Why Foxes Are Better Forecasters Than Hedgehogs

もっと見る

Future Continuos

Maja Ribicにより

GP

Olivier Auberにより

English Will / Going to

elizabeth lizにより

Data Protection Stuff Amarja

Amarja Kathapurkarにより

Why Foxes Are Better Forecasters Than Hedgehogs

Aggregate of experts do better than single experts

Experts won with their students

Hedgehog-Fox Ecosystems

Foxes pick bits and pieces of big hedgehog ideas-and create eclectic mish-mashes that, on average, have more predictive power than the original ideas

Hedgehogs push big ideas as far as reasonable-and often beyond

Foxes vs hedgehogs

Mindless competition

Ideal-type foxes - "pursue many ends, often unrelated and even contradictory, ... entertain ideas that are centrifugal rather than centripetal; ...without seeking them to fit them into, or exclude them from, any one all-embracing inner vision."

Diffident about their ability to forecast

Skeptical of grand theories

Ideal-type hedgehog - "relate everything to a single central vision, ...in terms of which all that the say has significance"

Confident in their forecasting ability

Extend their theory into many domains

Parsimony

Two aspects of accuracy

Discrimination

perfect = 100%

Calibration

perfect = ZERO

Benchmarks

Do you know what you do and do not know

Scrutinize mental process

Keeping track of your guesses

Meta-knowledge indicators

Knowledge indicators

Prediction markets

Probability and humans

if you use complex statistical models virtuall all experts loose

People listen when you say 1 or you say 0

"Humans can only distingiush three levels of probability: impossible, maybe, inevitable", Amos Tversky

not linear

What do hedgehogs do well

Blak Swans

Assign higher probabilities to big changes that do materialize (but at steep cost in false-positive predictions)

Which makes them more interesting ;)

An 18-year Research Project

Good predictor is - How you think!

"Foxes" beat "hedgehogs"

Traditional divisions of opinion have little weight

optimistiv vs pessimistic

liberals vs conservatives

predictions pass clarvoyance test

28 000 forecasts

12 years average work experience

Almost all had post-graduate training

284 area experts

Russian-American tension - comittee

similar predictions with war in Iraq

then gorbachov comes

OLIS - outcome-irrelevant situation

usual NRC due dilligence

rather conservative predictions

Good political judgement