What can we deduce about “how they do this”

Google chose a specially‑constructed benchmark problem: random quantum circuits, high depth, 53 qubits.

Google measured performance: quantum machine runtime (~200s) vs estimated classical runtime (~10,000 years).

IBM’s rebuttal: they claim a better classical simulation (with optimized algorithm and storage) reduces the classical time to ~2.5days.

IBM emphasises that this task is highly specific, not broadly useful (yet).

So “proof” is given in the sense of: quantum processor executed something that classical simulation claims it would struggle with — but IBM argues that claim is weaker than portrayed.
---

What they didn’t show (based on IBM’s argument)

They did not show a general‑purpose quantum algorithm that outperforms all classical computers in a broad domain. IBM says threshold “has not been met.”

The classical runtime estimate by Google is disputed and depends on how you allocate classical resources (memory, storage, algorithm). IBM says the 10,000‑year number is overstated.

They did not “short a magazine” or run public broad‑market tests (i.e., mass‑visible industrial use) in that cut — the experiment remained within a lab/benchmark context.

OUT OF THE WAY YOU PUSSIES AND LET ME JUST REACH FOR THE TOYOBI;