This is a follow up to that 13 February Spinney piece on predicting the future, DARPA as Poster Child for Out of Control Budget. It belongs to the category of “we don’t make this stuff up.” Speaking of this category, I think this latest incident in Pakistan (CIA contractor shoots two Pakistanis in broad daylight, CIA SUV going to his aid runs down a Pakistani motorcyclist) ought to be a signal to disestablish CIA.
Phi Beta Iota: FATAL FARCE just keeps on growing.
Algorithm: in computer terms, a finite set of coded instructions directing a computer to solve a specific problem or execute a specific process.
The term ‘algorithm’ is a prosaic word that has taken on an almost occult meaning to hosts of middle and senior managers in the intelligence and military sectors of the U.S. National Security Establishment.
It appears that anyone claiming to have developed an ‘algorithm’ to solve any of the many issues facing those sectors will find a receptive audience and usually a wad of cash to pursue development of that program.
For years the U.S. Intelligence Community (IC) has vainly searched for an algorithm that would replace or at least supplement the research and analysis now done by humans. Another goal is finding an algorithm that would replace human translators and transcribers of foreign languages (commonly called “machine translation/transcription”). The problem of course is that the folks looking for these cyber panaceas usually have not actually reflected much on what involved in the cognitive processes that they are trying to mechanize. This has led them into all sorts of dubious and often costly projects.
On 13 February, Chuck Spinney called the attention of this Journal’s readers to what correctly identified as a lunatic DARAPA project paid for by the U.S. Military and the IC. It is to develop an algorithm to predict the future, specifically to predict outbreaks of political unrest. To date $120 million dollars has been spent on this project which has involved contractors from Lockheed Martin and is rapidly moving from the design to production phase. The problem of course is that the government folks who are sponsoring this project have conflated warning intelligence with prediction of specific events. As Spinney observed it is mathematically impossible to predict specific events because until they happen they do not exist and constitute only one of an infinite variety of outcomes. Warning intelligence is another matter entirely and indeed is a core element of strategic intelligence. For example President Obama was given strategic warning intelligence in August 2010 of growing unrest in the Magreb that could lead to political upheaval. This is what intelligence can and should do. It can never be the Oracle of Delphi.
An even more bizarre example of the search for omniscient algorithms was reported on 19 February in the New York Times. The U.S. military and intelligence communities over an eight year period provided over $20 million dollars to what now appears to be a con man who claimed to have developed algorithms to spot al Qaeda stenographic messages hidden in al Jazeera TV broadcasts, identify al Qaeda operatives from UAV video images, and to detect noise from hostile submarines. In addition the same individual claimed to have an algorithm that could track and detect terrorist plots. In point of fact none of these algorithms actually worked. This con game was able to go on for eight years because the project was probably cloaked in “need to know” secrecy and the chances are no really competent government systems analysts even heard of the project let alone vetted it.
Both these examples speak volumes for the level of ignorance, incompetence, and lack of integrity which seems to prevail in so many parts of our national security establishment.