“UK only” is four regulatory regimes, and we were citing the wrong one

Approved Document B has no force in Scotland. It never has. A building in Glasgow is assessed under the Building (Scotland) Regulations 2004 and the Technical Handbooks, and a report citing ADB as its guidance is citing a document that does not apply.

Our check register knew this and did it anyway.

How both things were true at once

The reference library had it right. The entry reads, verbatim:

And two study types in the scoping module cited "adb" unconditionally — in a list of references that gets attached to every assessment of that type, wherever the building is.

That is the shape worth noticing. This was not ignorance of the regulations. It was a piece of knowledge sitting in one file and a decision being made in another, with nothing in between — which is how almost every silent defect in this system has looked.

Four regimes, not one country

None of this is obscure. Any fire engineer practising across the border knows it, and would spot the wrong citation immediately — which is rather the point. The reader most likely to notice is the verifier deciding whether to accept the assessment.

What a jurisdiction has to be in software

The instinct is to treat this as a text substitution: swap the guidance name by region. That fails on the thing that matters, which is what happens when we do not carry the governing document.

We do not have the Scottish Technical Handbooks in the reference library. We do not have Technical Booklet E, or the Welsh ADB, or Ireland's Technical Guidance Document B. Assessing a Scottish building on BS 7974 and BS 9999 is entirely legitimate fire engineering — but the report must not imply the Handbooks were consulted, because they were not.

So a jurisdiction is three things:

• The instrument that has legal force, which is not the guidance document.
• The reviewing authority, because "Building Control" is not what it is called everywhere, and it is the first question a customer asks.
• Declared gaps. The governing documents we do not hold, stated by name — so the report can say what was assessed on and what was not.

The check that came out of it blocks a citation used outside its own regime, warns when a jurisdiction's governing documents are missing from the library, and — the part that took longest to get right — warns when no jurisdiction has been stated at all, rather than silently assuming England.

The database decision, which is the real one

The column is nullable, and we did not backfill it.

Every project created before this existed implicitly assumed England. That assumption is almost certainly correct — the product has only ever been used here. It would have been one line to write 'england' into every historical row and have a tidy dataset with no nulls in it.

An engineer looking at an old assessment needs to know whether the jurisdiction was decided or defaulted. Provenance you did not record cannot be recovered later, only guessed — and a guessed provenance on an issued document is worse than an absent one, because it looks like a fact.

A claim we got wrong, and corrected

While mapping this out we told ourselves that BS-based markets extend well beyond the UK, and that a Dubai engineer works to BS EN 12101-6 exactly as a London one does. That is the kind of thing that sounds right and turns into a market strategy.

It is wrong. The UAE Fire and Life Safety Code of Practice is largely NFPA-derived. Qatar's QCS references both BS and NFPA depending on the system. Hong Kong and Singapore have their own codes that reference BS rather than adopting it. None is a drop-in for a BS-based register, and a document citing BS guidance in front of a Civil Defence reviewer is citing guidance that does not govern there.

Those regimes are now listed in the product and marked unsupported, with the reason attached. A jurisdiction we cannot assess is a refusal, not an absence — and a picker that silently omits Dubai looks like a product that has never heard of it, rather than one that considered it and said no.

What this is really about

Every check register encodes a regulatory model, whether or not anybody wrote it down. The dangerous version is the one that encodes it implicitly — where the limits, the citations and the acceptance criteria all quietly assume one jurisdiction, and nothing in the system can tell you which.

If you are evaluating fire modelling software and it operates in more than one country, the question worth asking is not "does it support Scotland". It is: what happens if I tell it the building is in Glasgow? If the answer is that the output is identical, the regulatory model is implicit — and it is England's.