Path: utzoo!mnetor!uunet!husc6!bloom-beacon!mit-eddie!apollo!nelson_p@apollo.uucp From: nelson_p@apollo.uucp Newsgroups: comp.society.futures Subject: long distance monitoring Message-ID: <3b8abbf2.44e6@apollo.uucp> Date: 18 Apr 88 20:44:00 GMT Sender: user@apollo.uucp Lines: 33 To: comp.society.futures@news >In _The_Puzzle_Palace_ Bamford asserts that all long-distance >communications in and out of the country are indeed monitored, >and cites his reasons for asserting this. Crimethinkers have >always taken comfort from the fact that there is simply too much >traffic for human snoops to completely monitor, but .. > >Bamford also says that the NSA tries to stay 5 years ahead of the >state of the art. The "state of the art" is that machines are >very close to recognizing (and so transcribing) connected speech, >or at least keywords. What does this mean, 'tries to stay 5 years ahead of the state of the art'? It sounds like a contradiction in terms. Right now there is a a great deal of research going on in speech recognition both in academia and in business. IBM has had a major research program in this area for some time. There's a $bundle to be made in this by whoever comes up with something decent, first. Are you saying that everybody is just wasting their time because the government can already do this? And they're too stupid to know it? Are you saying that the government can keep a secret of this magnitude? Are you saying that they could attract the research talent to pull this off and still keep it quiet? Also, Bamford's book was published in September of 1983. His info had to be at least a little older than that. Yet here it is in 1988 and we still can't process connected speech of arbitrary speakers with much reliability. And then consider how many tens of thousands of overseas voice channels there are. There are not 1/10th that many Crays in the whole world. Count me as skeptical. --Peter Nelson