Looking for the robber

Given the numbers in the scenario, most people assume that

the smart place to start looking for the attacker in the incident

is among the indigenous community, because of the witness

testimony, even though a ‘possibility’ clearly exists that the

witness may have made a misidentification.

But suppose that 10,000 white people live in the city and just

1 ‘indigenous’ one. The best strategy is then a slam‐dunk for

the police, isn’t it? But then remember that the witness sees

‘indigenous people’ 20 per cent of the time. The evidence isn’t

so persuasive now, is it, because the witness will often say

someone is indigenous when they aren’t. The ‘mathematically’

correct answer to the original scenario is that a substantially

higher probability exists that the villain involved in the street

robbery was European rather than indigenous — so the police

should be looking for a European robber. Mathematicians use

a technique called Bayesian analysis to get an exact figure,

but the important thing is to be aware of the general issue.

The reliability of the identification of the robber as indigenous

is, in this case, 41 per cent, only about half the 80 per cent

reliability people comfortably opted for ‘without thinking’.

Astronomical wrangles

You can be forgiven if you think this is a bit of a no‐brainer —

of course the Earth goes around the Sun, and so you support

the new‐fangled theory. But when you do that, you have to

accept that you are throwing out any pretence that you think

scientific matters should be settled on the basis of the facts

and figures and evidence.

This is what the radical philosopher, Paul Feyerabend, meant

when he argued in his books that in science, the only rule is

that there are no rules, and what’s more, that only by breaking

rules that scientists have been able to make the progress for

which they are — later on — praised