Hacker News Reader

Comment by joefarish | Hacker News Reader

I use the word "unsubscribe" to filter out a lot of emails. There are occasional false positives but after combining the above filter with a whitelist of senders it has proven to be very effective.

View on HN

I prefer to filter on the word "hurry", among other things. That seems to catch all the mail lists I don't want to hear from while keeping those I do available. I don't hurry.

I only filter them into a pre-spam folder and batch spamflag later, that way I don't miss anything hit by a false positive (e.g. a Signal v. Noise digest had the phrase "Why the hurry?" at one point)

Using this method I don't think I had a single black friday email hit my inbox last year :)

Oh and of course "Greeting(s?) of the day" has never been a legitimate email. You can safely auto-spam that phrase.

That's why it's now "if you don't wish to receive these emails, click here".

Then, as mentioned in the article, the you could filter on the word "click" too.

The whole point of these Bayesian filters is that you don't have to spend mental cycles thinking of clever words to filter on. Just mark 'em spam and not spam, the algorithms are plenty good at finding which words are spammy or legitimate without human help.

I believe this raises the issue of short-term solutions (such as filtering on 'click') for which spammers will eventually find turnarounds vs long-term solutions. But do we need really need long-term solutions if an update of filtering terms every year is enough to keep the pace? Also, does a long-term solution against spamming even exist? Since the Graham solution from 2002 doesn't seem to have been implemented yet, the answer is no?

Another issue is the carbon/energetic cost of spamming. I'm no expert, but given the probably high figures, should this issue be escalated to a higher level, e.g. an international agreement along with heavy penalty if caught running a larger spamming farm? Compared to the difficult issues of drug trafficking etc. I don't see how there could not be a consensus for spam.

Reply on news.ycombinator.com