I mean, apply some bayesian reasoning about the probability of someone being a terrorist _given that_ he looks like an Arab and/or is Muslim — and you won't be surprised some people are screened more thoroughly. Had the world evolved differently, they might be paying increased attention to white males with freckles and red hair.
The US for some reason likes to pretend that everyone is a terrorist with equal probability and avoids "racial profiling". This is kind of like sticking fingers in your ears and shouting loudly "can't hear you!".
Statistical reasoning and Bayesian in particular, is frequently misunderstood in spite of its simplicity. It is also preferred that Bayes be spelled in caps. Just for arguments sake, let us assume that an Arab lookalike is more likely to blow up a plane. (digression: I also amuse myself by trying to look like one before flying, at least to the TSO.)
The short answer to the source of the misunderstanding is that one usually forgets to take into account that there are far less Arab lookalikes than non Arab lookalikes. A common mistake is to go with the conditional probability rather than the posterior.
The probability that an Arab is a terrorist, strictly speaking, the empirical conditionalprobability is the number of Arab with an intent to blow up the plane divided by the total number of Arab in the population under consideration. For arguments sake let me agree with you that this is higher than the corresponding conditionalprobability for a non-Arab. Lets represent these quantities by pa and pb respectively.
The likelihood ratio that an Arab lookalike will blow up a plane is then
(pa * Pa) / (pb * Pb)
where Pa and Pb are the probabilities that a random sample from the population is an Arab vs non-Arab. What are known as the priors. Here we have Pb >> Pa.
In order to justify that Arabs be selectively screened it is necessary that
pa > ( pb * Pb )/(Pa).
This becomes equivalent to the condition
pa > pb
when Pb = Pa.
So I suspect that if anyone is
sticking fingers in your ears and shouting loudly "can't hear you!".
isnt quite, as you put it, "The US".
There are other issues:
It is hard to argue that pa is indeed greater than pb, i.e. for random samples of equal size there would be more terrorists in the Arab sample than not. Its quite a radical statement and justifications should have been mandatory.
Also I disagree with your assertion that "Had the world evolved differently,..." things would have been different. My counterexample is that the Irish or the Catholics in US were not subjected to any of this on flights to Britain when the IRA were bombing away innocent civilians there. BTW this is not the only counterexample.(I also suspect that O'reilly's over the top assertion that terrorist = Muslim is a compensatory mechanism, conscious or unconscious I do not know.)
The logical explanation for the way the security response is playing out in the field of travel seems to be that people are afraid of unfamiliar people and all it takes to tone down that fear is just some superficial familiarity (such as skin tone).
So I am hopeful that this stage too will pass. Sooner the better, and I hope with no permanent scars.
I mean, apply some bayesian reasoning about the probability of someone being a terrorist _given that_ he looks like an Arab and/or is Muslim — and you won't be surprised some people are screened more thoroughly. Had the world evolved differently, they might be paying increased attention to white males with freckles and red hair.
The US for some reason likes to pretend that everyone is a terrorist with equal probability and avoids "racial profiling". This is kind of like sticking fingers in your ears and shouting loudly "can't hear you!".