When Spam Filters Go Bad
from the it's-a-feature,-not-a-bug dept
We’ve had articles about how spam filters can be worse than spam, itself, in some cases, but here is quite a story of just how bad a bad spam filter can be. It’s the story of a Salon.com writer, who discovered that her DSL provider, RoadRunner, had installed a spam filter that basically blocked nearly all of her legitimate email – including emails from editors at the NY Times, where she was pitching a freelance piece. RoadRunner, from Time Warner, in true big company fashion responded by telling her that if she got a single item wrong on the report she sent them, they would delete it and not tell her. When she finally spoke to people, they all told her there was nothing she could do, but that those who had their mail blocked could call. Deciding that it was unfair for people she was trying to get work from to spend their time trying to figure out why her email system was blocking them, she kept on complaining. At that point, they told her the spam filter was there to “benefit subscribers”. When she pointed out that it wasn’t benefiting her at all this way, someone actually responded that she shouldn’t use her email for “commercial, or revenue generating purposes.” Stunning examples of bad spam filters and bad customer service all in one package. The article starts off quoting Frank Zappa on another matter, saying that this is “the equivalent of treating dandruff by decapitation”. It’s a bit scary just how many stories you could apply that to.
Comments on “When Spam Filters Go Bad”
No Subject Given
Actually, what they said was:
“Our system has spam filters in place to protect our network from being overloaded by bulk unsolicited e-mail. The end result benefits our subscribers, who can expect less downtime and higher service levels.”
i.e.: The primary benificiary of the spam filter is Road Runner. The subscriber is an incidental beneficiary, at best.
The fact that some people subscribe to the service specifically to get email seems to have gone unnoticed.