Catégories : Tous - evidence - probability - uncertainty - experts

par Paweł Badeński Il y a 12 années

672

Nassim Nicholas Taleb: A Crazier Future

Nassim Nicholas Taleb explores the unpredictable nature of the future, emphasizing the concept of "silent evidence" and how our perception of probability is often skewed by focusing on survivors rather than considering all data points.

Nassim Nicholas Taleb: A Crazier Future

Nassim Nicholas Taleb: A Crazier Future

Ludic fallacy

we take risks - because we don't know we take them - we're blindfolded

Uncertainity principle

Black swan in History/Philosophy

Karl Marks
Doctor's - Empiricists
Philosopher's through time got domain dependent
Hume's Problem

Silent evidence

Neil Fergusson
overcausation
wrote a paper although we think IWW predictable the bound market did not predict it
fun with data
probability problem
people don't write biographies - "how I lost one million dollars"
we take only pool of information and dicard the rest
we compute probability on those who survive, not all
I world war - tension between England and Germany
you don't look at episodes of tension that do not let to the war

Ignore experts

Subtopic
Why are we suspicious of the bishop and be a sucker when it comes to an economist
Robert Tivers
we're not good at predicting wars
2 rules
Never take advice from someone wearing suit and tie
Never ask barber if you need haircut
Phillip Tetlock
processed 27 000 predictions
Tunnell
Mathematicians
Mediocristan is easily mathematized and will be mathematized
small number of equations is useful

Domains

Extremesitan
dominated by the exceptions
Whenever you take a large sample, a small number of observations in that domain will represent a big share of a total
Weigh thousand people and add one person that is the wealthiest you can think of
Mediocristan
paid by the hour
dominated by the collective
When sample is large, exceptions can happen, but they will not screw with the result

The law of large numbers

Weigh thousand people and add one person that is the heaviest you can think of

Dynamic Revision "palimpsest"

you don't remember what you actually predicted, but you revise your memory to adapt your prediction to current outcome

Black swan

History of the turkey - that gets fed for thousand days
an event not a bird
illusion of predictability post factum
low predictability, high consequence
I world war
a tie
very difficult to predict, based on prior information
Australia
you're sooner see a black swan