In article ,
"Gary Drescher" wrote:

"AES" wrote in message
...
To quote someone I overhead yesterday: "If there's a 2% chance, based
on all information known at the time, that the guy is a suicide bomber,
and 50 people in the subway car -- well, the choice is unfortunate, but
very clear."

I see several problems with this proposal.
(REMAINDER OF THIS RESPONSE APPENDED FURTHER DOWN)



As the one who quoted -- not made -- the assertion above, I recognize
the merits of essentially all the points made in the reply appended
below.

We're faced, however, with a new and very difficult situation in the
suicide bomber phenomena. The fact that it's primarily based in, caused
by, and supported by religious fanaticism (not primarily anything we do)
makes it all the more difficult to cope with.

Simply quoting Ben Franklin's "better that 100 guilty escape punishment
than 1 innocent be convicted" -- an aphorism that validly applies to a
very different situation or set of circumstances -- and concluding from
this, as some apparently do, that the bottom line is clear: the police
should never shoot in any suicide bombing situation, is not a conclusion
I find acceptable.


REMAINDER OF RESPONSE:
* The proposal presumes that an expected-gain calculation (about lives saved
or lost) is morally decisive. But there are well-known problems with such a
strict utilitarian criterion. For example, strict utilitarianism implies
that if you are healthy, killing you to harvest your organs would be
justifiable--doing so would save several lives by taking only one life. The
philosophical issues are too complex to analyze here; suffice it to say that
at present, it is (to put it mildly) less than "very clear" that killing for
organs--or killing someone who is known at the time to be 98% likely to be
innocent--is morally justifiable, even if (in both cases) the expected-gain
calculation is somewhat favorable.

Other difficulties concern the implementation of the proposal, even if the
utilitarian calculation were indeed morally decisive. Some are technical
problems:

* People's intuitions about small probabilities are notoriously inaccurate.
If a subject stands out because evidence shows he is hundreds of times more
likely than the average person to be a suicide bomber, then he may seem to
be at least 2% likely to be one, even though a trivial Bayesian calculation
shows that the actual probability is orders of magnitude less than that. But
we cannot realistically expect police to perform Bayesian calculations while
making split-second life-and-death decisions.

* The calculation, as framed, presumes that in the 2%-likely case in which
the person killed is indeed a suicide bomber, the killing is both necessary
and sufficient to prevent the 50 deaths. That might be roughly true if, say,
you shoot a bomber who is running toward (but still distant from) a crowd.
But the presumption's accuracy is much less clear if (as in the present
instance) the person is already on the crowded train (the bomb, if any,
might explode anyway due to a passive-release trigger, or just from the
impact of falling, due to TATP's instability), and is already flat on the
ground, surrounded by police, without having detonated any bomb (so the
police, at that point, may well be able to subdue him less lethally).

Other problems stem from the readily foreseeable deliberate or inadvertent
abusability of the proposed policy:

* If the 2%-likely suspects belong disproportionately to an identifiable
minority group, then the majority will (accurately) perceive themselves to
be much less at risk from the proposed policy, and will thus have diminished
incentive to give due consideration to the moral and technical problems with
the policy.

* If those disproportionately targeted by the policy belong to a widely
*disliked* minority, then their endangerment may be devalued even further
(often unconsciously, and thus unscrutinized); their endangerment may even
be perceived by many as desirable, rather than as a drawback of the policy.

* Our species has demonstrably inherited a primate quasi-sexual appetite for
(often-lethal) gratuitous cruelty (see e.g. Nell 2005, forthcoming in
Behavioral and Brain Sciences). Contemporary civilization has devised
various ways to help keep this tendency in check, but the power to engage in
socially approved killing of innocents inevitably serves in part to give
expression to this unfortunate inclination (the inclination can be
manifested --without necessarily being recognized as such--both by the
advocates of the proposed policy, and by those whose task is to execute it).

For these reasons and others, I believe a reasonable policy only permits
killing someone who is overwhelmingly likely to be posing a lethal threat
that cannot otherwise be countered.

--Gary