Spam is becoming a larger problem every day. In order to deal with this problem, RPI has provided a service called Respite. Respite acts as a filter that sits between RPI’s incoming mail servers and the mail storage queues for each user. Mail is analyzed, and any mail that is identified as spam is either tagged, held on the Respite server, or dropped.
Respite has been an opt-in service, meaning that users who want to use it could sign up for it. Over the weekend, however, Respite was turned into a mandatory service. Even if users select the “opt-out” option, Respite will be set to “average” settings.
This would be harmless if Respite only caught spam. However, spam filters only identify properties of spam. They essentially find a probability that a message is spam. If the probability is high enough, it classifies the message as spam. Thus, there is always a chance that a message that is not actually spam is identified as such.
Respite’s error rate is about 1%. One in every 100 valid e-mails that you should receive is, by design, marked as spam. Worse, for me—and some other users—it seems like the default policy is to drop spam mails without first holding them. If Respite thinks the message is spam, it is gone forever. DotCIO says that its server is set to bounce alleged spams back to the sender. However, if an automated process sent the original mails, the bounce process could be ineffective.
It’s the time of year when many students—especially seniors—are waiting to hear back from companies or graduate schools about job or admission offers. What if that one-in-100 message is an interview request? What’s worse is that many of these messages are sent by automated systems and may appear more statistically likely to be spam than they should.
There was no warning that Respite was going to be made mandatory, except a notice titled that the system was going to be upgraded. When I see a notice that a product that I don’t use is updated, I don’t assume that it will affect me. A more appropriate act would have been to e-mail students, informing us of the policy change. We already receive e-mail messages telling us of new Institute news and policies. Why couldn’t an e-mail be sent out telling us that our e-mail service policy would be changing?
In an e-mail that I got from an RPI system administrator, he claims that the mandatory policy “was driven by the massive increase in spam, and the side-effects it was causing on our systems, including RPI being blacklisted by ISPs [and] placed on RBLs.” How delivering e-mail to its own mail servers would cause other ISPs to blacklist RPI, I do not know, seeing how once rpi.edu e-mail makes it to RPI, it does not leave RPI. If this is referring to people who have their mail forwarded to an external address, outgoing mail filtering is sufficient.
Quoting the “What’s New” page, “To use e-mail today, one must accept that ... some legitimate e-mail will be accidentally trapped ... if you can live with [these] mistakes, you can lead a robust e-mail-experience. The rest of life—vast as it is—is up to you.” I don’t understand exactly what a robust e-mail experience is, or how it reflects on the rest of my life. I hope that I never have a problem with the service blocking a valid, important e-mail, but I am not that much of a clueless optimist.
