But I believe my long-term experiment using TMDA has reached the end. I was originally excited about the prospects of using C/R to alleviate my spam issues. Then the social drawbacks of C/R started to rear its ugly head: People don't want to deal with C/R. People take offense to being classified as a spammer unless they prove otherwise. People can't follow simple instructions, or simply lack the technological prowess to determine whether a challenge is valid. Unfortunately, it's a bit difficult to cop the attitude that they are probably not worth dealing with in the first place, because many of these people are friends and relatives.
So out went C/R. But I kept using keyword addresses. That has eventually evolved into a management nightmare: Keeping track of which mailing lists use which header fields, or which list managers use administrative From: headers during signup which are different from the regular From: headers. I found it hard to convince myself that the benefits gained from using keyword addresses was worth the time spent keeping all my mailing lists happy. So out went keyword addresses. I gradually merged TMDA with SA, using SA as my primary spam filter, while using TMDA to queue up questionable spam in the tmda-pending queue. I even wrote a cool little app to manage the pending queue via e-mail. Over time, I discovered something: The false positive rate of SA dropped to nearly nil once I figured out how to automate Bayes learning on those questionable e-mails. Over time, my pending queue ended up being 100% spam, so there was no longer anything to manage. So out went the pending queue. The one thing that was left of TMDA that I was still using was whitelists/blacklists. Then I discovered that SA would not only handle whitelists/blacklists, but wildcards as well. So out went TMDA white/blacklists. Which didn't leave much left for TMDA to do. Don't get me wrong: I found TMDA to be well-written, well-supported, excellently documented and very functional. But over time, the utility of TMDA gradually decreased until there was little return on time invested keeping TMDA up-to-date, tracking the mailing list, etc. I want to thank Jason (wherever he might be) for a great piece of software, even though we sometimes didn't see eye-to-eye on some things. And thanks to a great group of TMDA fanatics who were more than forthcoming with assistance. Maybe down the road I might revisit TMDA if the need arises. You all take care, and keep fighting the good fight. --Brian _____________________________________________ tmda-users mailing list ([email protected]) http://tmda.net/lists/listinfo/tmda-users
