Does (Spam|Phishing) Filtering == Email Censoring?
I was reading about the Gmail Labs option to display a key icon if the sender's domain is signed using DKIM and the sender is eBay or PayPal. This allows you to quickly verify if the email is legitimate by looking at the icon. Now it apparently takes some work for a domain to be "super-trustworthy", so this key can't just work for any domain. (I suggested two types of keys, one for all DKIM emails and one for these "super-trustworthy" DKIM emails -- almost like SSL vs EV SSL (it kinda hurt to say that though))
Anyways, to get back on track, as I was reading some of the comments on the Google Group, I came across this one, 'Censoring my Email'. It actually made me stop and think for a second. One one hand Gmail is indeed censoring the email you see, however they're doing it to filter spam... is it really censoring at that point?
I think we first need to consider what's being filtered. Any email from paypal.com or ebay.com (or their international counterpart domains) must be signed with DKIM. If Gmail can verify the DKIM signature, it delivers it to your inbox, however if they can't they send it to /dev/null. How much spam does this filter? Well, basically anyone who's set their own 'MAIL FROM' response to paypal.com/ebay.com. People who set their name to 'PayPal Support' with an email address of paypal-support@gmail.com will not be filtered and will show up as just 'PayPal Support', unless the recipient clicks 'Show Details'.
Now imagine that you're a non-technical Gmail user who's read an article that says paypal.com/ebay.com emails aren't even delivered to you if they are spam (that wasn't quite the wording Gmail used, but it's not hard to imagine it happening). You see an email that says 'PayPal Support' and you're going to click on it (after all, users are trusting... that's why phishing works in the first place). This could cause a lot of problems (maybe this is why the idea of showing the key for "super-trustworthy" domains came along even). So Gmail responds by introducing this key icon... and when you look at it this way, it almost seems required. Yet it was this introduction that made the filtering more evident to people and which prompted the commented that sparked this blog post.
So, back to the original question... is filtering spam and phishing emails the same as censoring email. I definitely don't think so. I applaud Gmail for making an effort to limit the spam that appears in a persons inbox (if only they were filtering my personal and work email
). However, I disagree with their approach and I see two problems with it.
The first is that they waited over a year between filtering email and providing verification for valid email. This could have lead to many cases like the scenario I described above and since the feature is only in Labs, not everyone will use it and it could lead to many, many more cases like the that.
The second is that they filter anything not signed via DKIM from ebay.com/paypal.com. After reading about this I went and setup DKIM on my server to get a better understanding of how it works. It requires a trust in two protocols that can't necessarily be trusted, SMTP and DNS. What happens when eBay/PayPal have a DNS issue and restart DNS and it doesn't start immediately... how many potentially valid emails could be dropped? What happens if someone gets it in their heads attack Gmail with DNS Cache Poisoning? What if someone at eBay/PayPal adjusts a mail server rule and the DKIM header stops being sent?
It's entirely possible that this email is "super-trustworthy" because work arounds have been implemented for every issue I've mentioned above, that still doesn't protect users that don't have the key icon yet. At this point, I guess the best we can hope for, is that this feature spends very little time in Labs before being implemented across Gmail.
So in the end... (Spam|Phishing) Filtering != Email Censoring and we should be thankful for it, not fighting it.


