How to Check Us

A practical guide for a reader who wants to verify Kronos claims rather than trust them — the sources, the data, and the questions to ask.

Do not trust us — check us

The honest position is that a reader should be able to verify our claims independently. This page is a practical guide to doing exactly that. If any step below is impossible for a claim we make, that is a bug in our transparency, and we want to hear about it.

Steps to verify a Kronos claim

• Find the claim's source: is it a paper with a DOI, or a page citing one?
• Check the verb: is the number simulated, projected, or measured? It should say
• Pull the open data and, where provided, re-run the analysis to reproduce the number
• Look for the conditions: operating point, scale, maturity, and the gate it depends on
• Ask what would falsify it — a well-made claim names its own falsifier

Questions worth asking

• Is this a design target or a hardware measurement?
• Which Q is this — scientific or engineering?
• What is the uncertainty, and which assumption dominates it?
• Is this one machine or a fleet, a peak or an average?

A claim that survives these questions is worth taking seriously. A claim that dodges them is not — even ours. We built the record so that this kind of checking is possible, because a program that cannot be checked has not earned the trust it asks for.