On Thu, 9 Oct 2003 13:13:03 +0100,
Jonathan Morton <chromi(_at_)chromatix(_dot_)demon(_dot_)co(_dot_)uk> wrote:
I understand someone figured out the vast majority of spam comes
from web scrapers - these don't know what kind of group the
addresses are coming from.
Minor quibble: When last we heard, best evidence suggests that most
spam comes from Usenet scraping, not Web scraping. We should have an
even better answer to this question in a few months' time.
Speaking of addresses... (clever segue)
I recently tried to inspire some interest in a trap-address
maintenance effort. Only three people responded (my sincere thanks
to all of you--you know who you are). Unfortunately, only one of
those three was able to commit the level-of-effort necessary to
launch the project.
So, I'm donning my "design curmedgeon" hat again, and offering the
following observations (and that's all they are...observations):
* There is very strong evidence that spam
volume is affected by at least one (and
possibly many) uncontrolled systematic
variable(s). While not truly "noise" in
the traditional sense of the word, it is,
nonetheless, statistically 'bad' variance.
Estimates regarding the size of that
systematic 'bad' effect range from
factor-of-6 to factor-of-10.
* Sadly, the "otherwise identical pairs of
addresses" approach I advocated many weeks
ago will *NOT* serve as an effective
"counter-measure" to this extraneous
variance.
* As a result, any ASRG study that requires
"active spam-seeking" to achieve its
research goals has an extremely low
probability of achieving statistical
significance, regardless of researcher
effort/activity.
* The off-list correspondence regarding trap
addresses has already spawned at least one
"passive" (i.e., archival data)
characterization project. It's easy to
envision several high-payback research
efforts in this venue. Anyone who's
interested in possibly contributing to
analysis/characterization of archival
data is invited to drop me an off-list note.
- Terry
_______________________________________________
Asrg mailing list
Asrg(_at_)ietf(_dot_)org
https://www1.ietf.org/mailman/listinfo/asrg