There was a Slashdot posting this morning on the Penny Black research project, looking at “computational signatures” for email to add a computational cost for sending email that is trivial for you and me but prohibitively expensive for spammers because of the scale.
Apart from the usual and customary trash-talking of Microsoft on Slashdot, which I’ve become pretty immune to, I think there are a couple of misconceptions in some of the comments that should be addressed.
First, there isn’t one anti-spam technology that will solve the spam program. It’s going to take a combination of complementary ones. Think of a probabilistic filter (choose your favorite variation) as the last step; it’s going to make mistakes because it has less than perfect information. We want to put as many deterministic steps in front of it so that we can make decisions about most messages long before we need to resort to educated guessing. That includes potentially some combination of: safelists/blocklists, digital signatures, HIPs, computational challenges, and other new ones that haven’t been invented yet.
The nice thing about this model is that no one technology needs to be applied universally. Sure, the wider the deployment, the better, but even “naked” email can just eventually fall through to a filter.
That said, a broad solution is going to need a combination of open standards and proprietary technologies. Microsoft has been working for several months now within a consortium to drive more widespread open standards and deployment of some technologies that should have interoperability. But the last-step filter doesn’t require interoperability, and the customer benefits from having many competing solutions. By the way, both Outlook and Exchange have standard APIs that a developer can use to write a plug-in spam filter.
The other thing that bothered me about the discussion thread this morning was the oft-repeated claim that Microsoft Research hasn’t contributed anything to Microsoft products. I’ll blog about that later today. I have to load up the kids and head to the airport to catch a plane…
