Ordinary computers and their limits
The computers and phones we use every day store information as bits, each of which is either 0 or 1. All the text, photos and video on your devices are long strings of these bits, and processors manipulate them with simple logic operations at enormous speed. For nearly all everyday tasks this works superbly.
Some problems, though, grow so quickly in difficulty that even the fastest ordinary computers cannot solve them in a reasonable time. Simulating the behavior of certain molecules is one example. Quantum computing is an attempt to build machines that handle some of these problems differently, by using the rules of quantum physics rather than ordinary on-or-off logic.
What a qubit is
The basic unit is the quantum bit, or qubit. Like a bit, a qubit can be measured as 0 or 1. Unlike a bit, before measurement it can be in a combination of both states, called a superposition. It is described by numbers that determine the probability of getting 0 or 1 when you measure it.
A common misunderstanding is that a quantum computer “tries every answer at once” and just picks the right one. That is too simple. Measuring a qubit gives only one definite result, and a superposition by itself does not make a computation faster.
Qubits can be built in several ways, including tiny superconducting circuits, trapped charged atoms and particles of light. These approaches are still being developed, and there is no settled winner.
Three key ideas
Superposition. A qubit holds a blend of possibilities rather than a single fixed value. With several qubits, the number of combined possibilities grows rapidly.
Entanglement. Qubits can be linked so that their measurement results are correlated in ways that have no ordinary counterpart. Entanglement is a resource that quantum algorithms use.
Interference. The probabilities in a quantum system behave like waves, which can reinforce or cancel each other. A quantum algorithm is designed so that paths leading to wrong answers tend to cancel out while those leading to the right answer reinforce. This clever steering of probabilities, rather than brute-force trying everything, is where potential speedups come from.
The underlying science is part of the wider field of quantum physics, which describes the behavior of matter at very small scales.
How it differs from ordinary computing
- Different building blocks. Qubits instead of bits, and quantum gates instead of classical logic gates.
- Probabilistic results. Answers come from measurements, so a computation is often repeated many times and the results analyzed.
- Fragility. Qubits are extremely sensitive to heat, vibration and stray signals. Interference from the environment, called decoherence, destroys the quantum state. Many systems must be kept at extremely cold temperatures or carefully isolated.
- Error correction. Because errors are common, much research focuses on combining many physical qubits into more reliable “logical” qubits. This requires large overheads, and building machines big and reliable enough remains a major engineering challenge.
A quantum computer is therefore not a faster version of your laptop. It is a specialized kind of machine that may beat ordinary computers on particular problems, and it is unlikely to replace them for everyday tasks like browsing, word processing or streaming.
What they might be suited for
Simulating nature. Molecules and materials are governed by quantum behavior, so quantum machines may eventually model them more naturally than ordinary computers can. This could matter for chemistry, drug research and materials science, although practical benefits are still being explored.
Certain math problems. In 1994, mathematician Peter Shor described a quantum algorithm for factoring large numbers, which is believed to be far faster than known methods on ordinary computers. Many current public-key encryption systems rely on factoring being hard, which is why researchers are developing new encryption methods designed to resist quantum attack. A sufficiently large, reliable quantum computer does not exist today, but the long lead time for changing security systems is the reason the topic receives attention.
Search and optimization. Quantum algorithms may give modest speedups for some search problems, and researchers are studying whether they can help with optimization. The size of any real-world advantage is an open question.
Machine learning. Whether quantum methods can meaningfully help with tasks in machine learning is an active but unsettled area of research.
Separating hype from reality
Claims about quantum computing are often exaggerated. A few points help keep a balanced view.
- Existing devices are limited in size and make frequent errors. Demonstrations that outperform ordinary computers have so far been on narrow, specially chosen tasks.
- Timelines for practical, large-scale machines are uncertain and vary among experts.
- Some problems will not benefit at all.
- Ordinary computers continue to improve, which keeps moving the target.
Learning more
You do not need advanced physics to build intuition. Start with the idea of probabilities that can add or cancel, learn what a qubit is, then look at simple examples of quantum gates. Many universities and companies publish beginner material, and some offer free simulators. As with any fast-moving field, favor sources that explain what has actually been demonstrated and are open about what has not.