The final spam solution: Social networking.
rzwitserloot posted in tech on June 22nd, 2006

Seth Godin reports on the spammers having found news aggregators. As you may have guessed, the news isn’t good for the aggregators.

Seth suggests counting a ‘known’, ‘respected’ voice more than a total stranger’s, where those terms (known/respected) are defined by their previous submission’s performance. The problem with that tactic, while it would work, is that you get into the groupthink problem: Only a small set of people will bother to submit news, because the rest would have to be noticed amongst an onslaught of SPAM. It’s been proven, time and time again, that making SPAM irrelevant through the use of spam filters and such doesn’t stop them: It’s so cheap for spammers to do it (they’ve aready written their scripts) that they’ll continue for the benefit of those few who check new submissions regardless of reputation of the submitter.

Another problem is hopping sites: If a reputable contributor from one thing decides to get into something else, he has to start all over again. Inefficient. A third problem is for those beginning their foray into internet contribution: They do have to start from scratch, but if you need a reputation with at least one site to get into others, our pilgrim will never get there.

What we really need is a leaf from the real world: Trust.

Here’s how it works: If a friend (or someone else I trust ‘in real life’ OR from the web) points me to something interesting, it tends not to be a spammer. We can use the web to extend this principle to half to all of the entire world population by using the social web: If a friend of a friend of a friend of a friend finds something interesting, its not spam.

if it does end up being spam, then I tell my friend that his friend’s friend needs to be told about a spam linkage in my social web. At some point in the chain, the spammer is found out, and either the spammer’s friend drops his trust, or he convinces the spammer to stop, or if spammer’s friend is unwilling to rat out the spammer, the spammer’s friends’ friend drops the trust of the spammer’s friend.

The problem is that it only works exactly as described if everyone’s personal social web is unique. This is a non-trivial computing issue, in that it requires huge resources and no small amount of cryptography to ensure that identities are reputable without also making people identifiable without being asked for it.

I have no definitive solution for how to set it all up technically, but I do think such a system is utterly impervious to spammers and widely applicable, from blog commenting to news aggregators to the original: email.