email addresses: @-sign needed

Brendan McKay bdm at cs.anu.edu.au
Sat Apr 19 13:01:04 CEST 2003


This site:
    http://www.cdt.org/speech/spam/030319spamreport.shtml
describes an experiment in spam.  One of the findings is that
even human-obvious obscuration methods are highly successful.

This can be expected to remain true until obscuration becomes very
commonplace and standardised due to being provided by web authoring
software.  Only at that point will spammer's care about the miniscule
fraction of the addresses out there that are obscured.  Even then,
the defence will be to use an obscuration method that is unusual.
Trying to authomatically understand lots of different obscuration
methods is a hard computational problem.  At the moment I can't
imagine why any spammer would bother.

Brendan.


* Brendan McKay <bdm at cs.anu.edu.au> [030419 20:48]:
> * N. J. A. Sloane <njas at research.att.com> [030419 20:34]:
> > 
> > Afterr thinking about the problem some more,
> > it seems to me that the simplest solution of all
> > would be the best:
> > 
> > just replace joe at broken.hill.com
> > 
> > by
> > 
> > joe _AT_ broken _dot_ hill _dot_ com
> > 
> > for example
> > 
> > As you once remarked, i think, that will defeat 99% of
> > the email harvesters, and will be much easier to implement
> 
> I wonder if even that level of disguise is needed.  Domain names
> are not really sensitive in the great majority of cases, so
> perhaps it will suffice to use
>    joe _AT_ broken.hill.com
> or more graphically
>    joe(AT)broken.hill.com
> 
> Brendan.





More information about the SeqFan mailing list