There has been some coverage in the news lately about the NSA/Bush Administration efforts to "wiretap" Americans. I write to clear up some of the mythology surrounding this subject.
First, a word about my qualifications to speak on this subject. In 2002, after my retirement from a 27-year career with the U.S. Information Agency, I was hired by a small company developing advanced software for the U.S. Army’s Intelligence and Security Command (INSCOM). My role was to use the new systems and help evaluate and improve them. In 2005, the commander of INSCOM, Gen. Keith Alexander, left for a new position as the commander of the NSA, and the small company, Object Sciences Corp., was purchased by SAIC. The research we were conducting was terminated in favor of shipping large data storage and processing hardware to overseas bases, and no one in SAIC had the imagination or desire to find a way that a non-engineer, non-retired military person such as myself could make a contribution.
When reading and hearing about the NSA programs that are under discussion, I recognized the programs in which I was involved, and knew that Gen. Alexander had brought those programs from INSCOM to implement at NSA, along with his advisor, Dr. James Heath. I have been reluctant to speak up about them for several reasons:
- Much as my release from SAIC was a bitter experience, I did not wish to be seen as just another disgruntled former employee. I was not let go because I held strong opinions about the programs, other than the belief which I will talk about below that the systems did not work. I was also reluctant to damage my chances of finding employment with another contractor/agency doing exciting research that was such a challenge with OSC. In the last two years, however, I have come to believe that SAIC or someone has placed a "do not hire" flag on my security clearance file, and that is what has happened.
- I had a legitimate fear that I should not compromise methods and sources that have real value. I am not like the former Vice President in feeling that the interests of the U.S. are secondary to my own ambitions. That fear no longer applies, as the existence of the program to record and store phone calls, e-mail, and financial records is now public knowledge.
- My own role and status within the organization was marginal and minor. I was aware that a convicted felon – Admiral Poindexter – was not only allowed to enter highly secure facilities, but was guiding its activities. But I never had to deal with the problem of directly dealing with the man. I was aware that there were vast stores of data on overseas communications, but not on the domestic programs. As small as my role was, however, it gave me an insight into and understanding of the subject that is missing from the reporting that I have seen.
- I understood, in a way that puts me in the minority of my former colleagues, and was part of the reason I was let go, is that the big untold story of the data collection and mining systems is that they do not work, and I said so.
We have to understand the stated goals of these programs. As Dr. Heath stated one time, quoting Gen. Alexander, they wanted to build a computer system that would allow an intelligence analyst to "look at" a billion documents at once. To me, the statement was both meaningless and misleading. It assumed several things that are not true:
- Text documents are the same thing as numerical data. In a computer, text documents are converted into numerical code as a matter of course. To a computer, there is no such thing as text or language; everything is numbers. But that very statement means that a computer cannot understand the "meaning" of a text document, and sort out documents with similar or different meanings from a vast sea of similar data.
- Data mining and its cousin pattern recognition are myths. As much as there are a library worth of books and articles on the subjects, as much as entire careers have been built around the concepts, these system do not actually exist. There is a very real possibility – I believe it to be the case – that they never will. There are approaches, there are very clever work-arounds, there are extremely useful and ingenious failures; but the fact is that a computer that can sift through thousands of documents or voice files and find not only what you are looking for, but also what you are not looking for, but should be, is not on the horizon. Anyone who has used the most sophisticated search system in the world today, Google, can understand that.
- That is the real danger with these systems. People will pretend that they work, and think that they work, and take actions based on this assumption, that will damage the lives of innocent bystanders. In the meantime, the real work of gathering real information is ignored.
There was a hint of this in the Olbermann interview Thursday night. Much of what passes for data mining is really simple (or even not so simple) key word searches. The problem is, the people we are looking for are aware of that, and have taken to using simple codes (soft language) that avoid the obvious words we can think of. Thus, a conspirator asking whether the equipment is in place for an attack will say: "Are the tools there to build the house?" The problem then becomes that intelligence agencies will look for these code phrases, which is where a legitimate building contractor stands in danger of being listed as a terrorist.
Our intelligence agencies are staffed by serious professionals, who are doing the best they can with the tools they have to find the enemies, provent attacks, and protect our safety. The real problem is that they are being led by political agents (such as Poindexter) with ideological agendas, short terms in office, and the desire to find a quick fix silver bullet solution that does not exist.
Yes, it is a crime and a Constitutional travesty that all this data is sitting on the data banks at the NSA. The good news is that there is so much data, and the so-called data mining techniques used to look at it so primitive, that there is little danger that anything, good or bad, will ever be done with it. The bad news is that all the money goes into useless enterprises like this, when the actions that would provide us real security are not taken.