Editor's Note: To me, the most interesting part of the above linked article is the explanation of how modern surveillance software and command & control technology works. That's the part I've excerpted below.
-Matt
. . . [this sort of software] combines datamining, artificial intelligence, and "interoperability," the capacity for one program to read, operate, and modify the source codes of other programs. Datamining is a technique for detecting and extracting meaningful patterns hidden within vast quantities of apparently meaningless data.
Programs based on datamining are powerful analytical tools; finding meaningful patterns in an ocean of information is very useful. But when such a tool is driven by a high-caliber artificial intelligence core, its power gets spooky. The datamining capability becomes a smart search tool of the AI program, and the system begins to learn.
In recent decades, great strides have been made by the mutually fertile disciplines of mathematics, computer science, and neuroscience. Among the results has been a new discipline called cognitive neuroscience, which constitutes a powerful new understanding of the way the human brain works (see Churchland, Gazzaniga, etc.) While this has illuminated some very fundamental and grand issues of philosophy, it also has applications so practical that they have reshaped our world. "Neural Network" programming is modeled on the computational techniques used by the human brain - an electrochemical computer that uses neurons instead of semiconductors; the firing or non-firing of neurons instead of ones and zeros.
With neural networking, software has become much smarter than it had been. Now it can perform multiple, related operations at the same time through parallel processing; now it can learn from setbacks, and use genetic algorithms to evolve its way out of limitations. Now it can respond to more kinds of data from the electronic environment, including "fuzzy" values that don't come in discreet numerical packages. This kind of computational power supports an inference engine that can digest the mined data into results that are not only descriptive of the system's present state but predictive for imminent and, to some degree, even middle-term outcomes. That's why the same family of programs that does enterprise architecture, which is descriptive (and prescriptive if you take its descriptions as a mandate for cutting costs by firing people - "process management"), comes to include risk management software, which is predictive of the future. It extrapolates from current trends in a more than quantitative way.
Conventional electronic surveillance finds patterns in the data of other instruments; [this sort of software] can exploit the patterns it detects and extrapolate future probabilities. Then it can integrate itself with the computers from which it's getting the information and intervene in their functioning. The result is a tool for surveillance and intervention.
Related: