[Tfug] [Bulk] Re: Stallman vs Ubuntu ryryma1a

[Bexley Hall]

Hi Keith,

On 12/14/2012 7:40 PM, keith smith wrote:

> You will probably want to read this
>
> http://www.wired.com/threatlevel/2012/03/ff_nsadatacenter/

I'm not overly paranoid about Big Brother.  Some of this
activity (define "some" as an unbounded value beginning at
0%) is the only *practical* way of ensuring safety, etc.
Ideally, systems with NO false negatives *or* positives
would be developed (for their goals).  And, I am willing
to tolerate some amount of "false positive" actions (e.g.,
the "meritless" patdowns) if it keeps the number of false
*negatives* more reasonable.

But, BB, for the most part (resource limitations), doesn't
peek in your bathroom window to see who's using the facilities
at a particular time of day (or night).  Chances are, that
sort of data -- even if easily available -- would just be
excess "noise" in their analysis.  (Do "bad folks" pee more
often than "good folks"?)

OTOH, a pharmaceutical company looking to peddle its products
(or a health insurer looking to limit its exposure) would be
far more interested in that sort of thing.

[E.g., I don't think BB cares if you eat a lot of hamburgers!]

The other problem is errors in data tend to have a higher
bar to cross for the sorts of things BB is watching -- they
don't (can't?) want to run around pursuing thousands of
"false positives".

OTOH, bad data in *commercial* environments has a low threshold
to cross before being acted upon/reacted to, etc.  Sending
solicitations to a male with a female-ish name, etc.  There's
just no incentive for the company to fix the problem... the
cost to the company is too low (easier to let the mailings
continue to go out than it would be to fix the problem).

[Contrast this with the cost to BB to do in-depth investigations
of lots of false positives]

If obtaining the "data" is an expensive undertaking, you place
more *value* (perhaps incorrectly) on that data.  And think a bit
harder before acting on it.