Slashdot Mirror
Computer Software to Predict the Unpredictable
Amigan writes "Professor Jerzy Rozenblit at the University of Arizona was awarded $2.2Million to develop software to predict the unpredictable — specifically relating to volatile political and military situations." From the article:
"The software will predict the actions of paramilitary groups, ethnic factions, terrorists and criminal groups, while aiding commanders in devising strategies for stabilizing areas before, during and after conflicts.
It also will have many civilian applications in finance, law enforcement, epidemiology and the aftermath of natural disasters, such as hurricane Katrina."
...the program will still fail to predict it. By definition.
The article (as would be unsurprising even from the professional press, and is less surprising from what seems to be a school newspaper of the school employing the professor getting the grant) seems to be a very uncritical regurgitation of an extraordinarily puffed-up press release that seems to suggest that the professor has gotten a grant to develop something that already exist and presently has the capacities sought by the grant. Sometimes. Maybe. Really, the shifting use of verb tenses gave me a kind of mental whiplash trying to read it.
Also, I think that while this may be useful, the danger of overreliance on a system where quite literally no one using it understands how factors are really being used to generate outcome predictions are immense; if you get something that works well predictively at all, it will likely be prone to fail wildly if any of the many factors it is adapted to based on the historical data used to train it shift. Unfortunately, it is quite likely that the particular sensitivities will be opaque, and thus no one is likely to know when it is likely to fail. This is rather distinct from conventional analysis which, even though it may fail in many circumstances, where it is rigorous analysis and not just guesswork to start with, its assumptions are transparent and its weaknesses and vulnerabilities in application to particular situations can also be evaluated.
"Did you know that the first Matrix was designed to be a perfect human world? Where none suffered, where everyone would be happy. It was a disaster. No one would accept the program. Entire crops were lost. Some believed we lacked the programming language to describe your perfect world. But I believe that, as a species, human beings define their reality through suffering and misery. The perfect world was a dream that your primitive cerebrum kept trying to wake up from. Which is why the Matrix was redesigned to this, the peak of your civilization. I say your civilization because as soon as we started thinking for you it really became our civilization which is of course what this is all about..."
-- Agent Smith
I disagree. A chaos predictor cannot be more accurate than simple guessing, for it is attempting to predict chaotic events. Look at the political scene in the US; given the current climate everyone is attempting to say that the Democrats are going to take the election (and they might). It is just as likely that one of the Republicans will nose over the line (especially with the right third-party candidate). We know his system well and experts can only call a race with 50% accuracy, even with just two candidates in most races. The human mind is far more complex and makes far more calculations, with random associations for a computer to predict anything about what a group of people will do. I will be impressed if we see a program, ever, that hits greater than 20% reliability on predicting events or trends.