Drumbeat project idea: Spam Warnings, "Firefox Phishing Warning"-style
hansschmucker
ideas, not an application (frankly, because I don't have enough time
to do it now)
-----------------------------------------------------------------------------
=== Collecting social spam and hitting spammers where it hurts ===
On social sites, there's a lot of noise. And a lot of it is spam.
Fighting spam so far is mostly about identifying it and blocking it,
but the awful truth is that this doesn't hurt the ones ordering the
spam.
== Using spam data to provide detection and hurt spammers ==
The basic idea is simple: Use the data from messages marked as spam to
provide a warning list, both for service providers looking to
automatically remove spam as well as end users trying to estimate the
trustworthiness of a website.
== Hurting the spammers ==
At the end of the day, spammers want to drive traffic. Make identified
spam post hurt their traffic and you've got them cornered.
= The addon approach =
As usual, a Firefox addon provides a great way to test drive the idea.
Maybe even with an already existing spam list provider as a starting
point. Investigate if people are interested in the spam activity
surrounding the page they are visiting. Is myprojectorshop.foobar
trustworthy? Spam volume can give you an idea. I know I'd install such
an addon on most of the computers in my family, but what do others
think?
== Collecting spam data ==
Machines can only do so much to identify spam. At the end of the day,
the most reliable source of spam information are users and moderators.
They can very reliably determine if a link is posted as part of a
conversation, or just for the sake of posting the link.
= Getting communities involved by giving back =
There's a very simple way to get service providers to work with us:
Give something back: A blacklist that is far more reliable than any
automatic spam protection can be.
*Integration
Of course, if it's a hassle to use the service, then nobody will: An
integral part is a stable API and plugins for common community
software.
*Motivation
Handing out badges, medals and the occasional reward for providers of
high-quality spam reports doesn't hurt either.
*Education
False positives can really be damaging in such a system. Educating
moderators so that they don't report posts that they simply want to
see deleted is essential.
*Trust
Making sure that only trustworthy services, i.e. only ones that we
know are not run by spammers themselves is essential. ( TODO No idea
how yet, besides manual verification which may be discriminatory)
*Technology
Of course, solid technology will be essential as we'll have to analyze
links for the actual target of the spam, which may be protected by
links, redirects or hiding among legitimate links. Analyzing those
redirect-pages, possibly using a spider to periodically recheck them
is far from trivial, but should be possible by cross-referencing with
the data from direct links.
== Pitfalls ==
= Legal =
One thing that's for sure is that the addon approach can be very
dangerous. Removing spam doesn't usually cause a lot of uproar, since
by blaming anybody for its removal, you admit to doing it. But if you
warn a user about visiting a certain page, you'll get a lot of
complains that you harm their business. Finding ways to warn the users
without bringing any legal harm to yourself is tough and should be
remembered.
= Spammers provoking false positives =
What would you do if you were a spammer and you'd see such a list gain
mass-market adoption? Well, you'd probably try to discredit it by
adding false positives for popular services.
= Spammers simulating spam to harm a competitor =
This isn't something you'll have to think about right away... not
until the harm caused for a spammer by being on the spam list is
bigger than the benefit of spamming. But then it might become a common
problem: What better way to get rid of a competitor than making him
seem untrustworthy.
-----------------------------------------------------------------------------
Again, it's just a suggestion. But at least it's a suggestion that
could produce a product that I would be interested in as a user.
--hans
pa...@appcoast.com
project since a general solution to the problem needs to be worked out
and the immediate benefits are a better web for everyone.
Perhaps initially we could setup this project on the wiki and then point
Dries @ Mollom to help us articulate the spam problem further ?
Best,
Paul Booker
> _______________________________________________
> community-drumbeat mailing list
> community...@lists.mozilla.org
> https://lists.mozilla.org/listinfo/community-drumbeat
>
