Thread by @KellyHereid

So, I imagine @oneconcerninc is currently formulating a response to this @nytimes piece, but here are some personal thoughts, from yours truly, sitting in a space that intersects with its field fairly closely. nytimes.com/2019/08/09/us/… 1/n

So to wade in... What One Concern has built is a catastrophe model. We can quibble about how it's done it, whether it uses public data/AI/magical fairy dust/whatever, but that's what it is - a model that predicts damage from a disaster 2/n

The insurance industry uses cat models to make realistic estimates of how damaging a whole suite of natural perils will be - hurricanes, floods, wildfires, earthquakes, and so forth. Thousands of years of hypothetical events, and huge catalogs of historical "as-if" scenarios 3/n

Want to know what would happen if the 1926 Great Miami hurricane hit again today? Ask a cat model. What about a magnitude 8 earthquake on the San Andreas? Again, cat model. 4/n

However, these models are not traditionally designed to perfectly match an event as it happens. Any big cat model vendor would surely agree with me in saying that event response, meaning figuring out how big a disaster will be in real time, is a difficult and frustrating job. 5/n

So this is the first innovation from One Concern - it wants to model disasters in real time, rather than picking a nearest approximation from a pre-compiled catalog. This is genuinely cool, if it works. 6/n

The second innovation is one of audience. Rather than selling primarily to insurance companies, it made its pitch to emergency managers of various stripes. This is where things start to get woolly. 7/n

A cat model is only as good as its underlying data. When you're an insurance company, an approximation of a building's location or characteristics is going to get you most of the way there on figuring out its level of damage in an earthquake, for example. 8/n

Any one individual building may be a little off in a given simulation, but on average across an entire portfolio of policies, a cat model will get you pretty close, making it an informative tool for pricing policies. 9/n

But a few problems emerge when putting this to use in emergency management. One, the margin for error is very low. If an insurance company's view of a cat model is wrong, it loses money, but if an emergency manager's view is wrong, people could die. 10/n

The second big problem is that your client base needs the chops to vet these things. Maybe One Concern has a model that performs better than the gut of your local emergency manager - but this hasn't been put to the test yet. 11/n

Insurance companies have whole teams of nerds whose whole job is to make sure that cat models are reasonable and accurately reflect their risk (hi). But this brand new client base has functionally never had to try to do this before. 12/n

Part of this is communication - no insurance company would expect perfect results at a building level out of a model. A cat model is a tool in a toolbox, and this one may be a good one for emergency managers - but they need to learn how to interpret it. 13/n

But the other part is that no one likes a black box. Insurance cat models get vetted to the teeth by industry partners, regulatory bodies, external peer review, you name it. One Concern will need to do this as well to gain trust. 14/n

One Concern has an opportunity to bring real innovation to an exciting space, but it needs to do it in a way that fosters trust and meets the needs of its intended audience. "Saving people better" is a noble goal, but it first has to prove it's not "saving people worse" fin/15

@oneconcerninc @nytimes I'd also be curious to hear how #EMGTwitter views what One Concern is trying to do, and the NYT piece as well. Thoughts, @SamLMontano?