A Modest Proposal to Kill Spam
This week, Tim Bray's posted a spamonomics model.
I've talked a lot on bubblegen about how I think economic approaches to spam will be more effective then technological ones, and last week me and some buddies came up with our own spamonomics model and business plan, which has different mechanisms than Tim's.
Look - I'm never gonna find the time to give this the attention it deserves. So, in the interest of saving the world, here's a proposal complete with business model to end spam (copied from an email proposal we sent to a friend).
Here's the rub:
Assume a closed network, so that only members can email one another.
Assume a small fixed fee to join the network - maybe $10.
Everytime you send an email that is not read, a microamount gets deducted from your deposit. If you send more unread emails than your deposit covers, we bill you - if you don't pay, we shut you out of the network.
If your ratio of read mails to unread mails is greater than or equal to 1, *you don't get anything actually deducted from your deposit*. You earn the full deposit back at the end of the year (or month, or day). But if your ratio is less than 1 (or some other number we decide after trials), you get the full amount deducted from your deposit, and are also billed for the rest of the unread mails you have sent.
As a member, you have the option, after opening an email, to mark it as spam. If you do, the sender get charged - a microamount gets deducted from his deposit. But these deductions only *really* happen if the ratio is as described above.
We redistribute the money collected from spammers to people according to the amount of spam they have received (at the end of the year).
We make money on the float of the deposits.
As far as we can tell, this scheme ensures that spammers pay a huge cost for spamming - eventually driving them out of the network. Because the notion of read mails versus unread mails sets up a moral hazard problem for subject lines, we added the bit about being able to tag an email as spam after you have opened it. In effect, in this system, real users pay nothing to send email - but spammers pay whomever thinks their email is spam. People collectively decide who is a spammer - by simply not reading their emails in aggregate.
There are a lot of real-world issues - authentication, billing, network closure, etc - to be worked out. But we think that the 'spamonomics' approach is much more promising than the 'spam filtering' (blacklist, whitelist) approach. Also, we want to add semi-permeable network boundaries, and are working on ways to do that.
If you decide to turn this idea into something that makes $$$, cut us in. Or else we'll spam you.
Small print: we know not everyone in the world is going to sign up for this. Essentially, we expect to see what economics predicts - a stratified market, with spam-free networks worth more - and thus costing more - than spamful networks.
The problem with Bayesian filters (which are really effective these days): they just haven't been hacked yet. Once the spammers figure out how to hack em, expect a viral epidemic of more effective spam