Spam On The Decline… Or Not?
from the maybe-possibly dept
Despite plenty of reports coming out this week that the spam problem is getting worse, AOL made a surprise announcement today saying that they’ve seen a “dramatic” decline in spam targeted at their network in the past month – down 27%. They’ve also seen a noticeable decline in spam complaints, which they attribute to better filtering technology. Of course, it sounds like Comcast isn’t so lucky with spam. Broadband Reports says that, despite a recent effort to cut off zombie machines sending spam, more Comcast machines than ever are spewing spam out into the world – just apparently not in the direction of AOL. Of course, as all of this goes on, people are beginning to realize that there are way too many ineffective anti-spam companies out there, and it’s time for a bit of consolidation in the space. If it meant the filters actually worked a bit better, I’m all for it. I’m still finding that free filtering technologies seem to work just as well, if not better, than all the pay anti-spam services.
Comments on “Spam On The Decline… Or Not?”
... not
To spam on the decline I’d say … not!
Fixing Spam Problems
I can’t profess to having the overall answer to this issue but there are SIMPLE (god-awfully SIMPLE!) ways to attack fundamental infrastructure problems that exist that allow spam to be a problem.
The first most important thing that people need to be railing about is proper ingress and egress filtering in service provider networks. This is simple. The IANA allocates networks that should be Internet routable. Thus, if IANA hasn’t allocated the network, there should never be [1] traffic coming from it and [2] no need for traffic to go *TO* it. Stopping provider networks from passing this traffic, so-called, bogon filtering, means that spammers now have to use legitimate networks, UNABLE TO HIDE BEHIND SPOOFED NETWORK ADDRESSES. And then as a matter of policy, providers can decline traffic from those legitimate networks that don’t enforce abuse policies. This practice of stopping traffic from entering your network that violates a given policy is simply “policy-based routing” and most providers do it to some degree anyway. Just none seems to have the nuts to do it w.r.t. spammers and their accomplice network providers. Not a total or complete solution, but damn helpful. And it would ratchet down the severity of all manner of other network problems, such as (D)DoSes.
As far as authentication, e-mail signing and encryption already exists. The big issue is making it easier. Working for the company I do (a large provider) and seeing how difficult this stuff is in Windoze (and I’ve used GnuPG on Un*x as well), the usability issue the biggest impediment to this. There is all sorts of flexibility that could be built around signing e-mail. E-mail isn’t signed OR isn’t signed by someone in your keyring? Delete it, or drop it in a generic box for later inspection. Mathematical whitelisting. In a large organization, an e-mail admin should be able to add all of the users to an organizational keyring, thus automatically accepting e-mail from internal sources and denying it (or applying policy to it) from external sources. Many of the tools are already there, just no one seems to be interested in putting them together into useful tools and architectures.