Last episode (check here) we saw how a quantum internet could ride on top of the internet we already have, and we ran straight into its biggest bottleneck: loss in optical fibers. Photons don't survive long journeys through optical fibers.
So let's look at one way around it.
What Happens in the Classical World?
We amplify. When a signal weakens, a station along the route reads it and sends out a fresh, strong copy.
That's exactly what we can't do in quantum. Remember why we're sending these photons — to establish a secret key. The last thing we want is someone opening the envelope on the way.
You might object: doesn't today's internet already do this? Don't the amplification stations see the message?
Not quite. What travels through them is encrypted. A station can see the encrypted message, copy it, and throw computing power at it later. That's precisely why today's internet isn't 100% secure, and why cybercrime exists at all.
You can do the same in quantum: open the message, re-send it, accept that the stations know what's inside. These are trusted node networks — what China implemented in its 2000 km link.
If your network is small, or you genuinely trust every node, it works.
But the interesting question is: what if you don't trust them?
The Real Problem
A quantum internet doesn't need messages delivered — it needs entanglement between Alice's particle and Bob's particle. Once they share entanglement, they can squeeze a secret key out of it that nobody else can hold a copy of. That's the whole game.
But entanglement isn't something you can mail. Two particles become entangled by interacting — by meeting, physically, in one place. So how do you entangle a particle in London with a particle in New York, when they have never been interacted with each other and neither can survive the trip?
Two Bits That Add Up to Even
Every qubit here will eventually give you a 0 or a 1 when you measure it. Not before. Before measurement, there is no answer sitting inside waiting to be read — the bit is genuinely undecided.
But two qubits can be entangled, and entanglement is a rule about their sum.
Alice prepares a pair, qubits A and B, in her lab where they can interact. The rule she builds into them:
A + B is even.
That's all. Both 0, or both 1 — 50/50, undecided, but linked. Measure qubit 1 and get a 0, and qubit 2 is instantly a 0 too. The sum was fixed from birth; the individual values never were.
Alice keeps A and sends B to a station halfway.
Bob, a thousand kilometers away, does the same thing from scratch. His own pair, qubits C and D, his own rule:
C + D is even.
He keeps D and sends C to the same station.
Now count what exists. Two rules:
A + B is even
C + D is even
And between qubit 1 and qubit 4 — the two Alice and Bob actually hold — nothing. No rule at all. Different labs, different sources, no shared history. Alice's bit and Bob's bit are independent coin flips.
The Trick
The station holds qubit B and qubit C, one from each chain.
It does not measure them individually. If it asked "is qubit B a 0 or a 1?", it would collapse Alice's chain and learn her bit — and we'd be right back to trusting it.
Instead it asks one question about the two together:
Is B + C even or odd?
Not what they are. Only how they add. And here's the point: that question has an answer and you don’t have to measure each of them to know it! The station can learn the parity of the sum while both qubits stay genuinely undecided.
Say the answer comes back: 2 + 3 is odd.
Now stack the three rules:
A + B is even
B + C is odd
C + D is even
Now what is A + D? Carefully figure out that:
A + D is odd.
A rule. A brand-new rule, between Alice's qubit and Bob's qubit. Two bits that were independent coin flips a moment ago are now locked: whatever Alice gets, Bob gets the opposite.
The station's question welded the two chains into one — and then dissolved itself. Qubits 2 and 3 are consumed by the measurement, gone. What's left is a rule stretching from Alice straight to Bob, across a gap their particles never crossed.
They were never entangled. Now they are. That's entanglement swapping.
Think for a minute what happens if B + C is even? Convince yourself that it is useless and we discard that particular attempt.
Why the Station Learns Nothing
This is what makes swapping the hero, not just a clever alternative to amplification.
A trusted node holds your actual secret and you simply hope it behaves. The swapping station never holds the secret — the secret doesn't exist yet when the station does its job. Alice and Bob only create the key afterward, by measuring qubits 1 and 4, which the station never touched.
Yes, the station has to announce its result — "odd!" — publicly, so Alice and Bob know whether to flip their bit. But announce it on live television. "Those two qubits summed to odd" says nothing about whether Alice's qubit is 0 or 1. It was undecided then. It stays undecided until she measures.
The station learned a relationship. Alice and Bob own the values. Those are different things, and that gap is the entire security of the scheme.
The middle node can be run by your worst enemy. It still works.
The Quantum Repeater
Now the picture completes itself. If one swap can weld two short links into one long one, many swaps can weld many.
Take a distance no photon can survive. Chop it into hops short enough that photons do get through. Build entanglement across each hop independently. Then swap, and swap, and swap — parity rules chaining into parity rules, cancelling out every qubit in the middle, until one unbroken rule stretches from Alice to Bob across the entire route.
Divide, conquer, and combine. That's a quantum repeater.
And did you know you can do all this with satellites instead? Next episode!
Bonus for serious people
As you might have realised, entanglement swapping is the idea that makes the whole quantum repeater and quantum internet possible. Here is one of the first experimental papers if you are interested!
