start mutt without having to see bad messages

annoying.filter

<limit>
!(
~s token ||
~s reward ||
~s data ||
~s feedback ||
~s news ||
~s free ||
~s crunchbase ||
~s "!!" ||
~s integrate ||
~s save ||
~s alert ||
~s free ||
~s marketing ||
~s daily ||
~s views ||
~s soon ||
~s trial ||
~s best ||
~s soon ||
~s rapid
)
<enter>

Starting mutt with mutt -e 'push "<limit> !(~s free || ~s "!" || ~s news || ~s alert ) <enter>" ' will start mutt without annoying messages.

I thus define a bash function Mutt() { mutt -e 'push "<limit> !( ~s free || ~s "!" || ~s "deep learning" || ~s marketing || ~s daily || ~s data ) " ' ; } in ~/.bashrc (and ~/.zshrc)

and then start reading mail with Mutt instead of mutt.

What are some other good filter words?

Alternately T ~s free and then D will delete any of these patterns from the mbox.

PS I’m aware that Bayesian methods can generate signals like this. The point is to see the filter typed out, whether you generated it mathematically or personally.