Been watching the Spam Conference webcast (indexes at Oliver Schmelzle’s TechBlog).
- CRM114 – >99.9% accuracy
- John Graham-Cumming’s (POPFile) “The Spammer’s Compendium” prsentation is great. The slice and dice technique really is dastardly (yet ingenious)
Actually, many of the talks are really good. Other standouts: Paul Graham, Arc Project, “Better Bayesian Spam Filtering”, Matt Sergeant, MessageLabs, “Spam Filtering at the Network Level”, Joshua Goodman, Microsoft Research, “Spam Filtering: From the Lab to the Real World”, Michael Salib, MIT, “Integrating Heuristics with n-grams using Bayes and LMMSE”, Jon Praed, Internet Law Group, “How Lawsuits Against Spammers Can Aid Spam-Filtering Technology: A Spam Litigator’s View From the Front Lines”, David Berlind, CNET, “Desperately Seeking: An Anti-Spam Consortium”, and Ken Schneider, Brightmail, “Fighting Spam in Real Time”
While spam filtering is good, I don’t think anyone really questions that email as it currently stands it just plain broken. Ken Schneider had some very telling numbers. The percentage of spam (of total messages) has increased from something like 10% last September to over 40% today. And that’s on average. It seems that the larger the organization (company/isp), the larger the percentage grows.
I’ve posted some of these before, but I decided to round up some possible long-term solutions:
- Tagged Message Delivery Agent (TMDA)
- Confirmed Mail Delivery
- The Spam Problem: Moving Beyond RBLs
- The Penny Black Project
- The Spam Solutions
- Will FIlters Kill Spam?
- Keeping Spambots Out: A New Anti-Spam System – well, it’s related
There’s a new and what looks like a fairly complete spam analysis (Spam Control: Problems & Opportunities) online, but it looks like it’s a for-pay report. Also: spam costs