Hacker News Reader

Comment by whatever1 | Hacker News Reader

I don't think that probability theory provides a useful framework for decision making.

Let's say I, a dictator, interrogate the casino manager on the suspicion that the dice are rigged. I flip 100 times I get all 6's. Do I kill the guy or not? Probability cannot help. I need to measure the dice weights myself.

As an insurance should I insure a house for $1000? Again probability cannot help when an earthquake demolishes the entire state and I have to write a check for every damn house I Insure. I actually need to keep in the bank amount equal to the price of each house

I am now inclined to believe that probability is a nerdy way of lying to our educated selves.

View on HN

> Again probability cannot help when an earthquake demolishes the entire state and I have to write a check for every damn house I Insure. I actually need to keep in the bank amount equal to the price of each house

I’m not sure what you mean. Insurance companies don’t hold liquid capital sufficient to cover 100% of their active policies paying out simultaneously.

See also: Reinsurance

> Reinsurance is insurance that an insurance company purchases from another insurance company to insulate itself (at least in part) from the risk of a major claims event.

https://en.wikipedia.org/wiki/Reinsurance

> As an insurance should I insure a house for $1000? Again probability cannot help when an earthquake demolishes the entire state and I have to write a check for every damn house I Insure. I actually need to keep in the bank amount equal to the price of each house

Not really, no. Actuarial tables can be quite amazing when they're accurately applied. All you need to do is structure the policy so that it is unlikely to pay out in such a situation, or has a limited payout. That's easy to do: you just make a list of exclusions such as "Acts of God" or war, or terrorism, etc. So basically, any instance of mass destruction will fail the coverage and you pay out $0.

Oh no, under my rule insurance companies don't get to cherry pick what they will cover. We have banned belief in acts of God.

Chances are there might not be many viable insurance companies in that scenario.

The probability of you getting 100 6's in a row on a fair die is dozens of orders of magnitude lower than your guy testing the die either getting it wrong, or lying to you for any number of reasons.

As for your second example, the problem is your own naive assumption of independence of all events. It is an extremely popular naive assumption, one I have ranted about how the school system accidentally teaches this more than once, but nevertheless is not a problem with the concept.

I think you have correctly identified that your understanding of probability is wafer-thin. That's actually progress, a good thing, and not an insult. But you've misattributed the problems to probability rather than your understanding.

Find me one person who you think he has good understanding of statistics and I will find you 10 that they think he is a complete idiot.

Our community is great at insulting each other :)

Btw for the first example, it is a deterministic one. You can count molecule by molecule and save the casino owner.

Reply on news.ycombinator.com