spamassassin-users January 2011 archive
Main Archive Page > Month Archives  > spamassassin-users archives
spamassassin-users: Need Volunteers for Ham Trap

Need Volunteers for Ham Trap

From: Warren Togami Jr. <wtogami_at_nospam>
Date: Tue Jan 18 2011 - 06:52:58 GMT
To: users@spamassassin.apache.org

Hi folks,

Here is an opportunity for non-developers to do simple tasks to help
improve Spamassassin.

I am seeking volunteers to help me build and administrate a "ham trap".
  The idea is to subscribe a list of unique e-mail addresses to various
retailers, airlines, government and other legitimate bulk mail senders.
  A sufficient variety of ham trap subscriptions should increase the
variety of legitimate senders represented in nightly masscheck and thus
improve the safety of Spamassassin's rules.

Benefits of the Ham Trap
========================
* Creation of an automated, synthetic source to build a corpus of very
recent ham for the nightly masscheck. Ham trap data will be expired
from the masscheck after 3 months. This will be fairly easy to maintain
in a 99% automated fashion, ensuring a constant stream of fresh data for
the nightly masscheck largely without the need for human sorting.

* Help to identify legitimate bulk senders who are performing poorly
with spamassassin. Our data may help legitimate senders to modify their
mail practices to avoid spamminess.

* Each subscription is a unique tracked address. This will make it
possible to definitively identify bulk senders who violate their
customer's privacy by selling their e-mail address list to others.
There isn't much we can do about these cases other than shame them on a
web page, but for spam fighters this is useful information.

Examples of Subscriptions
=========================
* Online Retailers - offers, coupons, etc.
* Airlines - offers, fare alerts, etc.
* Government - newsletters, emergency alerts, street sweeping reminders,
etc.
* Non-Profits - newsletters, begging for donations, etc.
* Announce Lists are good.
* Discussion Lists are a BAD IDEA.
* Some subscriptions have the option to opt-in to "partner" offers.
Always decline these.

Current Retail Subscriptions: llbean.com, lapolicegear.com, zappos.com,
armaniexchange.com, gucci.com, ralphlauren.com, jimmychoo.com,
coach.com, chanel.com, louisvuitton.com, prada.com, diorcouture.com, ysl.com

I will have a website soon listing all current subscriptions so we can
avoid duplicates.

Volunteer Tasks
===============
Volunteers live in different states and countries, so they are exposed
to different regional and local subscription opportunities. For this
reason we need input from many different people to make the ham trap
representative of the world.

* EASY: Simply prepare a list of reputable sites you think I should
subscribe to. Local government/companies in your area would be
especially good. Myself or a volunteer will handle it.

* EASY: Using a list of e-mail addresses, subscribe to reputable sites
and record each site to the corresponding e-mail address. This requires
no access to my server although I need to trust you to keep accurate
records and keep the e-mail addresses secret.

* MEDIUM: Read the ham trap inbox. Write procmail rules to
automatically sort certain known senders to a specific address as Ham.
Manually sort past unclassified mail as Ham or Spam. This requires IMAP
and ssh access to my server. This group will be small due to the level
of required trust.

Procmail rules ensure that ham corpus mail actually came from the
desired sender. All else is put into a separate Unsorted folder for
human identification. The Unsorted folder becomes smaller and the
process becomes more automated over time as the procmail rules are crafted.

Please let me know if you are able to help!

Warren Togami
warren@togami.com