Analysing Uncertainty (David Spiegelhalter) RSS 13 October

Discrete areas in probability-hazard space

Discrete areas in probability-hazard space

So this was all quite interesting.  David Spiegelhalter’s main point was that there was a lot more to uncertainty than uncertainty in parameter estimates.  Could you for instance have a probability-hazard space when nothing was properly scaled?  Maybe it would be better to use discrete clusters as suggested by Ortwin Renn (and as above) where for instance Cassandra had high probability and a high degree of damage.

Parameters might be subject to random variation (aleatory uncertainty), or they might be fixed but you just didn’t know them–(epistemic uncertainty).  But then your model might be–undoubtedly was–wrong.  So could you average over models?

At the end (and thinking of the banking crisis/credit crunch) he concluded that one needed a clear separation between (1) modellers presenting their work with due humility and (2) decision-makers accepting this work with due caution.  But this leaves out a further level of uncertainty–you don’t know what the right question is either.  If the modellers and decision-makers work together they can negotiate a question–often along the lines of ‘do we need to do something about this now’–that the modellers can actually help with.

And along the way we learned many interesting things:  for instance, agnotology, the deliberate production of ignorance by for instance the tobacco industry and climate change deniers…

Tags: ,

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: