As you may recall, Abaca has a really interesting spin on the spam filtering problem. Finely-tuned mathematics and a big database of receiver statistics give back up some truly impressive claims. As I said last year, I'm reasonably convinced that it's not just a silly FUSSP.
For over a year, Abaca has been working on a deal with Yahoo! to add the technology—which they now call CLX—to the spam filtering mix. A few months ago, I heard unofficially that Yahoo! agreed to roll it out.
Now, Abaca is announcing that the rollout has been hugely successful, and Yahoo! is extremely satisfied with the result. Nice going.
As an update, here's the (claimed) highlights of the Abaca technology:
- Guaranteed accuracy of at least 99% catch rate (with money-back contract terms)
- Claimed false positive rate is infinitesimal (I calculate their claims equate to one in a million messages)
- After bootstrapping with recipient email statistics, no user training is required, but can be individualized by users clicking the Spam/Not-spam buttons
- By its nature, it's extremely scalable—a single small server can handle 90 million messages per hour
Equally, I don't know how close to reality the false positive figures are -- at best they're based on user reports alone, which usually tend to significantly under-state the reality. But, again, if the Yahoo! user reports are anything close to 1:1,000,000, then Abaca has something really worth shouting about.