A platform like this is only useful if it sends you to people who can measure. NPL exists for a reason. So do tank tests, hangar surveys and the nurse who will tell you the lid is on the wrong side.
We will keep publishing work packages that ask for a number. Metaphors are welcome afterwards, in the pub.
A claim without an uncertainty is a hope with units attached.
The pattern repeats across every domain we list. A capability is demonstrated, genuinely and honestly, under conditions that are not the conditions of use. The demonstration is then described in language that quietly drops the conditions. Within eighteen months the language is in a strategy document, and the original researcher is being asked why the technology has not been adopted.
A claim without an uncertainty is a hope with units attached. This is not pedantry — it is the difference between a number a customer can rely on and a number that will be renegotiated during a dispute. The organisations that understand this best are the ones who have been through such a dispute.
What is striking is how cheap the correction usually is. A traceable measurement campaign costs a fraction of a funding round. An independent validation on data a model has not seen costs less than a month of engineering time. A tank test costs a fraction of a marine recovery operation. In almost every case, finding out early is the economical option, and it is chosen surprisingly rarely.
Our modest contribution is to keep asking, in every package we help scope, what will be measured, against what, and what the report will be permitted to say. It is not a glamorous question. It is the one that separates the projects that finish from the projects that get quietly reprofiled.