Yet this, or any analysis that seeks to predict the future based on
current knowledge, cannot help but overlook the possibility of Black
Swans. The largest event to date is no guide to even larger events
that could occur but have yet to. So is there a fundamental obstacle
to mass killing by an individual, or are we less safe than we (or at
least Kelly) think we are?
The article offers two main reasons why this should be so, which are
that killing large numbers of people is a complex task, and that
social resistance hinders recruitment of resources. Which got my
inner evil genius wondering if there were ways to bypass these
difficulties.
Perhaps these restrictions only apply to certain types of resources.
The capacity to recruit and use physical resources scales linearly
with manpower, whereas that for abstract resource does not. David
Deutsch has argued that any physical states that are not forbidden by
the laws of nature are necessarily achievable with the right
knowledge, matter and energy. Insofar as any such villain seeks an
achievable state of affairs, their limiting factor must be knowledge.
If their plan has minimal physical dependency, relying on ideas over
matériel, then they are no longer limited by their solitude.
Supervillains seem to run up against Ashby's Law, which suggests that
the number of states one system must be able to take on in order to
modify an autoregulatory system must be greater than the number of
states that autoregulatory system can take on, i.e. you need to be
more adaptable than the thing you are trying to change. But the mind
has (if used right, anyway) potentially more Kolmogorov complexity
than most systems, since systems can be represented by simpler models
in the mind. And even that level of complexity is not requisite if a
momentary disturbance is all that is desired; the climate will
outlive any butterfly, but the much-vaunted butterfly of chaotic
alignment can effect a storm by tipping a fragile system over the
edge, without being nearly as large as Mothra. As in tai chi,
it is wiser to defeat a system by using its momentum against itself.
And it is always easier to perturb something trying to stay still
than to defend it from threats of unknown nature.
What if we rethink our conception of aetiology? We live in a complex
world where causal webs are densely interconnected, and Aristotle's
four causes become myriad. The assassins of Archduke Franz Ferdinand
and his wife triggered World War I and caused the death of much more
than two people, even if they cannot be held responsible. Perhaps
they weren't even a sine qua non, and war would have broken
out eventually anyway, but in our history they became causes. From
this perspective, can we say that a warmonger requires the assistance
of thousands of soldiers or an institute of nuclear scientists to
kill millions? After all, those are already in position, waiting for
just such a one to turn up. Of course, there are often checks and
balances in most political systems, but the possibility remains. But
what of dangerous ideologies propounded by a single revolutionary?
These move populaces, which are not obliged to listen to the other
side of the argument. Much more covert triggers are also possible -
consider how the professor who rejected Hitler's application for art
school could have toppled a chain of dominoes leading to the deaths
of millions. This also gives us, chillingly, the possibility of what
Edogawa Rampo termed the perfect crime, one for which the criminal
could not be discovered or judged guilty. Of course, as in Rampo's
short stories, criminals view these as works of pride and can never
help boasting of them. But by then it is already too late for us.
Conversely, some actors may be upstream of many more agents and
events that lead to mass murder. They are crucial nodes in the
network, and stopping them could have a tremendous positive impact if
they could be identified beforehand. But perhaps it is safer to
reduce our vulnerabilities instead. It is too slow for us to be
stationary and reactive; we must go on the offensive. Testers could
seek out weaknesses in society and infrastructure. We could even hire
potential supervillains to carry out the reviews, much like how
companies may hire hackers to hacker-proof their networks. Because if
vulnerabilities exist, they're going to be discovered eventually,
whether intentionally or inadvertently. We might as well make sure
they are discovered by someone on our side. Crowdsourcing to the
evil-genius-wannabes lurking out there would work, as would designing
a supervillain emulator. So perhaps a hostile post-Singularity
superintelligence wouldn't be such a bad thing for humanity after
all. It would make for a bloody good game of cat-and-mouse, at the
very least. Or Pinky-and-the-Brain, as we plot our own demise
together.
No comments:
Post a Comment