Re: [syndication] RSS errors due to template mistakes

[Rick Bradley <roundeye@roundeye.net>]

* Bill Kearney (ml_yahoo@ideaspace.net) [030307 11:18]:
> So it falls back to the question of what (and where) to ask the feed providers
> to add to their content?  I'd prefer to avoid having them add anything
> service-specific.  It's hard enough getting cooperation out of people without
> making them think you're trying to "own" them.  If anything an MD5 hash of
> whatever address they'd be willing to use would work.  They could hash it on
> their own.  Anything that already knows their address could cross-reference it.
> If the remote service already knows their address then the hash would match up.
> The trick continues to be getting the users to start the ball rolling.

All good points.  As is normal, getting user adoption is the biggest
issue.  Having a few providers agree to start using contact-hashes might
start the ball rolling.  There will always be the worst cases where
there really is no contact info available, but at least the problem
could be mitigated somewhat over time.

> > We made our code available [1] under a BSD license so anyone can use it,
> > and we even documented the process of setting up full harvest-proof
> > hypermail archives. [2]
> 
> That's a nice bit of code.  I'd wonder, however, whether spammers would OCR
> those pictures?  I suppose if they're willing to do to that much effort there's
> not much more you can do.  I raise this question because many free mail sites
> now use visually obfuscated signup screens precisely because of auto-signup bots
> reading the images.  The arms race continues....

So far there's been no indication that list-only addresses have been
harvested, but it would indeed be possible to OCR the images (it's also
possible to subscribe to the mailing list ;-).  It's only a protection
in degree, but possibly due to the infrequency of the use of this exact
tactic no harvesters seem to have OCR'ed the archives so far.  I'm sure
it's only a matter of time (and the more people who use this code the
more likely it is that someone will hack an OCR-bot for it), but the
value provided by having the archives online has been significant.

Rick
-- 
 http://www.rickbradley.com    MUPRN: 559
                       |  in skin tones, and
   random email haiku  |  that's something I'm kinda
                       |  picky about. Heh.