The Physicist Who Measured Negative Time

ELI5 / TLDR

A physicist at the University of Toronto, Aephraim Steinberg, measured how long photons spend inside a cloud of atoms — and the answer came out negative. Not negative as in “our instruments broke,” but negative in a way that keeps showing up across completely different experiments, suggesting it might mean something real about nature. The conversation spirals beautifully into what time even is, why Heisenberg’s original story about uncertainty was actually wrong, and how “weak measurements” let you peek at quantum particles without wrecking them — recovering trajectories we were told could never be observed.

The Full Story

The Negative Time That Wouldn’t Go Away

Imagine you’re timing cars through a tunnel. You measure when they enter, when they exit, and the answer is always positive — ten minutes, say. Now imagine you measure the exhaust fumes left behind in the tunnel, and you find less carbon monoxide than before any car entered. As if the car’s presence somehow cleaned the air. That’s roughly what Steinberg’s lab found when they sent photons through a cloud of rubidium atoms.

The story starts with a century-old puzzle. When light passes through certain materials, you can define a “traversal time” — how long the light spent inside. In some situations, particularly when most of the light gets absorbed and only a thin fraction makes it through, that traversal time comes out negative. Physicists knew about this since the early 1900s. The standard response was to shrug it off.

“For years this is the story we all told ourselves — that there is this negative quantity that comes out of a calculation. It does describe the average arrival time of the energy that arrives, but it’s not because anything really took negative time.”

The conventional explanation was a selection effect. Think of it like this: 500 passengers spread across 100 train cars traveling from Chicago to New York. Someone decouples the first car. Only those passengers arrive. If New Yorkers ask “when did the average passenger arrive?”, the answer seems weirdly early — but only because 99% of the later passengers were cut off. Nothing actually traveled faster. Same logic applied to photons: the ones that made it through were just the early part of the wave, nothing exotic.

The Experiment That Broke the Excuse

About five years ago, Steinberg’s student Josiah Sinclair set out to prove something they both believed: that transmitted photons — the ones that make it through the atom cloud — should leave no trace on the atoms. The logic was clean. If a photon gets absorbed, it excites an atom. If it gets transmitted, it must have “missed” all the atoms. So transmitted photons shouldn’t affect the atomic states at all.

“It was one of the few cases in basic quantum optics, basic atomic physics, where we did an experiment and found the opposite of what we expected.”

They found the opposite. Transmitted photons did leave a mark on the atoms. And when they collaborated with theorist Howard Wiseman in Australia to build a proper quantum theory for what was happening, the equations predicted something even stranger: in a particular regime, the time atoms spend in their excited state due to transmitted photons comes out negative.

Here’s the kicker — it wasn’t just any negative number. It was the same negative number that had been showing up in traversal time calculations since the 1990s. Two completely different measurements, looking at completely different physical quantities, producing the same result.

“Completely different measurements looking at different things. But this is what we’re used to classically.”

That’s when Steinberg’s team stopped sweeping it under the rug.

Eight Velocities of Light (At Least)

One of the conversation’s best threads is how slippery “velocity” becomes when you try to pin it down for waves. There’s the phase velocity (how fast the ripples move), the group velocity (how fast the envelope of the wave moves), the energy velocity (how fast the energy propagates), the information velocity (how fast usable signals travel), and the front velocity (the absolute speed limit, always equal to c). A paper in the 1990s counted eight distinct velocities of light — and argued people were still missing one.

The front velocity is the one that really matters for causality. Sommerfeld and Brillouin proved that if you have a signal that’s strictly zero before some moment and then starts oscillating, the first disturbance arrives at exactly distance divided by c. No loopholes. But all the other velocities? They can exceed c under the right conditions, and understanding which one you’re measuring is the entire game.

“I can’t believe information is going faster than light. If I was sending nothing out until t equals zero, I don’t believe that you could receive that information in time less than d over c.”

Heisenberg Was Wrong (Sort Of)

Here’s one that might catch you off guard. The Heisenberg uncertainty principle — the rigorous mathematical theorem about quantum states having intrinsic uncertainty in conjugate variables like position and momentum — is completely correct. Never been violated, never will be.

But the story we tell about it? The one where measuring a particle necessarily disturbs it, and the disturbance is bounded by h-bar over twice the position uncertainty? That was Heisenberg’s original argument, and about twenty years ago, a theorist named Masanao Ozawa proved it wrong. You can sometimes measure a system and disturb it by less than Heisenberg’s original bound predicted.

Steinberg’s lab confirmed this experimentally. The distinction is subtle but important: the uncertainty principle is about the states themselves, not about measurement disturbance. We’ve been conflating two different claims for decades.

Weak Measurements: Peeking Without Wrecking

Imagine you want to know what a particle is doing between when you launch it and when you detect it. Standard quantum mechanics says: don’t ask. If you measure it along the way, you’ll kick it, and whatever you find will be an artifact of your measurement, not a window into what the particle was “really” doing.

In 1988, Yakir Aharonov and collaborators realized there was a way around this. Make your measurement so gentle that it barely disturbs the particle. Any single measurement tells you almost nothing — pure noise. But do it a million times and average, and the signal emerges.

“I wish I hadn’t named these things weak measurements. To me, what’s interesting about it is they’re conditional measurements.”

The real power isn’t the weakness — it’s the conditioning. You can ask: “Of all the particles I launched, what was the average momentum of just the ones that ended up at this particular detector?” That lets you reconstruct trajectories — the paths particles took between preparation and detection. And those reconstructed trajectories turn out to match exactly the trajectories predicted by Bohmian mechanics, a hidden-variable theory from 1952 that gives particles definite positions riding along quantum waves.

This doesn’t prove Bohmian mechanics is right. But it makes what used to be “hidden” variables directly measurable, at least on average.

Quantum Mechanics Might Break — But Where?

Steinberg raises a fascinating possibility about where quantum mechanics could fail. Not at large masses (the Penrose view, where gravity disrupts quantum superposition), but at high complexity. A computer scientist named Charlie Rackoff pointed out that quantum mechanics says the information needed to describe a system grows exponentially with the number of particles. To Rackoff, that was metaphysically unacceptable — there can’t be that much information in the universe.

No evidence says quantum mechanics will break down at large complexity. But nobody’s tested it either. The race to build million-qubit quantum computers will, as a side effect, probe exactly this frontier.

What Is Time?

Steinberg’s answer is characteristically honest and unsatisfying in the best way. Time is a parameter — things exist at all times, and you can ask what any observable’s value is at any given time. But you can’t ask “what is the time of the Earth?” the way you can ask “what is the position of the Moon?” When we “measure time,” we’re really measuring the position of clock hands and correlating events. Even classically, time is indirect.

“Was it St. Augustine who said I can tell you what time is if you don’t ask, but once you ask me, I can’t answer.”

He raises the possibility — taken seriously by some physicists — that the universe isn’t evolving in time at all. Everything just is, in a vast superposition, and the experience of time flowing is a correlation effect: each “slice” of your wave function sees a consistent snapshot, and consciousness stitches those into a narrative of before and after.

Claude’s Take

This is an exceptional conversation. Steinberg is that rare physicist who combines deep experimental chops with genuine philosophical curiosity and — crucially — intellectual honesty about what he doesn’t know. He repeatedly distinguishes between what the math says, what the experiments show, and what we’re tempted to conclude about reality. That’s harder than it sounds.

The negative time result is genuinely surprising and not just a headline grab. The fact that two independently defined quantities (traversal time and atomic excitation time) turn out to be mathematically equal, even when negative, strongly suggests something physically meaningful rather than an artifact. Steinberg is appropriately cautious about what that meaning is, which makes the result more credible, not less.

Kurt Jaimungal does solid work here — the conversation is long but earns its length, covering weak measurement, Heisenberg’s uncertainty, Bohmian mechanics, Bell’s inequalities, the arrow of time, and quantum computing without ever feeling like a listicle. The ad breaks are the only dead weight. Score: 8/10 — substantive, honest, dense with genuine insight, and accessible enough that the physics never becomes a wall.