> Prerequisite: Episode 004 (interference) — this piece is essentially an application of it. > Companion piece: "Quantum Computers Are Not Faster Computers" covers the same topic from the perspective of the machine's nature; this article targets the most widespread reasoning flaw within it. > Next episode: 006, "The Toppling Top."
The Popular Misconception of "Exponential Parallelism": More Superposition ≠ Faster Computing
Almost every viral article about quantum computing contains a passage like this: "A qubit is 0 and 1 at the same time; two qubits are simultaneously 00, 01, 10, 11 — four states! n qubits means 2ⁿ states existing at once! 300 qubits embody more states than there are atoms in the universe — so a quantum computer can try all answers simultaneously, and every hard problem falls."
The previous episode planted the seed of criticism (superposition without interference is just noise); this one tears down the phrase "every hard problem falls." This is not a minor slip of the tongue — it is the most popular and most misleading way of understanding quantum computation.
Having Many States Is Not the Same as "Trying All Answers"
The sentence "2ⁿ states exist simultaneously" is itself correct — the superposition of the 300th qubit does contain an astronomical number of arrows (in the wording of episode 004). What's wrong is the jump from there to "trying all answers at once." Break it into three flaws:
Flaw one: capacity you can store is not capacity you can use. The 2ⁿ candidate answers to a problem can indeed be spread across 2ⁿ arrows. But then what? If every operation treats all arrows identically, measurement hands you uniform noise — every answer roughly equally likely, meaning you computed nothing. The exponential capacity pays off only if you can execute exponentially fine interference choreography on it, and that ability is precisely the hard part — it cannot be waved away in one sentence.
Flaw two: exponentials go in, only linear comes out. This is the harder gate: n qubits, measured once, output only n bits — you saw the principle in episode 002 (one collapse yields one definite outcome). The exponentially many pieces of information sit in the superposition, but the exit is a narrow n-bit-wide door. Any design of the form "try all answers at once, then pick the right one" must first compress the 2ⁿ pieces of information into an answer of at most n bits — and what does the compressing? Interference again: arrows of wrong answers cancel each other. We come full circle to episode 004's line: all the power lies in interference.
Flaw three: problems that can actually exploit exponential space are few. Thirty years on, the families of general quantum algorithms rigorously proven to have exponential advantage over classical computation remain just a few: factoring (episode 038), certain simulation tasks (episode 042)… Even Grover's search algorithm only gets a square-root speedup — a hundred million checks become ten thousand. A quantitative gain, not "instant." The blueprint of exponential space has always lain there; designs that turn it into genuine speedups are rare, each one a blade honed over decades by physics and computer science.
A Practical Litmus Test
From now on, whenever you see "our quantum approach is faster because it exploits exponential parallelism," measure it with this thermometer:
> What is its arrow-cancellation structure? Which arrows are reinforced, and which piles of arrows are cancelled out?
If it can't answer, it is most likely marketing copy treating "superposition" as a get-out-of-jail-free card. In real algorithm papers, this part is called "interference pattern analysis" — the hardest and most essential section.
Clear up a related misconception along the way: quantum computers will not replace your laptop. They are not faster general-purpose computers — quite the opposite: for many everyday tasks (writing documents, watching videos) they have no advantage even in principle. They are specialized amplifiers for particular mathematical structures: only when the problem hides a period or search structure that interference can choreograph do they take the stage. Which problems hide which structures is the main thread of the fourth and fifth batches of this series.
Dismantling the Analogy
This article uses no new analogy — because its job is to demolish one. "Trying all answers simultaneously" is essentially an analogy equating a quantum state with "countless tiny parallel computers." Its fatal flaw: the little computers could each compute a result and report it, compare notes — while the arrows in a superposition cannot. Any "report" is a measurement, a collapse (episode 002). The exponential resource is trapped inside a whole that cannot report piecemeal; the only way out is interference-based compression. That is the fundamental crack between the analogy and reality.
What to Take Away
1. "2ⁿ states exist simultaneously" is true; "therefore it tries all answers at once" is false — capacity is not computing power; interference choreography is. 2. The exit bottleneck: n qubits measured once yield n bits; exponential information must be interference-compressed into a single answer before it can be extracted. 3. Proven exponential advantages remain rare (factoring, simulation, etc.); even Grover gives only a square-root speedup. A quantum computer is a specialized amplifier for particular mathematical structures, not a faster general-purpose computer.
Hook for the next stop: the arrows are so precise, the environment so meddlesome — how long can a quantum state actually "live"? Why does a spinning top eventually fall? Episode 006: "The Toppling Top: Coherence, Decoherence, and the Fragility of Quantum States."