[Eridrus]

At some point, what are you insuring against, if you know which machines will fail? This reminds me of the pre-existing conditions debate in healthcare - if you only insure healthy machines, then what is the point of buying insurance? Sure, catastrophy insurance is good, but I wonder if better data may reduce the size of the insurance market. Not necessarily bad for those being insured, since they could do preventative maintenance, but I wonder if the insurers are at all concerned.

[dsr_]

The insurance company wants to pay out less money, so they offer a reduction in premium cost for behaviors that they think are worth that differential.

Suppose you have a thousand doohickey machines that cost 10,000 each to replace in an emergency, of which 50% is the doohickey cost and 80% of the rest is the emergency labor cost ; a ten year lifespan, and an observed failure rate of 1% per year.

In a normal year, you need to replace 100 doohickeys at a cost of a million. Over ten years, you replace 1000 doohickeys at a cost of ten million. Your insurance company charges you 1.02 million a year whether they have to replace 900 or 1100 in that particular year. It costs you a little more on average, but it keeps you from experiencing a catastrophe.

Now the insurance company gathers data from your doohickeys that predicts with 90% reliability that a doohickey will fail within a month. If they can pay the 1000 non-emergency cost of the labor (plus the 5,000 part cost), then they go from a 10,000 outlay to a 6000 outlay. 4000 savings x 90% x 100 doohickeys needing replacement is 36000.

So the insurance company offers you a reduction from 1.02 million per year to 985,000 per year if you install the realtime doohickey monitoring system. That's a great savings for you, a good savings for the insurance company, and everybody is happy...

unless it turns out that the realtime doohickey monitors have lousy security and leak valuable personal information to anybody who guesses the password (which is password321).