Hacker News Reader

Comment by roenxi | Hacker News Reader

It exposes an interesting flaw in people's thinking where they hear percent and immediately convert it to a binary (75% -> Yes, 29% -> No). If pressed most would probably claim to understand that 75% means "every 4th time we see this it won't work" and 29% "nearly every 3rd time it will", but in my experience most people act more appropriately for the binary outcome than the probabilistic one.

Thinking and acting to the probability model is almost a superpower; I keep finding I'm ready for 1 in 10 style events that other people don't see coming. After being told authoritatively there is a 10% chance of it happening, no less. Eg, I will look at what a 1:100 year rain event in an area will cause then make long term decisions around it.

View on HN

Yes, this drives me nuts.

I'd also say that when viewing something like a "1:100 year rain event" to understand that it's just a model there, and not always the actual probability, and to try to understand what limitations might exist. Climate change is likely making some events like that more frequent for example.

I think there is an additional problem with "1 in 100 year" events: it looks so small that a lot of people don't realize that they are more likely than not to see it in their lifetimes.

It depends on framing. Most people would decline to play Russian roulette even with a 100 chamber cylinder with but a single live round.

Of course they would. What have they got to gain?

Now put a prize in there and I might well play.

> I keep finding I'm ready for 1 in 10 style events that other people don't see coming.

Surely if you’re ready for more than 10% of 1-in-10 events, your probabilities are just as wrong as other people’s; it’s just that your risk management is better.

That is why I don’t like probabilities for what are fundamentally one-off events.

I posted a comment saying "a common mistake is thinking 10% = No" and you've responded by with something that looks a lot like "you're making a mistake thinking 10% = yes".

My suggestion is you don't have enough information to claim I'm making any specific mistake.

You prepare for a 1 in 10 chance event which other people don't prepare for; i.e. is you act exactly as if you think it will certainly happen, but you tell yourself you don't think it will certainly happen?

This sounds like a distinction without meaning. Using "people mistake my preparation for me believing that it will happen, so they are dumb" is weird. You're acting as if you think it will happen, of course people think you think it will happen. Why else would you bother preparing? You're not buying 10% of the sandbags you'd need, you're buying 100% of them because nothing else would do.

"Oh but I'm smarter" but your belief is leading you to take the same actions as the person who thinks 10% == yes, so in what way is it smarter? In what way is it different?

Are you absolutely sure you are across all the flaws in my character and decision making process after my 2 short comments? You've got practically no idea at all how I approach planning and you're already telling me a bunch about myself that I never said.

> "Oh but I'm smarter" but your belief...

Case in point.

The article says Nate Silver had Trump's chances at a little worse than 1 in 3. The average person is going to sit through 17 presidential elections.

I don't have to be very clever to say that Trump winning is totally consistent with the model; and anyone who was caught by surprise has failed to understand probability. 29% is not saying no chance of victory and it isn't really that surprising. It isn't even a forecast of defeat if you think about it. It is evidence that a defeat was likely, but a modeled probability is not a forecast - it is evidence used in forecasting.

I'm really claiming that most people will hear "10% chance of disaster!" then after it didn't pan out 9 times they'll start talking about the boy who cried wolf or stopped clocks when the disaster strikes the 10th time. I have a massive advantage over them, because that is exactly the wrong attitude towards probabilities.

> so in what way is it smarter? In what way is it different?

As Alexander said repeatedly, it's cost-benefit analysis. See my other post in this thread.

If you go through life only preparing for the things you are certain will happen, you are quite likely to experience some avoidable unpleasantness.

This doesn't answer my question. If you prepare for it, then you are acting as if you think it will happen. That is, if you thought it would definitely happen you would do the same thing you are doing now, therefore the two states are indistinguishable except for the story in your head - and the main point about the story seems to be "I don't think it will happen therefore I'm better than the people who do think it will happen, even though we are both expending the same effort, doing the same preparations, to avoid the exact same problem"

Chanting "cost benefit analysis" doesn't change anything.

> If you prepare for it, then you are acting as if you think it will happen.

That isn't the case. Preparation isn't a binary state either - people can be more or less prepared for an occurrence. The estimated likelihood of an event changes how much time and effort should be spent preparing to it.

There is a big difference in how people behave if they thing there is a 0%, 0.05% or 98% chance of being mugged when they go outside for example. The 0.05% case would rationally start avoiding people late at night or not carrying valuables. The 98% person would not go outside.

> This doesn't answer my question.

Of course, if you persist in dismissing everything that makes a difference, then you will never figure out what the difference is.

You are being inconsistent in the way you presume (mostly mistakenly, I believe) various opinions and motives on the part of Scott Alexander, while attempting to reject opinions, motives and rational thought as being relevant to the behavior of people who prepared for C19.

> Chanting "cost benefit analysis" doesn't change anything.

Well, I agree that "chanting" (i.e. explaining to someone who will not listen, apparently) will not change anything (and least of all your understanding of the issues), but actually practicing cost-benefit analysis effectively does make a difference - as we can see, for example, in the difference in how C19 is progressing in South Korea and in Britain.

> You prepare for a 1 in 10 chance event which other people don't prepare for; i.e. is you act exactly as if you think it will certainly happen, but you tell yourself you don't think it will certainly happen?

"Prepare for" is not the same thing as "think it will certainly happen". It's a case of making your cost-benefit calculations appropriately. You don't do all the things that you'd do if you thought it was 100% going to happen, you do a cheap subset.

I bought 2 bottles of hand sanitizer and 2 reusable n95 facemask and some filters in mid January.

I didn't think I'd need it - I thought it'd end up like SARS or h1n1. But it cost me $30 or something and I can afford to lose that so why not.

I bought some Bitcoin fairly early. I thought it was stupid but there was a small chance I was wrong.

In both these cases if I'd been sure I'd have acted very differently!

I'm unsure about a lot of things so sometimes I try to get on both sides of a subject.

> Surely if you’re ready for more than 10% of 1-in-10 events, your probabilities are just as wrong as other people’s.

Not necessarily; as Alexander mentioned repeatedly, it depends on the cost of repeatedly preparing unnecessarily versus the cost of not being prepared that one time it happens.

Just about the whole of aviation safety, to give one extended example, is based on this principle.

> Surely if you’re ready for more than 10% of 1-in-10 events, your probabilities are just as wrong as other people’s

Disagree, it depends how much it costs to be ready, and how much it costs to be unprepared if the 1-in-10 event happens.

So if you know of 10 different events which may occur with probability 10% each, you would prepare for exactly one of them?

I think it's clear that there are some events that may be rare but which are cheaper to plan ahead for than to deal with after the fact, and vice versa (e.g. for many people self insuring against minor losses like appliance failures makes sense).

Here are some risks I'd like you to maybe prepare for:

A: 10% chance you lose 100$. Preparing costs 5$

B: 10% chance you lose 100$. Preparing costs 15$

C: 10% chance you lose 200$. Preparing costs 15$

D: 10% chance you lose 200$. Preparing costs 30$

E: 10% chance you win 100$. Ticket costs 20$

F: 10% chance you win 100$. Ticket costs 5$

You prepare for A & C. You buy tickets for F

Think of it like choosing when to fold in poker

It's useful to think that way, until you add more options:

G: 5% chance one of your parents die, preparation costs $X

H: 3% chance you die, preparation costs $Y.

What at what values of $X and $Y do you prepare?

Refusing to answer the question doesn't mean you won't have to make that decision. Better to've considered these questions ahead of time. People run into this when they haven't decided how much they're willing to pay before putting their pets down

The 5%/X 3%/Y question is also hard to answer since one also has to consider how much is it mitigating, if I survive with chronic pain or something then that loses value. Assuming full mitigation:

I'd be willing to pay ~16000$ to skip a 1/30 chance of death (valuing ~480000$ to live ~50 years, or 9600/year being-alive tax)

For my parents I'd be willing to pay ~9000$ to skip a 1/20 chance of one of their's death. (valuing ~180000$ for them to live ~30 years, or 6000/year being-alive tax)

Reply on news.ycombinator.com