News for hackers. Stuff that matters.
Re:The Logic of Risk Taking by ianamartin on Wednesday October 04, 2017 @08:03PM (#15405853)
My god, this guy is so irritating. As usual, he’s not completely wrong, but he is entirely useless. It’s tautology wrapped in a bunch of garbled language and name-dropping.

Lesson #1: you don’t know where you fall in a statistical distribution.

No shit. Really? There is a 1% chance of failure in a game. If I play the game an infinite number of times, I will fail to win 1% of the time.

The alleged twist: if the nature of the game is that a single failure results in you no longer being able to participate in the game, you cannot play the game an infinite number of times.

Color my mind blown. I’m absolutely shocked that Murray Gell-Mann was able to grasp this concept so quickly. A friend of mine has a PhD in Statistics, and I remember once telling him that the probability of a fair coin tossed an infinite number of times and landing on heads approaches 50%. He grasped that concept immediately. I was impressed. We started a company to explain this.

I’m kidding. That didn’t happen because it would be stupid in every possible way.

Lesson #2: Cost benefit analysis is impossible if you don’t know where you are in the distribution.

Somehow, not knowing where you are in the range of possible outcomes means that you can’t know the range of possible outcomes.

The alleged twist: even when you know the range of possible outcomes, you apparently don’t.

I don’t even know where to start with this. I can’t even make fun of it properly.

When you’re gambling, the range of possible outcomes is well-defined for most people. You lose all your money, you break even, or you win. This model is incomplete because it doesn’t take into account human pathology. Some people cheat. Some people take out loans from shady people and lose everything and get a leg broken. Some people commit suicide because they lost a few hundred dollars.

Guess what, if you include human behavior in the model of Russian roulette, it’s also a broken model. People will cheat. People will chicken out. People will want to die and pull the trigger twice. That does not make a cost/benefit analysis “undefinable,” it just makes the range of possible outcomes larger than most people are comfortable with. That’s why most people choose to not play the game.

What Taleb does here (again) is conflate a legitimate criticism of frequentist statistical models with a critique of incomplete decision models.

But the critique of frequentist models is entirely sophomoric. It boils down to, “You can’t do something an infinite number of times. Therefore, everything we know is wrong.”

Fuck off, man. This gets addressed in every basic statistics intro. I’m not saying there aren’t problems with the idea. There are. But it’s the same problem as saying that the limit of x as y approaches infinity is 1. Is calculus fucked and completely broken because y never gets to infinity?

No. Not even a little.

Lesson #3: social sciences are broken.

Completely agree.

The alleged twist: They are broken because the basic concept of probability is broken.

The absolute comically stupid apex of all of this linguistic garbage is that Taleb uses the social sciences to prove how little his argument makes sense.

He can’t tell the difference between a bad model and a broken system for creating models. He proudly claims that probability is broken because the models that people create to estimate outcomes are bad. And he uses the worst possible collection of models for the most complex systems that we know of as proof that the system is broken.

There are absolutely legitimate criticisms not only of frequentist statistics, but of inductive logic in general. It’s only a conversation that’s been going on since, I don’t know, Pythogras? Maybe before?

There is nothing new or insightful here. Taleb is wrapping up and idea that everyone already knows in a word cluster so obtuse my 5 fifth-grade English teacher would destroy as some deep new insight.

But hey, he posted it on Medium, so it must be awesome.

Give me a break.

Re:The Logic of Risk Taking by lz400 on Wednesday October 04, 2017 @10:27PM (#15406355)
I think Taleb likes to think of himself as a scholar so smart that has destroyed mainstream scholarship and his contributions are not being recognized by the lesser scholars. But he's smart enough to recognize that he's just not that deep of a thinker or has anything particularly insightful to say (outside of some financial modeling topics perhaps) so he uses all that pedantic prose and name dropping to cover for it. He's certainly cashing on it so it's not like he's going to stop but I can't shake the feeling that he knows, is bitter and that shows in his writing.
Re:The Logic of Risk Taking by dsjoerg on Wednesday October 04, 2017 @09:57PM (#15406248)
I was hoping to find a comment about how terrible the writing was, thank you. The writing is terrible.
Re:The Logic of Risk Taking by ianamartin on Wednesday October 04, 2017 @10:37PM (#15406385)
I’m not completely convinced that Taleb isn’t an elaborate hoax designed to win the Turing test.

Look, I don’t like being so harsh. But the guy gets really fundamental things really really wrong. He makes me angrier than Gladwell combined with the Freakonomics guys. Which is a lot.

There’s a genre of pop-statistics that is genuinely bad for people. And it all follows the same pattern. Science says x is true, but if you think about it, surprising y thing is true. Because I’m a special snowflake, and so are you.

It’s absolute garbage.

And I don’t even know that much about stats. I dropped out of a violin performance and music theory degree. I just read a few stats textbooks and had some good professional mentors because some of this is relevant to my work as a data engineer.

This is basic level stuff that’s going wrong. The guy is clueless about everything he writes. But I’ll give it to him that he’s a genius at marketing.

He publishes books and makes money off of them. I do not. So he wins that way. But if I ever write a book, it won’t be in the fiction section where his belongs.

Re:The Logic of Risk Taking by pizza on Wednesday October 04, 2017 @11:36PM (#15406546)
> He can’t tell the difference between a bad model and a broken system for creating models. He proudly claims that probability is broken because the models that people create to estimate outcomes are bad. And he uses the worst possible collection of models for the most complex systems that we know of as proof that the system is broken.

> And I don’t even know that much about stats. I dropped out of a violin performance and music theory degree. I just read a few stats textbooks and had some good professional mentors because some of this is relevant to my work as a data engineer.

eye roll

> What Taleb does here (again) is conflate a legitimate criticism of frequentist statistical models with a critique of incomplete decision models.

double take

..bullshit?

His critique:

* E[f(x)] is different from E[x] ~ X

* Jensen's inequality => use a convex decision/utility function/behavior whenever you are exposed to events out of your control

* Things like coin flip outcomes or binary true/false observations ( i.e. not simply just probabilities! ) depend on their zeroth-moment; no dependency on magnitude of the moment, such as in outcomes that are made complex by outcome-dependency upon higher-order moments

* Bet on processes whose payoff f(x) has a stochastic first moment, but don't do so stupidly, and bet rarely

* Any uncertainty about the generating process concerning higher-order moments of very small probability outcomes makes the payoff more attractive for this class of complex outcomes

If you care at all, the preprint of Silent Risk will more than satisfy for a collection of proofs. Hell even the page I just happened to be looking at would [0]

[0] https://i.imgur.com/W1V3jyV.png?1

Re:The Logic of Risk Taking by ianamartin on Thursday October 05, 2017 @03:57AM (#15407344)
All of that hinges on your 3rd bullet point.

And translating into plain English means that if you bet your life or your life saving on the outcome of a coin toss, you're an idiot.

That's not an effective critique of frequentist statistical theory. That's obvious.

The absolute best and most courteous interpretation of this is, "Don't be an idiot." Thanks, Taleb, for that most insightful piece of advice.

The problem here is that this series of thoughts ascribes an attribute to statistical theory that is absolutely non-existent: that there is any weight given to any individual trial of an outcome. No one thinks that. He, and now you, are arguing against a completely non-existent point.

Look, if you have anything approaching a decent model, you know the range of possible outcomes.

If you want to understand flipping a fair coin, the outcomes are that there's a heads and a tails and it has to land on one of them.

What Taleb and you are arguing is that there's a third possibility you didn't think of where the coin lands in an alternative universe, and Margaret Thatcher is the queen of the United States, and all the puppies die.

And if that's the real set of possibilities, I want to play that game. Because I would trade all of the puppies for Queen Thatcher right now. (No offense to you if you are actually a puppy. I just really don't like Trump.)

Saying that statistics is broken because infinity doesn't exist is like saying that basic arithmetic is broken if 1 equals 0.

No shit. Yeah, when you change the basic rules of how things work, things get broken really quickly.

I'm sorry if I'm being obtuse here, but I don't see anything even remotely worthwhile in his article, your post, or your link.

Trying to work backwards from a stochastic model to the outcome of an individual event will never work. And no one thinks that it will. If you want certainty, you have to use an entirely different model of logic.

Taleb is criticizing inductive logic for not being deductive, and then claiming that you'll get just as good a result by limiting the inductive model and using only a portion of what makes it work.

It's utter and complete bullshit, as far as I can tell.

Happy to be corrected, if you think I'm wrong.

Can we agree that a charitable plain english version of what he's trying to say is this:

Don't play the long odds if you aren't playing the long game. If you're playing the short game, don't risk more than you can afford to lose.

If he's saying more than that, please enlighten me. In my first post on this topic I said that he wasn't entirely wrong, but he is entirely useless. If you've made it past the age of 5 and haven't figured this out already, well, reading his book won't help you because you won't understand a goddamn word in it.

I will read his next book because it's unfair to criticize something you haven't exposed yourself to. But I can't imagine any possible way that it could comprise a collection of proofs.

Again, what is there to prove? That probability doesn't work good when you change the rules about how it works? That you can't take a distribution and apply it to a single point in time? That you can't take a collection of inductive data and use it to prove a universal truth?

No sane person has ever claimed that you can do any of those things. All I can tell is that Taleb is claiming to be both interesting and original by writing a book that says you can't do any of those things.

Welcome to the grown-up world, Taleb. We already knew that.

P.S. Santa Clause isn't real. Maybe I should write a book about how not real Santa is. I bet Murray Gell-Mann would immediately grasp this concept.

Re:The Logic of Risk Taking by pizza on Thursday October 05, 2017 @06:06AM (#15407932)