Two fires, two coffee cups, two badly named descendants#

Here’s what makes this satisfying rather than merely annoying: the two oldest branches of modern insurance each left behind exactly one terribly named child, and both children are still sold today.

1
1666  Great Fire of London  →  fire insurance    →  dwelling fire
2
1686  Lloyd's coffee house  →  marine insurance  →  inland marine

Twenty years apart, in the same city. Let’s take them in order.

1666: the fire that started fire insurance#

The Great Fire of London burned through most of the city in September 1666. What followed was the first serious market for insuring buildings: around 1680 — sources differ on whether it was 1680 or 1681 — an economist and property speculator named Nicholas Barbon founded the first fire insurance company, tied to a parallel business making mortgage loans to people rebuilding their houses.

(Barbon is worth one aside. He was the son of a Puritan preacher, and his baptismal name is recorded as something along the lines of “If-Jesus-Christ-had-not-died-for-thee-thou-hadst-been-damned.” Sources differ slightly on the exact wording, which is its own kind of appropriate. The man who invented fire insurance was named after eternal damnation.)

Early residential insurance was exactly one peril: your house burns, we pay. One contract, one cause of loss. Calling it a fire policy was simply accurate.

Then it grew, the way software grows. Somebody wanted wind and hail, so a bolt-on called Extended Coverage appeared. Then vandalism. Then the contents of the house. Then liability, when it became clear that owning property means occasionally being sued by people who visit it.

Nobody renamed anything. Each addition was an endorsement — a patch — stapled onto a contract whose title still said fire.

The refactor finally came in September 1950, when the Insurance Company of North America — founded in 1794, the oldest and largest of its kind in the country — put fire and theft and liability and loss-of-use into a single contract. That got a new name: the homeowners policy, which is what almost every American owner-occupier buys today. It cost about 20% less than buying the equivalent separate policies, and it sold immediately.

And here’s the part I had wrong until I checked. I’d assumed nobody bundled earlier because nobody had thought of it. It was illegal. Until US insurance law was reformed in the late 1940s with what were called multiple line statutes, a single company generally could not write both property and liability. A fire insurer was, as a matter of statute, a fire insurer.

Which reframes the whole naming problem. These labels weren’t lazy. On the day each was chosen, the name accurately described the entire business the company was legally permitted to conduct. The scope was the law.

But the old lineage didn’t die. It survived to serve everything the homeowners form can’t cover: rental properties, vacant houses, seasonal homes. And it kept the old label. Today those forms are DP-1, DP-2 and DP-3 — the dwelling fire forms, and DP-3 is written on an open perils basis, meaning it covers everything not specifically excluded.

So the flagship product in a line called dwelling fire treats fire as merely one of countless causes it happens not to exclude.

This has gone about as well as you’d expect. My favourite artefact of the whole investigation is that the name is bad enough that a state regulator has to disclaim it in writing. From the North Carolina Department of Insurance’s consumer page on dwelling policies:

Although the name states fire, these policies cover more than just fire peril.

That is a government agency formally apologising for an industry’s variable naming. If you have ever wanted evidence that bad names have real-world cost, it is a public-sector employee being paid to write that sentence.

1686: the coffee house that started everything else#

Twenty years after the fire, and a short walk away, a man named Edward Lloyd opened a coffee house on Tower Street. Coffee had only arrived in England in 1652 and had become the drink of the new professional class; London had more than eighty coffee houses, and each had drifted into serving a particular trade.

Lloyd’s specialised in shipping news. Ship owners, captains back from voyages, and merchants gathered there because it was the fastest place in the city to learn which vessels had made it home. It was, functionally, a Bloomberg terminal made of wooden tables and gossip.

And where information concentrates, risk gets priced. Lloyd’s began renting out its tables to men who would sell a ship owner a promise: if your vessel doesn’t come back, we pay. The earliest surviving reference to the coffee house is 1686; the first arranged contracts date to 1688.

Two things about that origin are worth sitting with.

The first is that it was frankly indistinguishable from gambling. The historical accounts describe patrons betting on which ships would return — including captains betting against rival ships. Insurance and wagering are the same transaction with different paperwork, and it took a long time and a lot of law to separate them.

The second, less comfortable: Lloyd’s coffee house was also frequented by merchants in the slave trade, and the market insured slave ships. That is part of the same origin story, and Lloyd’s own historical timeline includes it.

Now here’s the part that makes this an article about naming rather than about coffee. Lloyd’s is a global market today, and it still speaks in the furniture of a 1680s coffee shop. An underwriting syndicate works at a box — because Edward Lloyd rented out boxes, meaning tables. The policy document is a slip. The individual members who bear the risk are called Names.

The names aren’t fossils in that case. The furniture is a fossil, and the vocabulary is the impression it left.

When the two branches collided#

So: by the 19th century you have two established guilds — and, as we just saw, the separation between them was enforced by statute, not just by custom. Two legally distinct trades with two different habits.

Fire insurers wrote named perils — a contract listing exactly which disasters it covered. Conservative, enumerable, easy to price.

Marine insurers wrote all risk — everything covered unless excluded — because that’s what makes sense for cargo crossing an ocean, where you genuinely cannot enumerate what might go wrong.

Then the Industrial Revolution produced a new category of property that fit neither habit: telegraph lines, railroad equipment, bridges, radio towers. Things on land, but strung across the landscape, hard to inspect, hard to categorize.

Fire insurers looked at that and offered narrow, enumerated coverage, because that was their instinct. Marine underwriters looked at it and offered all-risk, and took the business.

That’s what inland marine means. Not “marine, but inland.” It means “the marine underwriters’ book of business, on dry land.” The word records which guild won the account.

And once you know that, the modern line stops looking incoherent. Inland marine today covers jewellery, fine art, film equipment, contractors’ tools, mobile medical equipment — and also bridges and radio towers, which are neither expensive-and-portable nor anything like cargo. The formal definition covers property in transit, property in someone else’s custody, and anything that is an instrumentality of transportation or communication.

The common thread was never “things that move.” It was “accounts the marine desk took.”

Two more artefacts of that history, both delightful:

• Ocean marine is called wet marine inside the industry. The real pair is wet and inland — which is honest, and far clearer than “ocean and inland.”
• Inland marine policies are also called floaters, because the property floats around. Which means the scheduled-jewellery endorsement on an American homeowners policy is formally a floater — a term from cargo insurance, applied to a wedding ring, on a policy descended from the Great Fire of London.

The other direction: when the word moves and the term stays#

Everything so far is a product outgrowing its name. There’s a second, sneakier version: the term never moved, but ordinary English did, so a modern reader imports the wrong meaning entirely.

casualty did not originally mean a dead person. In the early 1400s, casuelte meant “chance, accident” — built from casual plus a suffix, on the model of royalty and penalty, from Latin casus, “chance, event,” from cadere, “to fall.” The military sense of “losses from a troop” arrives only in the late 1400s, and “an individual killed or wounded in battle” dates to 1844 — four centuries after the word itself.

So casualty insurance is chance-event insurance. It always was. Modern readers push an 1844 meaning back onto a much older term and end up picturing morgues.

Which also means casual clothes and a hospital’s casualty ward are the same word. Both are just “by chance.”

And this explains the most-used acronym in the industry. P&C — property and casualty — means “property, and other things that happen by chance.” The C is a junk drawer. It’s been a junk drawer so long that in 1956 a textbook author was already writing in his own preface that some insurance men expected the category to shrink to liability and disability lines only. Seventy years later it’s still there, still a drawer.

indemnity names a state, not a payment. In- plus damnum: not-damage. Un-damaged-ness. The sense “security against damage” is mid-1400s; “compensation for loss” only appears around 1793.

That is the entire legal principle of indemnity, sitting inside the word: the goal is not to give you money, it’s to return you to undamaged — and specifically not better than undamaged. You cannot profit from a loss. The constraint is etymological.

Bonus: damnum is also the root of damn. Indemnity and damnation are relatives. Given that fire insurance was invented by a man named after damnation, the family tree here is tighter than anyone intended.

subrogation is just surrogate. Latin subrogare, to put one party in the place of another. When your insurer pays your claim and then goes after whoever actually caused it, it is acting as your surrogate. The word isn’t misleading, only opaque — and it stops being opaque the second you notice the cousin.

umbrella is precise, and almost everyone reads it wrong. Most people assume an umbrella policy stacks extra limits on top of your existing coverage. That’s excess insurance, which is follow-form: it mirrors the underlying policy’s terms exactly and only raises the ceiling.

An umbrella is not follow-form. It can define risks more broadly than the policies beneath it, drop exclusions they carry, and act as primary coverage from the first dollar for losses they don’t cover at all — the industry calls that last move “dropping down.”

So the metaphor is right and the intuition is wrong. An umbrella is wider than what’s underneath it, not merely taller. Excess is taller. Umbrella is wider. One of the two names in this entire article turns out to have been chosen carefully, and it’s the one people misread.

A third kind: names that point at the problem, not the protection#

The two mechanisms above are both about time — a name that stopped being true. There’s a third that was never quite true to begin with: coverages named after the thing going wrong rather than what you get.

Uninsured Motorist is the one people misread most, and the misreading is close to universal. It sounds like insurance for drivers who don’t have insurance. It is the exact opposite: it covers you, for when the driver who hits you has none — or drove off and was never identified. Its sibling Underinsured Motorist handles the case where the other driver does have insurance and it simply isn’t enough, which happens constantly, because state minimums can be startlingly low. Pennsylvania requires $5,000 of property damage liability. Florida requires no bodily injury liability at all.

Ordinance or Law is my favourite of these, because once you see what it does, the name is almost perverse.

Replacement cost promises to give you back what you lost. But building codes change. A house built in 1965 burns down; the insurer owes you a 1965 house; the law will not let you build a 1965 house. You have to build to current code — wiring, seismic, insulation, sprinklers, accessibility. And basic replacement cost won’t pay that difference, because it isn’t a restoration. It’s an upgrade.

Three costs live in that gap: demolishing the parts of the building that survived the fire but no longer comply, the lost value of those surviving parts, and the higher cost of building to the new standard.

So this coverage exists because two rulebooks disagree with each other. Not because of a peril. Nothing burned twice. The insurance contract says restore it and the municipal code says you may not — and Ordinance or Law is the money in between.

Now read the name again. Ordinance or Law. It names the obstacle. A homeowner scanning a policy for “will you pay to bring my house up to code” would never stop on it.

It has one more trick, which confused me for a while: it appears as both a coverage and an endorsement, on the same policy. That isn’t a duplicate. The ISO forms build in a base amount — 10% of your dwelling limit — and HO 04 77 is a separate endorsement that raises it, to 25% or 50%. One row is the included 10%; the other is the top-up you bought. Two genuinely different things wearing one name.

Loss Assessment pulls the same trick and has the same problem. It names the invoice your condo association sends after their own insurance falls short on a common-area loss. What you’re buying is protection from that invoice. What the name describes is the invoice.

Errors and Omissions names the mistake you might make. Not the lawyer who will defend you for making it.

There’s a pattern worth naming here. Property coverages are named after what you own — dwelling, other structures, personal property, loss of use. This family is named after what goes wrong. Two naming conventions living in the same document, and nothing marks the switch.

A fourth kind, added later: names that point the wrong way#

The three mechanisms above are all failures of aim — a name that went stale, a word that drifted, a label pointed at the problem instead of the protection. Since writing this I’ve run into a fourth and worse category: names that actively send you in the opposite direction.

Comprehensive, on an auto policy. It sounds like the coverage that covers everything. It is the narrow one. ISO’s own name for it is Other Than Collision — damage to your car from things that aren’t a crash — and when you look at how state insurance departments enumerate it, it’s a finite list of named perils: fire, theft, hail, flood, vandalism, riot, glass breakage, contact with a bird or animal, falling objects. A list with an end. Hit a deer, covered; engine dies of old age, not covered.

So in the vocabulary of the property article, the coverage marketed as comprehensive is a named-perils coverage — the restrictive kind, the thing that separates an HO-2 from an HO-3. Other Than Collision never misled anybody. It just didn’t sell as well.

No-fault, on a state’s whole auto system. This may be the worst name in the business, because the plain reading is simply false. No-fault does not mean nobody was at fault. Police still cite, adjusters still assign responsibility, it still lands on the at-fault driver’s record, their premium still goes up. What “no” negates is far narrower: when paying your medical bills, fault is not consulted. It is a name about a process that reads like a name about a fact — and the honest version would have been “no-lawsuit,” since what you actually trade away is your right to sue. The mechanism, if you want it.

And two more fossils I didn’t have room for the first time.

Tort — the word underneath the whole at-fault system — is Latin tortus, “twisted.” Same root as torque, torsion, torture, contort, and tortellini, pasta twisted into a little coil. The metaphor is genuinely good: the world was straight, someone bent it, the law’s job is to bend it back. A thirteenth-century Norman French word still doing daily work in American car insurance, having never once entered ordinary English.

And Kentucky still calls its no-fault benefits Basic Reparation Benefits, a phrase lifted from a 1970s model statute — the Uniform Motor Vehicle Accident Reparations Act. Every other state re-worded it. Kentucky is still speaking draft-statute English from fifty years ago.

One counter-example, in fairness. British Columbia renamed its no-fault regime Enhanced Care — a name describing what you receive rather than negating an abstraction. The only case I’ve found of this industry noticing a name was bad and fixing it.

What this is actually about#

Three mechanisms, one outcome:

Either way, the name is a fossil — of the product’s origin, of which desk sold it, of what a word meant in 1400, or of the problem someone was staring at when they wrote the clause.

Which gives the one practical rule I’d hand anyone starting out in this domain, and the reason I went down this hole in the first place: never infer coverage scope from a line name. Not as a caution. As a hard rule. dwelling fire will tell you it’s about fire. inland marine will tell you it’s about water. casualty will make you think about ambulances. Every one of them is wrong, and the document is right there.

And if you build software in a domain like this, there’s a second lesson underneath the first. Every one of these names was accurate on the day it was chosen — several of them were accurate by law. Not one was wrong at the time. They became wrong because the thing kept changing and the label didn’t, and each individual decision not to rename it was completely reasonable — too expensive, too disruptive, everyone already knows what it means.

Three hundred and sixty years of completely reasonable decisions, and now a state regulator has to publish a sentence explaining that the fire policy isn’t only about fire.

I think about that every time I decide it’s not worth renaming something.

Credit where it’s due#

The thread I pulled on started with a laugh and a throwaway line from Brigitte Viola, an Industry Solutions Consultant on the insurance side of our team, who told me inland marine was the worst name in insurance and then explained why. Most of what I understand about this domain arrived the same way — as a real example rather than a definition.

She writes about this work from the insurance side of it: AI for Insurance Renewal Review Automation.

If the structure of these policies is what you’re after rather than their names, that’s the companion piece: The Onion and the Ladder: How U.S. Property Insurance Is Actually Organized.