ietf-secretariat(_at_)ietf(_dot_)org wrote:
...we are planning to turn on SpamAssassin on all IETF mail...
I have serious concerns about the use of spamassassin to filter IETF
mail, but it depends hugely on the details.
If the secretariat is just tagging mail, I don't have a big problem with
that.
If the secretariat is proposing to use spamassassin as a way to
implement a few very carefully chosen filters, that's fine, provided
those filters are actually reliable indicators that a message is not
appropriate for the IETF list. For instance if IESG were to set a
policy that all messages are plain English text (I'm speaking
hypothetically here), then a filter that excluded other content-types
and other languages would be defensible. If IESG were to set a policy
that says messages should have valid return addresses, then a filter
that checked return addresses for validity would be defensible.
In all cases the filters should be well-documented and subject to public
review prior to implementation, not something imposed at the whim of the
secretariat and/or IESG.
If the secretariat is proposing to use SpamAssassin's default filters,
or even very many of the filters included in SpamAssassin, I would have
serious objections to that. Many of SpamAssassin's criteria are
completely without technical justification, and it is not unusual for
SpamAssassin to block legitimate mail (as it did in that example I cited
a day or two ago).
Even if SpamAssassin only blocks one out of one hundred legitimate
messages, it is still unreasonable to impose a significant barrier on
that one poster. SpamAssassin is not a good judge of what is correct or
reasonable, and we have had too much arbitrary censorship already. It's
hard enough to contribute usefully to IETF without imposing the
additional burden that contributors second-guess SpamAssassin's filters.