How should we approach this? By sketching personalities, and mapping hierarchies of dominance, or by moving tokens on a board? In “Al Qaeda Hostage Release, & Lantana Weed Control”, it’s asserted that Qatar cannot be tokenized. In other words, Qatar has a complex internal structure that spits out behaviors considerably more complicated than a shocked lab rat.
Some problems become simplified en masse, while others become more complicated. The physics of bodies with gravitational fields provides an example of magnificent clarity:
The mechanics of a single body is trivial.
The mechanics of two bodies has a simple formula solution.
The mechanics of three bodies has no formula solution. All you can do is calculate the evolution of their motions in tiny increments of time, and keep doing it till you get to the time you want. This is called “iteration.”
- The mechanics of more than three bodies becomes increasingly miserable, until, when you have about 13 or more, provided the bodies are identical, and you don’t care about their individual identities, statistics emerge, and can be solved for. Statistics are numbers that characterize the particles as a bunch.
Inexact analogies to human behavior are obvious: individual, cabal, crowd, tribe, society… The imprecision, and perhaps, lack of practical exploitability of these relationships is as frustrating for intelligence as it is for physics. Could one, for example, predict a revolution by monitoring the behavior of a crowd, as, for instance, with tweets before the 2011 Egyptian revolution?
The U.S. intelligence community observed a large scale social instability prior to the resignation of Hosni Mubarak, but was unable to predict the revolution itself. To predict the fall of Mubarak would have required analytic tools that work from the large to the small. One group (citation missing) published claims of a model-with-software that predicted the fall of Mohamed Morsi, supposedly by analyzing the claims, grievances, and resources of the “actors” of Egypt’s political scene. But as the claim was made after the fact, it and a couple of bucks will buy you a latte.
While informatics is still largely preoccupied with models and statistical quantities thereof, it has also moved past that, with the new understanding that some important information may be uncovered, if we are only willing to cede the need to know the reasoning. The neural network is one such device. The details of how a complex neural network actually functions cannot be known, because the states are too complex to interpret individually. But if it says yes, or it says no, and it has a track record, what the hey?
This was one of the inspirations behind crowdsourcing intelligence; put lots of warm & wet neural networks on the case, and feed their outputs into some kind of algorithm that would cleanse the data of their bizarre individual opinions, exposing hidden gems. Suppose, for example, you had assembled a collection of opinionated heads who always, without exception, gave the wrong answer. You pose the question: Will Bashar Hafez al-Assad be among the living after June 16, 2015, the end of the month of Ramadan? The group reply is “without a doubt, absolutely, bet your life on it”…and the group is always wrong. Would you ignore the prediction?
If this seems farfetched, the financial markets use methods like this, called contrarian sentiment indicators, and they are valued by many for market timing. The simulated market is a crowdsourcing methodology for intelligence work, perhaps the most widely used. There was an operating “terror market”, devoted to predicting terror events, which was terminated by ethical considerations, such as the possibility of encouraging the act.
You should not underestimate the possibility that your own warm & wet and conveniently available neural network can tackle prediction problems with results better, at least, than the “opinion pieces”. It does require a high level of self-awareness to create the required virtual “detached little man.” Do some meta-analysis about your own thinking. When you think about Iraq, ISIS, Iran, and Syria, do you gravitate to:
- Individual players?
- “Peace of Westphalia constructs”, with political maps populated by men wearing western business suits?
Without tipping my hand before the next post, this is a big problem. Each presents a different way of abstracting and simplifying the situation. Each is very incomplete, yet there is no combinatorial principle that provides completeness.