I’m guessing that the “Erroneous domain” column describes domains where the named biases allegedly don’t function as such. I’m less sure that the table actually describes situations where that’s true; it is somewhat less than obvious to me, for example, that our naive dread of terrorism vs. falling from ladders closely mirrors expected experience.
it is somewhat less than obvious to me, for example, that our naive dread of terrorism vs. falling from ladders closely mirrors expected experience.
Terrorism is a black swan (rare high impact) event. Thus the statistics rationalists-types use for dismissing it are almost meaningless. Consider attempting to estimate the danger of asteroid impacts or rogue AIs by looking at the last several millennia of data, terrorism has a similar problem.
Terrorism is a black swan (rare high impact) event.
No, it’s not. The biggest terrorist event ever is… what, 9/11 with something like 5k deaths? That’s not high impact by any means. Annual variations in flu virulence outweigh that.
Consider attempting to estimate the danger of asteroid impacts or rogue AIs by looking at the last several millennia of data
One of those is not like the other, one of those does not belong.
Terrorism is a black swan (rare high impact) event.
You are misusing the term. Black swans are not rare high-impact events—they are events that nobody has considered possible. The original black swan—finding black swans in Australia—certainly wasn’t a high-impact event.
Consider attempting to estimate the danger of asteroid impacts or rogue AIs by looking at the last several millennia of data, terrorism has a similar problem.
I disagree. Terrorism has been a constant presence throughout human history. It’s not rare, at all. There is lots of historical data.
EDIT: I was wrong. Taleb actually defines black swans (at the beginning of The Black Swan book) as outlier events that have three characteristics: (1) rare; (2) high-impact; (3) not foreseen but back-explained in hindsight.
Nassim Taleb
I don’t understand the table. What do “Erroneous domain” and “Justified domain” mean? Also, “Prediction markets” is not a name of a bias.
I’m guessing that the “Erroneous domain” column describes domains where the named biases allegedly don’t function as such. I’m less sure that the table actually describes situations where that’s true; it is somewhat less than obvious to me, for example, that our naive dread of terrorism vs. falling from ladders closely mirrors expected experience.
Not sure why prediction markets are on the table.
Terrorism is a black swan (rare high impact) event. Thus the statistics rationalists-types use for dismissing it are almost meaningless. Consider attempting to estimate the danger of asteroid impacts or rogue AIs by looking at the last several millennia of data, terrorism has a similar problem.
No, it’s not. The biggest terrorist event ever is… what, 9/11 with something like 5k deaths? That’s not high impact by any means. Annual variations in flu virulence outweigh that.
One of those is not like the other, one of those does not belong.
You are misusing the term. Black swans are not rare high-impact events—they are events that nobody has considered possible. The original black swan—finding black swans in Australia—certainly wasn’t a high-impact event.
I disagree. Terrorism has been a constant presence throughout human history. It’s not rare, at all. There is lots of historical data.
EDIT: I was wrong. Taleb actually defines black swans (at the beginning of The Black Swan book) as outlier events that have three characteristics: (1) rare; (2) high-impact; (3) not foreseen but back-explained in hindsight.
I’m using Taleb’s definition.
Most of which the people who try to compare it statistically to falling of ladders tend to ignore.
He means that alleged bias of people not trusting/creating prediction markets.