Postini bug

Ben over at MailChimp has an article talking about a recent experience with Postini and an actual bug that causes Postini to interact badly with another spamfilter and block non-spam.

Related Posts

Troubleshooting a Postini block

Mail from one of my clients is being filtered at Postini and they asked me to look into this. Not that there is anything that can be done, of course. Even before they were bought out by Google, they were the poster child for a spam filtering company that believed they could do no wrong. It was difficult, if not impossible to get a straight answer from Postini about filtering, and the only statement they would ever make in regards to blocking problems was ‘have the recipient whitelist your mail.’
It is not just that Postini will not talk with people who are blocked, they will not talk to their own customers, either. Many years ago, I was dealing with another Postini issue for a customer. This customer was a Postini customer and was sending mail to themselves to test their new ESP. Postini was blocking the mail and the customer wanted me to find out why. After a couple days of digging I did actually find a really-o truly-o human at Postini. [1] He explained to me that a single line of text, followed by an unsubscribe link was spam, always spam and nothing but spam. He also explained that the only way for that mail to be let through, was for my customer to turn off his Postini filters.
Fast forward 4 years and I once again have a customer blocked by Postini.  Usually, I tell customers there is nothing to be done for Postini blocks and that no one can find any information about them, but this customer is insistent. This particular customer has extremely clean mailing practices, sends highly relevant and wanted mail and consistently gets 95+% inbox delivery. They are not spammers, not even a little bit. Because I know this customer is so clean, I poked around a little to find some information about them. They do use the ReturnPath Mailbox Monitor so I have a copy of the headers Postini is adding. I also discovered that Postini is now providing a decoder service for their headers at https://www.postini.com/support/header_analyzer.php
The response you get back from pasting in a header is not that useful if you have found any of the numerous explanations of Postini headers, but it does show some willing. Note, there is no way to ask a question or provide feedback to Postini on the listing.
There is not much that can be done to deal with Postini filtering your email. The best you can do is have your recipients whitelist you.
[1] I believe I am the only person on the delivery end that has ever been able to actually talk to a live human at Postini, and I think that is only because I called them from the same area code they are in and some engineer decided to return the message I left on their corporate voicemail.

Read More

Forgery and spamware

Recently there has been a massive uptick in forgeries. I have been seeing hundreds of bounce back messages, peaking at more than 1000 in an hour. I have been talking about this with people who monitor large spamtrap feeds, large MTAs and spamfilters and it seems this is not an isolated experience. The consensus seems to be that there is new spamware out there which is using email addresses on the spam list as a From: address
The volume itself is annoying. Thousands of messages a day from “mailer-daemon” telling me that the mail I sent with the subject line “Get a longer tool” cannot be delivered to some random address some where. These are coming to at least 3 separate email addresses. One of them was given to Intuit back in 2001/2002 when I registered a copy of Quicken, and ended up leaked to loan spammers and is all over spam lists. The other two are addresses scraped from websites. Same spammer has them, same spammer is using them as part of his spam run.
Even more annoying than the volume, though, is the challenge/response emails. “Your email to jobobjimbo@example.com cannot be delivered until you click this link.” I have been adding every domain I can find that is using c/r to my filters, and just discarding the c/r emails so I do not have to deal with them. That is not my ideal solution, it does mean that if someone using c/r ever tries to contact me I will not see the challenge and our communications cannot happen.
Some people have recommended that the right way to deal with challenges from forged spam are actually to answer the challenges. As the reasoning goes, if someone using c/r is going to outsource their spam filtering to a victim of spam forgery, then they should expect that the “spam filter” may have a different opinion than they do. While I always sympathized with this viewpoint, I was not sure I would ever confirm spam forgeries. The sheer volume of c/r stuff I have received in the last few weeks has almost convinced me that people who use c/r deserve every bit of spam they get. If a c/r filter lets in spam, then perhaps they will reconsider their choice to spew challenges out to forged email addresses.
The amount of c/r spam I am getting as part of the forgery runs is decreasing, I think I have finally managed to block the primary sources. It does mean I will not be able to communicate with people who use c/r in the future, but I find this a small price to pay for not having to be an outsourced spam filter. I get enough of my own spam, I really do not want to have to deal with yours.

Read More

More on spamfiltering feedback

Al wrote a post commenting on my post from last Thursday on spamfilters talking to senders who are being filtered. I think his take on it is close to mine. I would point out that Google has a pretty opaque system and no feedback to senders, but a lot of people seem to think their filters are accurate and do a good job.
Overall, I think there is room for discussion and feedback between senders and recipients, but on both sides the goal needs to be improving the enduser experience.

Read More