On Takatoshi Kato's predictions / suit-and-tie-wearing hotshots

"These were days before I decided to climb up the mountain, speak slowly and in a priestly tone, and try shaming people rather than insulting them. Listening to Kato's presentation, I could not control myself and flew into a rage in front of two thousand Koreans – I was so angry that I almost started shouting in French, forgetting that I was in Korea. I ran to the podium and told the audience that the next time someone in a suit and tie gave them projections for some dates in the future, they should ask him to show what he had projected in the past – in this case, what he had been forecasting for 2008 and 2009 (the crisis years) two to five years earlier, 2004, 2005, 2006, and 2007. They would then verify that highly Venerable Kato-san and his colleagues are, to put it mildly, not very good at this predictionizing business. And it is not just Mr. Kato: our track record in figuring out significant rare events in politics and economics is not so close to zero; it is zero."

"What was getting me in that state of anger was my realization that forecasting was not neutral. It is all in the iatrogenics. Forecasting can be downright injurious to risk-takers – no different from giving people snake oil medicine in place of cancer treatment, or bleeding, as in the story of Geaorge Washington. And there was evidence. Danny Kahneman – rightfully – kept admonishing me for my fits of anger and outbursts at respectable members of the establishment (respectable for now), deeming such behavior to be unbecoming of the wise member of the intelligentsia I was supposed to be. Yet he stoked my frustration and sense of outrage the most by showing me the evidence of iatrogenics. There are ample empirical findings to the effect that providing someone with a random numerical forecast increases his risk taking, even if the person knows the projections are random.
All I hear is complaints about forecasters, when the next step is obvious yet rarely taken: avoidance of iatrogenics from forecasting. We understand childproofing, but not forecaster-hubris-proofing."

The financier Warren Buffett states that he tries to invest in businesses that are "so wonderful that an idiot can run them. Because sooner or later, one will."
Nassim Taleb, in Antifragile

Models tie us to a mast of logic

Pintura de John William Waterhouse

Curso: Model Thinking, por Scott E Page 
Aqui

Why Models?
Reason #1: To be an intelligent citizen of the world.
Reason #2: They make us better clear thinkers.
Reason #3: To understand and use data.
Reason #4: To decide, strategize and design.

Models are the new lingua franca, they're the language of business, they're the language of politics, they're the knowledge of the non profit world.

Models tie us to a mast of logic, and by doing so, we figure out which ways of thinking, which ideas are useful.

Models are better than we are.

The accuracy of different ways of predicting ('Formal Models' do better):

Model by Philip Tetlock

Models are fertile. Once you learn a model for one domain, you can apply it to a whole other bunch of domains. You can also use them to find out who wrote a book.

Models make us humble. They make us see the full dimension of a problem.

Thinkers who use many models, know best. So what is really important, is to have many models.
The 'Foxes', who had lots of models, did much better than the 'hedghogs', who had no models (Philip tetlock). Formal models did better than 'Foxes'. People with lots of 'Formal models' will do even better. So the idea is to use multiple models.

Models make us clearer thinkers. They work through logic. They inductively explore. They allow us to understand the class of outcome [05:42]. They help us to identify logical boundaries [07:26]. models allow us to communicate our ideas and what we know, really simply [08:21].

Steps to build a model:
— Name the parts (people, money, places, preferences, etc)
— Identify relationships (and how things play out)

People use models to:
— Understand patterns
— Predict points [02:07]
— Produce bounds
— Retrodict
— Predict other stuff
— Inform data collection
— Estimate hidden parameters
— Calibrate

How can models help us?
— They can be real time decision aids.
— They can help us with comparative statics.
— They can help us with counter factuals.
— We can use them to identify and rank.
— They can help us with experimental design.
— They can help us design institutions, themselves.
— They can help us choose among policies and institutions.

Foxes and Hedgehogs

Só sei que nada sei

"Experts are just human in the end. They are dazzled by their own brilliance and hate to be wrong. Experts are led astray not by what they believe, but by how they think."

"The line that separates the possibly predictable future from the unpredictable distant future is yet to be drawn."

Daniel Kahneman, in Thinking, Fast andSlow

Será que o Marcelo alguma vez leu Kahneman, ou Tetlock?

"People that spend their time, and earn their living, studying a particular topic produce poorer predictions than dart-throwing monkeys who would have distributed their choices evenly over options. Even in the region they knew best, experts were not significantly better than nonspecialists."
Daniel Kahneman, in Thinking, Fast and Slow

Philip Tetlock explicou as, denominadas, "previsões dos especialistas" num proeminente estudo que durou 20 anos e que publicou no seu livro Expert Political Judgement: How Good is it? How Can We Know? Entrevistou 284 pessoas cuja actividade principal era o comentário político ou económico. Recolheu mais de 80.000 previsões e verificou a sua veracidade. Os resultados foram devastadores para os peritos.