powered by STREAMPAD
Click to launch FredWilson.FM music player

« One Step Closer to Conversion | Main | Been There, Done That »

False Positive Penalties

I've argued many times on this blog that blocking emails I want (the false positive) is a big problem. I value the email I get and I don't want anyone blocking legitimate emails that I want to recieve.

Most people in the spam filtering business take the approach that blocking a few legit emails is a small price to pay for protecting everyone from spam.  Well Verizon found out yesterday that that approach can be costly.

In a settlement of a class action lawsuit announced yesterday, Verizon is providing refunds to customers who had their legitimate email blocked by overly agressive spam filters in Verizon's online service.

Hopefully this will lead to a more balanced view in the spam filtering business.

Thanks to Brad Feld, who forwarded me the link to the Verizon story.

Comments (6) | | TrackBack (1)

TrackBack

TrackBack URL for this entry:
http://www.typepad.com/services/trackback/6a00d83451b2c969e200e5502230a58833

Listed below are links to weblogs that reference False Positive Penalties:

» Recent Links #7 from RadicalStu Blog
Go for a Presence Walk Go for a walk alone — anywhere you like, as long as you want. As you walk keep... [Read More]

Tracked on Apr 7, 2006 4:20:56 AM

Posted April 5, 2006 in Venture Capital and Technology

Comments

I've found that gmail does a very good job of this. I've never had a false positive, and if I should get one, i'll probably spot it immediately in the spam folder. False negatives appear only about once every few weeks, too.

Posted by: Hans Van Deun | Apr 5, 2006 10:28:09 AM

Lovely. The individual user gets (up to) $28. The lawyers want $1,400,000.

Posted by: Oskar Austegard | Apr 5, 2006 10:48:17 AM

Funny how this has been our point since day one of rolling out our anti-spam products. False positives are the dirty little secret of the anti-spam industry - especially for those products based on "guessing" technologies like spam assassin, etc. Remember folks, computers guess at spam, human's KNOW spam. That's why we use human editors to create our rules - killing more than 95% of spam isn't difficult, but not catching any legitimate emails at the same time isn't.

Posted by: David | Apr 5, 2006 12:36:15 PM

I have to disagree with you and your other commenter. I'm in the business, and it's not just me that thinks a few false positives is a small price to pay, it's the users. And it's not a dirty little secret either, everyone's perfectly straightforward about what's going on. It's the difference between being able to use mail, and having it be completely useless.

Also, you're misinterpreting what Verizon did. We're not talking a few false positives here, we're talking about some very ham-handed blocking, as in all of Europe.

This will change nothing. It also doesn't change the fact that e-mail is, and always will be a "best effort" system, not a guaranteed delivery system. I don't like it any more than you do, but that's how it is, until the SMTP mail system is completely replaced by something no one's thought of yet.

If you find someone who's got that something, I suggest you invest in it.

Posted by: Ron | Apr 5, 2006 10:10:46 PM

I'm disappointed that you'd be in favor of lawsuits as a punishment.

The lawyers are set to make $1.4 million of this, reportedly.

While false positives are indeed an issue -- using lawsuits to fix the problem is not a good solution.

Posted by: Rick | Apr 6, 2006 12:49:51 AM

I have to disagree with the above point about Gmail. Its spam filter works terribly for me, I get both false negatives and false positives.

My work mailservers pretty much have to be able to accept mail from anyone, so I do that, and then filter spam based on various content bits. I've never been a fan of rejecting email. (Though I do have friends who fall into the blacklist-rejecting camp. We argue about this over drinks.)

Posted by: candice | Apr 6, 2006 4:10:55 PM

Post a comment

This weblog only allows comments from registered users. To comment, please Sign In.