Skip to main content

View Diary: 99.80% Error rate! NSA Leaks destroy legal defenses of wiretaps! (93 comments)

Comment Preferences

  •  Text consists of patterns (none)
    It's not significantly difficult to use a computer to identify a language from a small sample of text. I've done this using simple modules written in the Perl programming language. My job at the time was to cull out anything but English, but it's not any more difficult to watch for any of a few dozen languages and dialects and run different key words analyses on those.

    That said, I think they would definitely watch English language traffic as well, expressly because there are so many languages spoken in the Muslim world. The lingua franca in cases where people of many nations need to communicate is often English. In Muslim countries it would more likely be Arabic, though....

    But you're right in your implicit assumption that it is not the NSA's job to try to catch domestic terrorists (i.e. those speaking vernacular US English). That task resides firmly with the FBI.

    •  Correction (none)
      It's closer to about 30 different dialects of Arabic, not just 8.  
      •  One would need (none)
        to toss all the major European languages into the mix as well. Esp. french.

        I don't really see this as a big problem anyhow. It's all 1's and 0's to the hardware and what are American ctizens paying all those nerds at the NSA for anyway if not to capture, decrypt and translate foreign language communications?

Subscribe or Donate to support Daily Kos.

Click here for the mobile view of the site