Read my lips


Daily Kos has a good article about the NarusInsight, a little box from AT&T that can monitor the contents (the semantic content, the meaning) of internet traffic at ten billion bits per second. Deep stuff.

Just to help you understand how it works, try Googling something, anything, just like many of us do every day. Look over on the right, at the “Sponsored Links” section. Google reads the content of your search, and uses its text recognition software to understand what you are asking for, and responds with advertisers who are offering what you are looking for. Many of you now have Gmail, Google email accounts. Read your mail. There it is again, those “Sponsored Links”. Yes, Google is reading the contents of your mail, analyzing it, and responding. That’s how they make their money offering you free services like free searches and free email. They read and understand everything you read and write, and respond. Advertisers pay good money to Google because they believe Google does an excellent job understanding the meaning of everything you write.

And that’s what the NarusInsight does. It reads everything you read or write on the internet, understands the meaning, and responds by sending “interesting” stuff to the NSA spy agency.

AT&T is not developing expensive equipment that can process ten billion bits per second out of charity. If it makes it, it will be purchased and used. Whether anyone tells you or not.

So what about when you talk on your telephone or your cell phone? Well, AT&T is right there, and has been for decades. Let’s see what they are doing right now:

Accurate recognition of spontaneous large vocabulary telephone speech is of vital importance to AT&T in numerous applications such as customer care and information retrieval. We developed a system for Large Vocabulary Conversational Speech Recognition (LVCSR) which for a third year in a row had the best performance of any industrial lab and finished second overall out of seven participating systems in the recent NSA sponsored and NIST administered Speech-To-Text (STT) part of the RT-02 Evaluation. Also, AT&T was the only participant that managed to field a system running at faster than real-time speed.

The annual evaluations aim to evaluate speech recognition technologies on difficult problems involving spontaneous conversational speech over the telephone, including cellular in different environments (office, street, moving vehicle).

So now we have the technical ability to analyze every bit of text that flows across the internet. And we have the ability to turn everything spoken into a telephone, landline or cellular, into text. That’s pretty much everything folks.

Analyzing text has been around for a long time. And turning speech into text has been around for a long time too. But it has been used in small scale commercial operations. Now we finally have the ability to do it on such a grand scale that it becomes possible not only to spy on those few people in the world who are terrorists, but to spy on every human on earth who might conceivably be a part of your political opposition.