point than we knew, and a worse one than the agency needed, and both at once — which is, in the end, the only honest grade prediction ever earns. The crystal ball is cloudy. It is also, these documents show, the best one we've got — and now, for the first time, you can read it yourself.
At My Audio Books dot A I, you can create fiction, non-fiction, and turn your documents into audio, all stored in one place with a single subscription — plus get instant access to thousands of audiobooks and deep-dive investigations. Learn more today at My Audio Books dot A I.