Please Stop Counting Badly
Sun Aug 31 2025
A recent quantum computing event I attended marked about the millionth time I’ve heard people in quantum computing make a particularly misleading statement, and it finally pushed me to actually write something about it. This will be a touch more technical than my usual pieces, but also short so I’d suggest giving it a go, but no hard feelings if you decide to take your leave early.
The Statement of my Pain
Almost any time someone talks about quantum computing to the general public, they will say something along these lines:
Quantum entanglement is key to a quantum computer’s power, it means qubits scale exponentially and can store and manipulate exponentially more information than classical bits.
(See for example McKinsey).
There are two glaring errors in this thesis, and I intend to destroy the first and improve the second.
Part 1: The counting is the same
The basic unit of a computer is called a “bit,” just an extremely simple switch, on/off, 1/0, whatever dichotomy you like. It is shocking how much we’ve been able to do just manipulating gigantic collections of these switches, you should take a minute to be impressed.
The idea behind quantum computing is to replace “bits” with quantum bits, or “qubits.” Qubits are like switches in that when you poke them you will find them to be either on or off, 1 or 0, but they are statistically tunable, you can dial how likely it is that they will be found to be on or off when you poke them (see my earlier post for more detail). There are clever ways to use this tunability to solve a couple of problems quickly and the industry of quantum computing is constantly, aggressively searching for more use-cases.
But here’s the key point. The data in a qubit is the same as the data in a bit, the 1 or the 0. The tunable likelihood is a new operation you can do to manipulate the data, but it is not a new place to store data. If you want to use the qubit’s tuning knob as its data, you have to design an entirely new architecture and spoiler: quantum mechanics says that device would be worthless1, so good job.
When people say that “qubits can store/manipulate exponentially more data,” they are implicitly transferring the meaning of data from the 1s and 0s to the tunable likelihoods, secretely designing their worthless new devices. I want to say a bit more about that but it’s not important for my story today, the important thing is this claim is wrong and deceptive and I need people to stop saying it. A bunch of qubits stores the same amount of data as a bunch of bits, that’s it, the counting is the same.
Part 2: Thank goodness the counting is the same
Now for the fun part: the counting is the same but it didn’t have to be, it could have been worse.
See, mathematically bits and qubits are different things. If you grab a handful of switches (bits), you can just count how many different possible combinations of on and off you can have (there are ). But qubits are different, they are vectors2, and there are different ways that vectors can combine with each other. Realistically there were two options Nature could have chosen: one exponential and one not, and the difference between these two is the existence of “entanglement.” I won’t explain what that means physically (I talk about it in detail in a previous post) because all that matters for me here is the counting.
Here’s a quick example. Say we have 3 bits: they can take any of 8 unique combinations,
In a world where Nature chose to let quantum things “entangle,” these are also all unique combinations for qubits, but in a world without entanglement, they are not all independent. In fact, any two of them are equivalent to some combination of the remaining 6. Depending on how you write it, you could for example say .
Again, the point of using qubits is not representing data differently, it’s about doing more with the same data. In a world without entanglement, we would suffer diminishing returns using more qubits as the new things you could do would not scale the same as the data you want to manipulate. Entanglement evens that out, so computationally, it’s really just a relief that we have entanglement because otherwise quantum algorithms would have to adapt for this kind of redudancy and would in a sense lose power as systems grow.
In Summary…
Computationally, entanglement means quantum algorithms generically scale, you don’t have to adapt them for mathematical vector shenanigans. What it doesn’t mean is that qubits are exponentially “more powerful” than bits. So please. Please. Please stop saying that.