[p2p-research] Drone hacking

[Andy Robinson]

The question of whether people are fundamentally predictable is key here.
But there is also the problem of data collection.

I still insist there will never be the amount of data available on *marginal
* populations to make this possible, certainly not on an individualised
basis.  The kind of thing that can be mathematically modelled based on
observations of individual behaviour is the kind of thing that leaks a lot
of data to be collated and which is relatively regular, such as online
purchasing practices.  This is simply not going to apply in the cases:
firstly, of indigenous societies which are resistant to modern practices
generating this kind of paper/data-trail; secondly, of the global poor who
are working, buying, sharing and living mainly by means which don't leave
such a trail - in everyday networks and in the informal economy; and thirdly
of people who have specifically decided to live 'off the grid' and who are
thus observed a lot less than other people.

Since the context in which this arose is the context of drone hacking and
the war in Afghanistan, this is absolutely crucial.  Where are the inputs
going to be coming from from which extrapolations can be made about the
future behaviour of villagers in Helmand province or informal economy
workers in Kandahar?

Another problem.  Since the predictions are based on previous individual
behaviour and/or on comparison with others' patterns of similar individual
behaviour, they will not be able to predict how individuals will respond to
situations which are previously unforeseen, or predict cases where
individuals undergo large transformations in their social behaviour - *
except* where they have a similar case from which to extrapolate already on
record.  When the group of occurrences is very small, they will not be able
to make valid extrapolations, because they will not know which individual
differences between the cases are relevant for prediction.  They are not for
instance going to be able to tell how people in London would respond to a
tsunami, because they have no data on these people individually or
collectively in relation to tsunamis.  They would either have to extrapolate
from how each of the individuals in Birmingham responds to personal crises
and assume that this adds up to their response to the collective crisis, or
they would have to make aggregate guesses based on the behaviour of similar
individuals in somewhere where there was a tsunami.  In this case, the
accuracy of the predictions depends on the accuracy of the analogies drawn
by the computer - that Fred Smith in London has the same personality-type as
Dita in Banda Aceh and hence will act the same way; or that Fred Smith will
react to the tsunami in the same way that Fred Smith reacted many years ago
to the 7/7 bombings or the way he related to his wife dying.  Neither of
which is a very solid basis for predicting how Fred Smith will act, no
matter what algorithms the computer is using to establish the best possible
analogy.

And given that they only have data from massified processes, their data is
always incomplete.  Since the context if *complex* and *chaotic*, this could
easily be fatal: the missing piece of data, however apparently trivial,
could make the difference in predicting how someone would actually act.

People may well display predictable patterns which could be collated but of
which they are unaware, but these are expressions either of the unconscious
or of habituated practices.  These could often be spotted by an astute
anthropologist or psychoanalyst.  Just supposing these kinds of observation
could be imitated by a computer (and this is*not* what's being discussed in
relation to mathematical modelling) - it still would not add up to the
ability to predict individual behaviour, firstly because these bases of
predictability are malleable, and secondly because the factors which can
cause the disruption of predictable patterns are themselves complex and
chaotic.

Hence, the calculations in question will only be useful for
predicting/directing massified behaviour, and only for as long as people
remain massified and their lives reproduce familiar patterns which have
already been observed a great many times.  Actually I would like to see real
information rather than assertions here - what proportion of the time do
these 'primitive' versions of the coming supercomputers actually
successfully predict even something as simple as online buying behaviour?
How often do they actually manage to induce people to buy something?  If
they were as successful as is being made out, people would in effect be
addicted to these sites, since the sites could induce them to constantly
come back and constantly spend up to their limits there.  They would be
gaining market share exponentially at the expense of other sites not using
this technology.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listcultures.org/pipermail/p2presearch_listcultures.org/attachments/20091222/bd8f3f33/attachment.html>