Monday, November 16, 2009

100% sales job

OK, so back in May I stumbled upon a link to this TED presentation from February 2009, TED Talks: Bruce Bueno de Mesquita predicts Iran's future What aroused my curiosity was the teaser that said "3 predictions on the future of Iran, and the math to back it up." Hmm, sounded promising, so I checked it out and found (in no particular order):
  • Although this was supposed to be based on a mathematical model, it more or less was a "black box," and based on some super secret algorithm...
  • It seemed rather unlikely to this Tweeter that the number of people with influence over the Supreme Leader's decision, 87 per the presentation, would be that much greater than the number influencing the President's.
  • BBM's asserted that the 'Quietists' "see Iran going in an unhealthy direction, contrary to what Khomeini had in mind..." Hmm, really, so he believes that the quietists are upset at deviations from the Khomeinist line? Tell that to one of the most senior of the mujtahid, Grand Ayatollah Hussein Ali Montazeri, who broke with Khomenei! Or Ali al-Sistani, who differed with Khomenei re the definition of the doctrine of Velayat-e faqih. In fact, by definition the Quietists are those who (unlike, and in contrast to, Khomenei!) "hold that a cleric’s proper calling is not to rule, but to advise, teach and guide adherents according to God’s law as found in the Koran and the holy writings. This pure Islamic duty transcends involvement in politics." Hmm, this doesn't invalidate his overall point, though it sure doesn't do much for his credibility that he apparently isn't very clear re the facts on the ground...
  • BBM then provides this factoid re the accuracy of his model: "This is an assessment by the Central Intelligence Agency, of the percentage of time that the model I'm talking about is right in predicting things who's outcome is not yet known, when the experts who provided the data inputs got it wrong..." Apparently he is referring to this: "FACTIONS and Policon: New Ways to Analyze Politics" taken from "Inside CIA's Private World: Declassified Articles from the Agency`s Internal Journal 1955-1982" by H. Bradford Westerfield. Hmm, if you read the article, it is true that it says that Policon (BBM's model) forecast specific outcomes more often than the "regular" CIA analyses (60% versus 33%); and that it also did better at showing the political dynamics that lead to the outcomes (60% versus 35%). However, it then goes on to say that "forecasts done with traditional methods and with Policon were found to be accurate about 90% of the time." So, bottom line, "while traditional and Policon-based analyses both scored well in terms of forecast accuracy, Policon offered greater detail and less vagueness..." It also says that FACTIONS, the software developed internally by the CIA Directorate of Science and Technology, Office of Research and Development essentially did the same. OK, so very impressive, but hardly the same as the "What Percentage of Time Is The Model Right Even When The Experts are Wrong: 90%" slide in his presentation...
  • In the end his predictions ended up being rather bland, and he never really answered the question re the future of Iran's theocratic regime. Given that this talk was given barely four months before Iran's presidential election, it would have been a lot more impactful if he had 'predicted' the sequel to the election... the unrest, the "Green revolution," etc., etc. Apparently, when predicting this future he did not achieve the specificity touted by the CIA...
  • BBM ended up, "If you can predict what people will do, you can engineer what they will do, and if you engineer what they will do then you can change the world, you can get a better result." Hmm, OK, so this begs the question of why exactly the U.S. government, which has access to this great tool, seems to be unable to get the outcomes it wants in so many areas around the world...
  • Finally, given that this is supposedly based on mathematics, it seems strange that when he predicts outcomes he does not seem to provide any confidence limits. It would seem that mathematically-predicted outcomes would not be so categorical..
OK, so, all in all, a somewhat interesting talk... with no small amount of hyperbole in it. Subsequently this Tweeter has come across several laudatory articles re BBM's uncanny ability to predict the future e.g. see some of the links below. This Tweeter is not convinced re this supposed ability to predict situations, and its purported accuracy. What he is most impressed by is BBM's amazing sales ability and his great success in the marketing of himself!

Bruce Bueno de Mesquita: Learning the Methods for Making Accurate Predictions
Can Game Theory Predict When Iran Will Get the Bomb?
Brain food: can maths really let you see into the future?
If A Computer Model Says It, It Must Be True
The New Nostradamus
Reading Tarot on K Street
Recipe for Failure
A counter-opinion:
Criticism of Bueno de Mesquita's Theories

OK, so if you are willing to believe that mathematical models can predict the future with "90 percent accuracy" (or at least BBM's super secret black box model can!), you shouldn't have any problem swallowing all the hyperbole and claims in this following article... According to 'How Team of Geeks Cracked Spy Trade' a software program is unraveling terrorist networks, and also has "... helped root out terrorist financing networks, revealed new trends in roadside bomb attacks, and uncovered details of Syrian suicide bombing networks in Iraq, according to current and former U.S. officials familiar with the events. It has also foiled a Pakistani suicide bombing plot on Western targets and discovered a spy infiltration of an allied government."

With both these ultimate tools in its arsenal one has to wonder why our government is so singularly inept so often...

No comments:

Post a Comment

back to top