Earlier this week, Charles A. Blanchard floated a provocative idea: “As strange as it may seem, quantum mechanics might help us illuminate the best approach to restrictions on surveillance.” Blanchard asserts that bulk-collected metadata amounts to “nothing more than a collection of ones and zeros,” comparing it to an opaque state of quantum superposition that is difficult to describe and impossible to observe. But when a government analyst searches the data, Blanchard writes, “[t]his all changes”: the “ones and zeros become meaningful,” just as a “quantum wave function collapses into certainty by measurement.” Blanchard concludes that quantum mechanics teaches us that “[o]ur true privacy interests are implicated by the use of the compelled collected data, . . . and not by the collection itself.”

Blanchard’s approach is inspired, but his conclusion is flawed. Here’s what he says:

Just as quantum mechanics teaches that matter has no certainty in position, spin, momentum and other properties until we actually seek to measure or observe matter, the privacy interests at the core of the Fourth Amendment are only implicated when officials search the data. . . .

Here is where quantum mechanics comes into play. According to quantum mechanics, matter such as an electron or photon is described by a probability wave function and does not have a precise physical state. This does not mean merely that we don’t know the precise state of a proton or electron—it means that the proton or electron truly has no precise state at all. But when we attempt to measure the state of a proton or electron, something almost magical occurs—the probability wave function collapses into a measurable physical state.

But this isn’t quite right.

Quantum mechanics is fundamentally about probability: A thing is not just a thing, but a range of probabilities concerning its thingness. That we observe just one “reality” among these probabilities when we observe or measure that thing is one of the nagging problems at the core of modern quantum mechanics—what’s the relationship between the math and what we see?

Remember “Schrödinger’s Cat”? A cat is placed inside a box with a radioactive substance, a Geiger counter, a vial of poison, and a hammer. When the radioactive substance decays, the Geiger counter detects it and sets the hammer in motion, breaking the vial and killing the cat. Because radioactive decay is random, we can’t know when it will happen. But we set up the experiment so that we know that, by the end of an hour, there’s a 50% chance the cat will be dead, and a 50% chance it will still be alive.

When we open the box, we observe that the cat is either alive or dead. But—more interesting—before we open the box, all we know for certain is the array of probabilities surrounding the cat; thus, we might say (as Schrödinger did) that the cat is both “living and dead . . . in equal parts.”

That idea strikes us as absurd—yet that’s what the math suggests. And that’s the point—the paradox is supposed to show us something about the weirdness of quantum mechanics. It also shows us that probabilities are often more important than the observed result, because they tell us more about the world than what we can observe from our own narrow existential vantage point.

What’s true for cats is true for privacy as well. Thus, we could use quantum mechanics—or simple probabilities, anyway—to help us think about the likelihood that different arrangements (boxes) would invade American privacy. And that’s effectively what the entire metadata debate is about, with the familiar variables: government collection or retention by telecoms; court orders before queries or executive-determined “reasonable articulable suspicion”; one hop, two hops, or three; the computer programming we use to structure storage and queries; the oversight mechanisms we put in place; data security; &c.

Call it “Schrödinger’s Metadata”: How can we set up this “experiment” in a way that will increase the odds that Americans’ privacy will remain protected—irrespective of whether we observe a particular invasion or not? After all, we shouldn’t accept a system that makes privacy violations extremely likely, even if—by chance—we don’t observe those violations directly.

Thus, Blanchard is on shaky ground when he describes the bulk collection of metadata as “ones and zeros” without “meaning[].” Just as our observation of the cat upon opening the box is just one expression of a range of potential outcomes, so is a single search of collected metadata. And by focusing on that range, we can make it more likely that the cat (and our privacy) makes it out of the box alive.

Incidentally, Blanchard’s assertion—that it is observation itself that “collapses” the array of probabilities into one observable reality—is just one of several competing theories that purport to resolve the paradox that Schrödinger’s cat seems to present. Another far weirder but equally prominent theory holds that while we observe just one reality, there are other equally valid, parallel realities representing the other probabilities.  Even under Blanchard’s chosen theory, though, it’s misleading to say that our own act of observation causes the collapse of the probability wave function—and Blanchard’s use of the word “magical” perhaps admits as much. It’s more accurate to say that our observation changes the nature of our knowledge about the reality before us, and eliminates the need for us to rely on probabilities to describe it.

The point is that we’re far less certain about what happens when we open the box than we are about how to describe what’s inside of it. If quantum mechanics has any application here at all, it’s to suggest that how we set up the “box” in the first instance—that is, how we regulate collection of sensitive data—is what truly matters.

Brett Max Kaufman is the National Security Fellow in the ACLU’s National Security Project.