There was a time when finding your way required more than a signal.
In the early years of my military service, I used a PLGR, a Precision Lightweight GPS Receiver. The name suggested something more refined than the device itself. It was large, deliberate, and slow to orient. Acquiring a signal could take time, and until it did, you were left with little more than a set of expectations and your own sense of direction. It did not guide so much as confirm. You still had to know where you were going.
A few years later, I purchased a commercial hiking GPS. It was smaller, more responsive, and capable of saving routes. But it offered no map, no broader context. It could tell you where you had been, and where you were in relation to that path, but not what surrounded you. It assumed a kind of prior understanding. You navigated first, and verified after.
I still have that device. It struggles now to lock onto satellites, as if the sky itself had grown more distant.
Not long after, I began using a cell phone that approximated location through nearby towers. It worked well enough, until it didn’t. Somewhere between San Diego and Palm Springs, the signal dropped, and with it, any clear sense of where I was in relation to where I intended to be. The system had not failed dramatically. It had simply stopped providing an answer.
Now, navigation is ubiquitous. The map is always present. The route is calculated instantly, adjusted continuously, and presented without hesitation. We no longer ask where we are and then determine a path.
The path is given, and we follow.
It is easy to experience this as progress, and in many ways it is. But I find myself wondering what, exactly, has been exchanged. Earlier systems assisted judgment. They confirmed, corrected, or refined what you already understood. The newer systems do something quieter. They begin to replace the need for that understanding. It is not only navigation that has shifted in this way, and they do so by asking for something in return.
Verification, it seems, has not only been automated. It has been delegated. In earlier systems, to be trusted required recognition. Someone, somewhere, had to accept that you were who you claimed to be. Even when that recognition was mediated by documents or credentials, it still retained a human element. There was, at least in principle, someone to convince. That condition has quietly changed.
We now exist within systems that do not know us in any meaningful human sense. They do not remember us, except as patterns, and do not recognize us except through signals. To move through them is not to be known, but to be evaluated. There is, increasingly, no one to convince.
This is not always obvious, because the language remains familiar. We still speak of logging in, verifying identity, and confirming access. But the underlying process has shifted. The system does not ask who you are. It asks whether the attributes associated with you satisfy its conditions. To be verified is no longer to be recognized, but simply to pass.
The decision remains, but the decision-maker is no longer present. Or rather, the decision-maker has been displaced, distributed into a system whose logic is visible only in its outcomes. I find myself unsure what to make of that absence. It offers a certain efficiency. Decisions are made quickly and consistently, without the variability that accompanies human judgment. There is a clarity to it that is difficult to deny, and yet it alters the experience of being judged.
To be refused by a person is frustrating, but intelligible. There is, at least, the possibility of explanation. One can ask why or attempt to persuade. Even when those efforts fail, they retain a certain shape. To be refused by a system is different. The outcome appears, but the reasoning does not. There may be an appeal, though it often leads only to another system. The process continues, but the point of contact remains elusive, and it is not always clear what, exactly, is being asked to be proven.
Up to this point, such systems have concerned themselves primarily with what has already occurred. Did you enter the correct credentials? Does your device meet the expected standard? Does your behavior resemble what has been observed before? These are questions about the present, or at most, the immediate past.
But systems that collect enough signals do not remain confined to verification for long. They begin to anticipate. A system that can evaluate behavior can also model it. If it knows where you are, how you typically act, and what sequences of behavior tend to precede certain outcomes, it can begin to form expectations. Not certainty, perhaps, but probability.
At that point, the question begins to change. It is no longer simply whether an action is valid. It becomes whether that action is likely, and from there, whether that action should be allowed at all. In some cases, the question extends further, to whether the action should be made possible in the first place at all.
We are already familiar with early forms of this logic, though we may not always recognize them as such. Financial systems flag transactions before they are completed. Platforms suppress content before it is widely seen. Security systems restrict access based on patterns that deviate from expectation. Each of these is presented as a form of protection, and often, it is. But the structure is worth noticing. The system is no longer responding to what has happened. It is acting on what it expects to happen.
There is a faint resemblance here to older speculative fears. In Minority Report, intervention occurs before the act, justified by the confidence of the system. Our systems are less certain. They deal in probability, not prophecy, but they require far less certainty to act.
I am not sure where the boundary lies. Prediction is not new. Human beings have always made judgments about what others are likely to do. We do it constantly, often without reflection. The difference, perhaps, is not that prediction exists, but that it has been formalized, scaled, and embedded into systems that operate without pause. The judgment is no longer occasional. It is continuous, and it carries a different kind of weight.
When a person misjudges you, there is at least the possibility that the mistake can be corrected through interaction. When a system does so, the error may persist quietly across contexts. We have seen something like this already in the quiet persistence of errors in financial records, where the burden of correction falls not on the system that made the judgment, but on the individual required to contest it. It may shape outcomes without ever being fully visible, and may not even be recognized as an error, because the event it prevented never occurs. When an action is denied before it happens, it becomes difficult to say what, exactly, has been denied. Nothing has occurred, and yet something has been prevented. The failure exists in a future that never arrives.
I find myself returning, perhaps too often, to the same uncertainty. These systems offer real advantages. They reduce certain kinds of harm and make possible forms of coordination that would otherwise be difficult to sustain. It would be easy, and perhaps comforting, to describe them as purely negative. I do not think that would be accurate, and yet I am wary of what they require in return.
We are asked, increasingly, to present ourselves as sets of attributes. Those attributes are evaluated continuously, and from those evaluations, expectations are formed and decisions are made, not only about what we have done, but about what we are likely to do. We no longer choose the path and then confirm it. Increasingly, the path is suggested before we decide to walk it.
It may be that this is simply the next stage in a longer development. Identity becomes infrastructure. Trust becomes verification. Verification becomes anticipation. Each step appears reasonable when viewed on its own. Taken together, they describe a system in which we are not only asked to prove who we are, but measured against what we might become.
I am not sure when verification became anticipation. Only that the question is no longer simply whether we can be trusted. It is whether we are predicted to be.
