The Naïve Democracy and Social Justice of Bayesian Spam Filtering
Project Honey Pot is a web community that gathers evidence and shares information to fight spammers. I saw a complaint on their message board by someone who, for some reason or another, lost a month’s income because they “missed an online meeting with a client because of this system“.
A year or two ago, I remember reading about the case of one of the Internet’s most tech-savvy electronic newsletters being blacklisted by one of the giant online email services. Unfortunately, I forget which techy website newsletter and which email service. But, as far as I could gather, the blacklisting occurred because too many people chose to cancel their subscriptions by clicking the “mark email as spam” button rather than haul themselves through a ‘laborious’ unsubscribe process.
I remember having no sympathy for that banished techy-website, they really should have known better. I stopped sending out my e-newsletters as soon as I learned about how Bayesian Spam Filtering works. At the same time, I made a thorough effort to spam filter every single message sent from my domain.
Perhaps I was erring too much on the side of caution. But, I feel that it does not matter what you, the website owner, thinks. It does not matter what the law says. All that matters is the opinion of your users.
At the very moment that they receive an email from you, the majority decide to either :
- flag you as a spam sender OR
- happily read your email.
Very few might be so kind as to unsubscribe.
Now, if too many users flag you as a spam sender, then that mass of subjective opinion goes up a level and the big email networks start to score you as a spammer, automatically, on behalf of their users. Your future emails will either end up in a black hole or, if you are really lucky, the recipient’s spam folder.
So, in my opinion, one has to think very carefully before you send out an email from an IP / domain name that is valuable to you. Is this a message from a boy who’s crying wolf, or is it essential information that your users will know they need?
The users really have no interest in the mechanics and side effects of spam filtering techniques. They are judging you behind a Veil of Ignorance. They are blindly injecting their true opinion into the collective intelligence. Its a truly democratic, black or white, personal, split-second vote about whether you are a sender of ‘good things’ or ‘bad things’.
I think that the more that users become aware of the power they hold over your website’s spam rating, the less true their ’spam or ham’ vote becomes. For those who care, the mark as spam button turns from a fly swat into a weapon, a tactical vote. Luckily I don’t think many people do care about the feedback loop in which they reside. They just care about getting what’s best for them at that precise moment.
Any way - i am starting to ramble. My main point is that it is nice to see ( something close to) John Rawls’ Veil of Ignorance being put into practice. Albeit a judgement on ‘whether or not it was right for you to send them that email’. A judgement that could tip your website’s emails towards a life sentence in the spam bucket. A judgement made in complete technical ignorance of the true consequences of that decision. Where they might judge the emails of friends and foes with equality. The simple decision being: “right now, in this moment, I want that email to come here or go away“.
Perhaps that is the truest social justice. Perhaps Naïve Democracy is the only truly democratic form.
When the CAN-SPAM Act of 2003 was being discussed, there was a lot of debate about how to define spam. Seven years and millions of users’ judgements later, surely the true definition of spam now sits in the data corpi of the major email networks. I wonder if there is any correlation between the legal definition of spam and the public’s definition of spam that has been aggregated by the likes of Yahoo, Hotmail and Gmail.
