What does measuring actually do?

In everyday language, measuring means a person reading a dial. In physics the person is optional. What matters is that something interacts with the quantum system and keeps a lasting trace of the result.

A detector that clicks counts. So does a speck of dust that a photon bounced off, or a single air molecule that bumped into an atom and flew away carrying information about where the atom was. Physicists call that trace a record. Once a record exists anywhere, it doesn't matter whether anyone ever reads it.

After a measurement, the system lines up with the result, just as each photon in chapter 2 came out of a filter lined up with that filter. Measure again straight away and you get the same answer.

Physicists describe this as the state "collapsing" to the result. That's a description of what we see. What actually happens at that moment, and why this result and not the other, is one of the oldest open questions in physics. Chapter 6 sets out the main answers. This page sticks to what every answer agrees on.

Chapter 4 covers interference properly. For now you need one fact: when a particle has two paths open to it, the two paths can add together or cancel out, just like two overlapping water waves. That produces a pattern of bright and dark stripes, and you only get stripes if the two paths stay unrecorded.

This is the heart of quantum measurement. Interference needs the paths to be indistinguishable. The moment a record anywhere could tell them apart, they stop adding and cancelling, and the stripes disappear. It doesn't matter whether anyone reads the tag. It only matters that the information exists.

Now think about something big, like a speck of dust floating in a room. Every second, enormous numbers of air molecules and photons of light bounce off it. Each bounce carries away a tiny record of where the speck is.

So the world is measuring the speck constantly, far faster than any experiment could. The surroundings wipe out any quantum behaviour it might show, like being spread over two places that interfere, almost instantly. The information leaks into the surroundings and never comes back.

Physicists call this leak decoherence. H. Dieter Zeh first described it in 1970, and Wojciech Zurek developed it through the 1980s. In 1996 Serge Haroche's team in Paris watched decoherence happen step by step in the lab, part of the work that earned him a share of the 2012 Nobel Prize.

Because they can't keep a secret. Its surroundings hit anything big so often that its quantum behaviour washes out before it could ever show. The quantum rules still apply to every atom in a coffee cup. Decoherence hides their effects at large scales.

That's why quantum effects seemed confined to the very small for most of the 20th century. It's not that big things follow different laws. It's that big things are never left alone.

Not quite. Decoherence explains why interference disappears and why the world looks solid and definite. It doesn't, on its own, explain why a particular measurement gives this result rather than that one. That's the famous measurement problem, and it's exactly where the interpretations in chapter 6 disagree. Schrödinger dramatised it in 1935 with his cat. Optional pages 3a and 3c tell that story.

Decoherence is the number one enemy of quantum technology. A qubit is only useful while its surroundings have no record of its state. Every stray bump, vibration, photon of heat or magnetic flicker is a tiny measurement that spoils it.

So engineers go to extreme lengths to keep qubits alone:

• Cold. Superconducting quantum computers run at around 15 thousandths of a degree above absolute zero, colder than outer space, inside the golden "chandelier" fridges you see in photos.
• Vacuum. Trapped-ion and neutral-atom machines hold their atoms in a near-perfect vacuum with lasers and electric fields.
• Shielding. Layers of metal block stray magnetic fields and radio signals.
• Speed. Do the calculation before decoherence wins.

Even then, decoherence always wins eventually. The best superconducting qubits made in the lab now keep their state for around a thousandth of a second. Trapped ions can last much longer. That's why chapter 8, on error correction, matters so much.