On Wed, May 14, 2014 at 09:03:50AM -0700, jim bell wrote:
> I guess I'm still not being clear. It would be my way of objecting to a court's ordering the telecom company that
>I might work for (or, one day, that I might own?!?) to present an "electronically-readable" form of the telephone >metadata of millions of telephones. The judge
ordered that; my sneaky response would be to generate an >"electronically-readable" file, basically a pdf file or a series of same, itself with an image that looks like >"captcha" information: relatively easy for a human to read, but rather difficult for any computer to turn into >easily-useable (searchable) information. In other words, the information would be presented to the NSA, but it >would be essentially unuseable without being (first) human-decoded.
>assuming this is correct:
>
http://googleonlinesecurity.blogspot.de/2014/04/street-view-and-recaptcha-technology.html>then googlestreetview tech is better at solving captchas than humans.
For a single, tiny
piece of "captcha", that might very well be true. But suppose the telephone metadata information for a billion phone calls per day is turned into "captcha's". How much CPU power would the NSA have to apply, each day, just to back-convert that metadata into computer-searchable form? Admittedly, that's irrelevant: The NSA would simply ask the court to order the company to stop being a wiseass, and to stop using the captcha technique.