A coffee house, and a question nobody could answer.

In 1760 a group of underwriters drinking at Edward Lloyd's coffee house in London had a problem that would sound familiar to any CISO today. They were being asked to insure ships they would never see, on the word of the men who owned them. The owner had every reason to describe his hull generously. The underwriter had no way to check.

Their answer was not a better contract or a cleverer premium. It was an institution: a society that would inspect ships, publish what it found in a book anyone could buy, and — the part that mattered most — own no ships itself.

That last constraint is what made the whole thing work. A surveyor with no stake in the cargo can afford to write down what he actually sees.

The same problem, wearing new clothes

Two hundred and sixty years later we are handing credentials, spending authority, and customer contact to software that acts without us watching. Ask an organisation what agents it runs, what those agents can reach, and who authorised it, and the honest answer is usually a shrug and a spreadsheet that is out of date.

Every casualty report we have written turns on the same thing. Not that the incident could have been prevented — mostly it could not — but that afterwards, nobody could answer what had happened, to which actors, under whose authority. That question consumed weeks. The technical failure was understood in hours.

The register was never valuable because it predicted wrecks. It was valuable because after one, everyone agreed on what the ship had been.

Why copy an eighteenth-century institution

Because it is the only one that solved this and kept solving it for two and a half centuries, through iron hulls, steam, steel, containerisation, and satellite navigation. The technology it classified changed beyond recognition. The institutional design did not need to.

So we took the parts that carry: a published rulebook that binds us as well as our members, a compact notation a stranger can read in four seconds, survey intervals with real consequences, conditions of class with deadlines, and casualty returns written like accident reports instead of marketing.

And we took the constraint that makes it honest. Nobody can type their own class — including us. Our own agents sit on our own register at A2, self-declared, because no third party has confirmed who answers for them. We could not award ourselves the better figure if we wanted to.

Where we actually are

1 fleet on the register, and it is ours. 148 clauses published, some already binding on us, with dates on the rest. 6 casualty reports. A relay that can no longer grade its own evidence, because we removed that after deciding a collector running inside the operator's trust domain has no business awarding itself a verdict.

That is a small beginning and we would rather describe it accurately than impressively. Lloyd's first register covered a few hundred hulls its underwriters happened to know about. The value was never the size of the book.