Err, no. That's not how SpamAssassin works. It works based on
looking at characteristics of mail messages, to see if they "look"
like Spam. It's what many of us do automatically. For example if
there is a HTML only body, and no text, I'm very likely to hit the 'D'
key and go on, unless there's some very good reason why I think it
might not be spam. Similarly, if the character set is Korean or
Chinese or Japanese, I will also assume that it is spam, and likely
hit the 'd' key very quickly.
And here goes the baby with the bathwater: a lot of our Japanese
colleagues send mail to IETF working groups with:
Content-Type: text/plain; charset=iso-2022-jp
This is a very natural set-up: they select "plain text" in their local
mail software, and they type their messages in English. They should be
fine, right? Well, not according to your rule. They would also have to
go tweak the preferred charset, a much harder proposition. I guess the
problem is the same from pretty much anyone who has to use a different
alphabet than ASCII.
The real reason why the IETF should study the countermeasures is
precisely that: it is damn easy to write spam filters that have many
"false positives". We need careful engineering, not slash and burn
tactics.
-- Christian Huitema