🔬 Myth Autopsy

Six claims you have almost certainly read about quantum computing. Each one is dissected below with a live simulation you can run yourself — not a diagram of the argument, the actual physics, computed in your browser.

We are not trying to make you feel clever for knowing better. Most of these myths exist because the honest version is genuinely harder to say in one sentence. So each autopsy ends with a sentence you can actually use — one that is both true and short enough to survive a dinner party.

The best one is #2. You get to hold both halves of an entangled pair and try to send a message faster than light. You will fail. Physics will not budge. That failure teaches more than any paragraph could.
1 · 0 and 1 at the same time 2 · Spooky instant messaging 3 · Tries all answers at once 4 · More qubits = faster 5 · Encryption breaks tomorrow 6 · Consciousness causes collapse
The claim A qubit can be 0 and 1 at the same time.
The most repeated sentence in popular quantum computing. It is also the first thing every physics student has to unlearn.

If a qubit really were holding both values simultaneously, then measuring it ought to sometimes give you both — or something in between. Let's check. Below is a qubit in a perfectly even superposition. Measure it as many times as you like and watch the list of values you have ever seen.

Live · every outcome ever recorded

No measurements yet.

Run it a thousand times. You will get 0s and you will get 1s, and you will never once get "both", "0.5", or a blend. Every single measurement returns one plain, definite bit.

So what is real before you look? The amplitudes — and they are completely definite. They fix the odds exactly. What is not yet a fact is the outcome.

A spinning coin is the honest picture. Mid-spin it is not heads-and-tails at once, and it is not secretly already heads. Its face is not yet a fact — but the way it is spinning determines precisely how likely each face is.

Say this instead A qubit in superposition has no definite value yet — but it does have a definite recipe of amplitudes, and that recipe fixes the odds of each outcome exactly.
The claim Entangled particles communicate instantly across any distance.
Usually followed by "…so quantum computers could enable faster-than-light communication." They cannot, and the reason is beautiful.

Here is a real Bell pair, (|00⟩ + |11⟩)/√2. Alice holds the left qubit; Bob holds the right one, light-years away. Your job is to break physics.

Try to send Bob a message. Measure Alice's qubit as often as you want, in whichever basis you want. Then look at Bob's histogram and see whether anything you did shows up in it.

Live · try to signal

👩‍🔬 Alice's results

—

👨‍🚀 Bob's results (he measures in Z, always)

—

Nothing measured yet.
Did you manage it? Switch Alice's basis. Measure, don't measure, alternate. Bob's histogram sits at 50/50 every time, to within ordinary shot noise. There is nothing you can do on Alice's side that changes anything Bob can observe on his.

And yet the correlation is completely real. Look at the third line in the readout: when Alice measures in Z, her results and Bob's match 100% of the time. The link is not an illusion.

So where does the message go? Nowhere — because Bob cannot see the correlation from his side alone. To find it, someone has to bring the two lists of results together and compare them line by line. That comparison travels by email, phone, or spacecraft: at or below the speed of light, like everything else.

Entanglement creates correlation, never communication. This is called the no-signalling theorem, and it is why quantum mechanics and relativity manage to coexist.

Say this instead Entangled particles give perfectly correlated results — but neither side can see that correlation alone, so nothing is transmitted. You need an ordinary, slower-than-light message to reveal it.
The claim A quantum computer tries every possible answer at once, then returns the best one.
The standard explanation of quantum speedup. The first half is nearly right; the second half is where it falls apart.

It is true that a register of n qubits can carry amplitude on all 2n possibilities at once. The problem is the word returns.

Below are 5 qubits in an equal superposition — genuinely all 32 answers, simultaneously. Now measure it and see how many of those 32 you get to read.

Live · 32 answers in, one answer out

Not run yet.

One bitstring. Every time. Measurement gives you exactly one answer per run, and the superposition is gone the moment you look.

Press the second button and watch what an actual quantum algorithm does. It does not read all the answers — it spends its effort rearranging the amplitudes so that the good answer becomes the likely one to fall out. Wrong answers are made to cancel; the right one is made to reinforce. That is interference, and it is the whole craft.

This is also why quantum algorithms are hard to invent. "Check everything in parallel" would work for any problem. Steering interference toward the right answer has to be designed problem by problem — which is why we have a handful of quantum algorithms and not a general-purpose one.

Say this instead A quantum computer holds many possibilities at once but only ever hands back one. The skill is arranging interference so that wrong answers cancel and the right one is the one you are likely to measure.
The claim More qubits means a faster computer.
Reinforced by every headline that reports a new chip purely by its qubit count.

Qubits without interference are not an advantage — they are a bigger haystack. Watch what happens to your chance of guessing right as you add qubits to a plain superposition.

Live · your odds of hitting the right answer by measuring

Every qubit you add halves your chance of measuring the answer you wanted. Scale alone makes the problem worse, not better. What turns those 2n possibilities into an advantage is an algorithm that concentrates amplitude — and if you do not have one for your problem, the extra qubits buy you nothing.

There is a second reason the headline number misleads. Real qubits are noisy, and noise compounds: more qubits and deeper circuits mean more chances for the state to decohere before you finish. That is why the field talks about error-corrected or "logical" qubits, where thousands of physical qubits may be needed to make one reliable one. A 1,000-qubit chip is not a 1,000-qubit computer in the sense the headline implies.

Say this instead Qubit count on its own means very little. What matters is how many stay coherent long enough, how low the error rate is, and whether an algorithm exists that can turn that space into interference.
The claim Quantum computers will break all encryption — any day now.
Sometimes sold with a countdown clock. The honest version is more interesting and much less dramatic.

Three corrections, in order of how much they change the picture.

1. Not all encryption. Shor's algorithm breaks cryptography based on factoring and discrete logarithms — RSA and elliptic-curve, which is most of what secures the web today. It does not break symmetric ciphers like AES-256, and it does not touch modern hash functions. Grover's algorithm weakens symmetric keys, roughly halving their effective length — which is why AES-256 is considered fine and AES-128 less so. Doubling a key length is an inconvenience, not an apocalypse.

2. Not on this hardware. Factoring a 2048-bit RSA key with Shor's algorithm needs on the order of millions of physical qubits once error correction is accounted for, running coherently for hours. Today's largest devices have hundreds to low thousands of physical qubits with error rates far too high. The gap is not a rounding error; it is several orders of magnitude, and closing it requires advances nobody can currently schedule.

3. The replacement already shipped. In 2024 NIST standardised post-quantum algorithms designed to resist quantum attack, and browsers and messaging apps have been rolling them out since. The migration is genuinely urgent — because of "harvest now, decrypt later", where an adversary stores your encrypted traffic today and decrypts it whenever the hardware arrives — but it is an ordinary, solvable engineering migration, not a cliff edge.

Say this instead Large-scale quantum computers would break today's public-key cryptography, but not symmetric encryption, and not with any hardware that exists or is scheduled. The quantum-resistant replacements are already standardised and shipping.
The claim A quantum state collapses because a conscious observer looks at it.
The word "observer" did enormous damage here. It is a technical term that sounds like an everyday one.

In physics, "observation" means interaction that leaves a record — a photon scattering off the system, a detector firing, a stray air molecule bumping into it. No mind is required, and none is involved.

The evidence is blunt: quantum states decohere constantly in sealed apparatus with nobody watching, which is precisely why quantum computers are kept in near-vacuum at a few millikelvin. Every engineering hour spent on shielding is an hour spent preventing "observation" by an entirely unconscious universe. If consciousness were the trigger, that hardware would be pointless.

It also runs the wrong way round historically. Decoherence theory explains the appearance of collapse through ordinary interaction with the environment, and it does so quantitatively — predicting how fast a given system loses coherence, numbers that are measured and matched in the lab.

There is a real open question underneath the myth — the measurement problem, about what collapse fundamentally is, and physicists genuinely disagree. But no mainstream interpretation gives consciousness a causal role.

Say this instead "Observation" means any interaction that records information about the system. A detector, a stray photon, or a warm wall will do it. Consciousness has nothing to do with it.

Why we built this

Every one of these myths is easier to say than the truth. That is why they spread — not because anyone is being dishonest, but because the accurate sentence is longer and the catchy one travels.

Our whole course is built on the bet that you can be completely honest and still be gentle, if you are willing to do the harder authoring work. Every "say this instead" above is our attempt to prove it: accurate and short enough to actually use.

Start Basecamp 1 → Test yourself first