Size & scale

The false common scale

Same word on the axis. Not the same thing being counted.

a.k.a. apples and oranges · incomparable units · the shared axis that isn’t · unit drift · mixed denominations

A bar chart makes exactly one promise: that its lengths are commensurable, because every one of them is so many of the same thing. Put rows on it whose unit quietly changes — counts of objects that do different jobs, figures collected under different rules, quantities that happen to share a word — and the promise is void while the picture looks unchanged. Nothing is exaggerated and no axis is bent; the ranking is simply about a quantity that does not exist.

How to spot it

Read the unit under the axis, then ask whether that word means the same thing in every row.
A word that a whole industry uses for several different objects — “users”, “cases”, “tons”, “qubits”.
Rows sourced from separate publishers or spec sheets, with no line saying how the definitions were reconciled.
One row an order of magnitude clear of the rest. A leader that far out is often a leader in a different unit.
Ask what a single unit of this actually does. If the answer differs by row, the lengths cannot be compared.

The fix

Split the chart along the thing that changed. One panel per unit, each with its own axis, is not a failure to combine them — it is the finding, stated plainly. Within a panel the lengths mean something again, and the panels sitting one above another show the reader exactly where the comparison stops. If one number really is wanted across the rows, convert them into a common unit and show the conversion; if no conversion exists, say so, because “these cannot be added up” is usually the most useful sentence on the page. And where the interesting quantity is quality rather than count — accuracy, duration, yield — put that on its own axis too, since no amount of counting will produce it.

In the gallery

Dark navy briefing slide, masthead “Quantum Desk · State of the machines · Flagship processors”, headlined “The qubit race has a clear leader”. A subhead reads: five companies, five flagship machines, one number each — the qubit count in the manufacturer’s own announcement; nothing estimated, nothing modelled; one zero-based linear axis. Under the label “Qubits · flagship processor · as published by the manufacturer”, five horizontal bars run from a zero line on an axis marked 0, 1,000, 2,000, 3,000, 4,000 and 5,000, with the unit “qubits” printed beneath. D-Wave’s Advantage2 takes a long gold bar labelled 4,400+; Atom Computing’s 1,225-site array a blue bar at 1,180; IBM’s Condor a blue bar at 1,121; Google’s Willow a sliver at 105; Quantinuum’s Helios a sliver at 98. Three ticked boxes run along the bottom: “Published specifications — every count is the figure in the maker’s own launch announcement”, “One unit — qubits, on a single zero-based linear axis, no log scale, no truncation”, and “Flagship machines — each company’s largest processor announced to date”. Down the right, a card carries 45× in large white type over the words “more qubits on the leading machine than on the smallest in the field”. A second card, “The thousand-qubit club”, lists D-Wave Advantage2 at 4,400+, Atom Computing at 1,180 and IBM Condor at 1,121. A third, “Still in three figures”, lists Google Willow at 105 and Quantinuum Helios at 98. A gold-bordered note at the bottom reads: “You don’t need any quantum mechanics to read this chart. One machine has 4,400 qubits and one has 98.” Beneath it, in small grey type, sits the disclosure: “Qubit counts are not comparable across architectures.” A footnote gives the counts as published — D-Wave Advantage2 (2025), Atom Computing 1,225-site array (2023), IBM Condor (2023), Google Willow (2024), Quantinuum Helios (2025) — and calls the slide a Quantum Desk demonstration drawn by us.

The qubit leaderboard that changes unit between bars

The false common scale