Most conversations about surveillance start from an assumption so obvious nobody bothers to state it: that privacy is the default condition of human life, and that we are currently losing it.
I want to argue the first half of that is wrong, which makes the second half wrong in a more interesting way than it first appears.
Privacy, in the sense we usually mean it, is not ancient. What we mean is the ability to move through the world without the people around you knowing who you are, where you came from, or what you did last year. That is not a birthright that industrial modernity handed us and digital modernity is taking back. It’s an invention, it has a rough birthdate, and if you squint at the timeline it looks less like a foundation and more like a window that opened and is now closing.
The village knew everything
Before cities, there was no privacy worth the name. In a village of four hundred people you were known completely and permanently. Everyone knew your parents, your debts, your religious observance, your drinking, who you were sleeping with and whether you should have been. There was no anonymity to lose. Social control ran through visibility, and visibility was total.
That changed when people began moving into cities in large numbers, and the change was not one that anyone set out to produce. Georg Simmel, writing in 1903 on the mental life of the metropolis, caught it early. What he described was genuinely new: in a dense enough crowd, you become unreadable. Not hidden. Visible to thousands of people, none of whom know you. You could move to a city and become someone else. You could have a first marriage nobody in your second city knew about. You could fail at something in one place and start clean in another.
Call it urban anonymity. It is roughly a hundred and fifty years old in any widespread sense, and it is the thing we have mistaken for the natural state of humanity.
Which means the story isn’t civilization had privacy, then computers took it. The story is: humans lived exposed, invented a way to be unknown, ran that experiment for about five generations, and are now watching the conditions that made it possible evaporate.
What made the old exposure survivable
What made that exposure livable wasn’t restraint on anyone’s part. It was structural, and it came down to four properties, every one of which has now inverted.
It was horizontal. Your neighbor knew your business. Your neighbor was also a person with a farm and no institutional power. Knowledge circulated sideways, among peers, rather than upward into an entity that could act on it at scale.
It was forgetful. Memory decayed. Gossip degraded. Ten years on, the specifics blurred and the story softened, and eventually the people who remembered died. Forgetting was free and remembering was expensive. The natural gradient of human information ran toward loss.
It was non-transferable. What the village knew about you stayed in the village. Move two hundred miles and the file did not follow, because there was no file.
And it was context-bound. The philosopher Helen Nissenbaum has the best formulation of why this matters: privacy isn’t secrecy, it’s contextual integrity. Information flowing according to the norms of the setting where it was shared. Your doctor knowing your weight is fine. Your employer knowing it is not. Nothing about the fact changed. Only the context it moved into. Villages leaked information constantly, but the contexts stayed roughly intact, because information moved at the speed of a person walking.
American law had a name for a version of this too. In a 1989 Supreme Court case, the phrase practical obscurity appears: the recognition that a fact scattered across a hundred courthouse basements is meaningfully different from the same fact compiled into a single searchable index. Same information. Different world. The friction was doing moral work, and nobody had to design it, so nobody noticed it was load-bearing.
The inversion
The researcher danah boyd describes networked information as having four affordances that older social environments didn’t: persistence, searchability, replicability, and scalability. It’s a dry list, and it describes the entire transformation.
Persistence kills forgetting. The gradient reverses. Remembering is now free, and deleting is the expensive, deliberate act that somebody has to choose to perform.
Searchability kills obscurity. Practical obscurity was never a right. It was a side effect of bad indexing, and indexing got good.
Replicability and scalability kill context. Nissenbaum’s contexts held because information moved slowly and locally. When a thing said in one setting can be reproduced perfectly and delivered anywhere instantly, every audience becomes one audience. Researchers call this context collapse, and most people have felt it without having the term: the specific dread of realizing that your coworkers, your mother, and a stranger in another country are all reading the same sentence.
None of this is a forecast. Readers mounted at intersections record vehicle movement continuously. Payment networks observe transactions across merchants that have no relationship with one another. Advertising infrastructure logs browsing across unrelated sites. Phones report position as an ordinary condition of functioning. Almost none of it was built to watch anyone, which is the strange part. The capacity is a byproduct of systems designed to sell advertising, reduce fraud, and lower municipal costs. Most of what gets collected is never examined by a human being, and attention was never the constraint that mattered anyway. What changed is that the past became queryable, by whoever holds it later, for reasons nobody has thought of yet.
Notice the direction all of it runs, because this is the fourth inversion and it isn’t a property of the information at all. It’s a property of who ends up holding it. Village knowledge sat with peers who could do very little with it. This flows upward, toward institutions, because upward is where the storage and the buyers are. So exposure returns, but transformed. No longer horizontal but vertical. No longer forgetful but permanent. No longer local but portable, following you across every border you cross.
This is why the historical analogies feel unsatisfying when you reach for them. We are not going back to the village. The village couldn’t do this. We’re going somewhere without precedent, and the absence of precedent is itself the problem, because our intuitions and our laws were both built for a world in which knowing something about a person took effort.
What the window was for
It’s worth being precise about what anonymity actually did, because we tend to defend it in the language of concealment, as though the point were having something to hide. That framing loses the argument before it starts.
What the window really provided was revisability. The ability to be unknown is the ability to stop being who you were. It’s the immigrant arriving without a reputation. It’s the person who failed publicly at twenty-two and was allowed to be someone else at thirty-five. It’s the possibility of holding an unpopular view provisionally, trying it on, and abandoning it without the attempt becoming permanent record. Anonymity wasn’t a shield for wrongdoing. It was the working space in which a person could change.
Forgetting was the mechanism. A society that forgets is a society that lets you become someone new, and it does that not out of generosity but out of incapacity. It simply cannot hold the file. Viktor Mayer-Schönberger’s argument is that forgetting was always doing this work quietly, and that once remembering becomes cheaper than deleting, we lose a function we never designed and therefore never thought to protect.
What replaces it isn’t punishment. It’s closer to permanent draft-lock. Every version of you stays live and retrievable, and there’s no mechanism by which the earlier ones expire. The self stops being something you author over time and becomes something that accumulates.
The window was open for about a century and a half. We built a good deal of our idea of individual freedom on top of it without ever writing down that it was contingent: that it depended on friction, distance, and the ordinary decay of human memory, none of which anyone designed and all of which we assumed would hold.
What replaces it is not the village. The village had no file. It’s a configuration we haven’t lived in before, and most of the vocabulary we reach for quietly assumes we have. Getting the description right seems like the prerequisite for everything that comes after it.
This essay is part of a series on privacy in the late digital age.