For all the talk about email overload, I'm surprised more people don't do stuff like this. It's simple and effective.
Instead, everyone wants an AI solution that automatically separates the important from the fluff. I hate to be a pessimist, but I'm doubtful we'll produce a notable success in this area any time soon, if ever. Priority Inbox certainly isn't it. I spent so much time double-checking its often-wrong results that it made more sense to just stop relying on it.
The unfortunate fact is that the concept of "importance" is vague at best. It differs from person to person, email to email. In many cases, the data that makes an email important can't be found in the inbox at all, and only exists in the real world. If I just met some guy in the hall, and promised that I'd respond to his next email, how will your program take that into account? What about online bank statements, which I normally delete instantly, but which can sometimes be helpful in reminding me to check my account? What about a Facebook message from a girl I find cute vs all the other Facebook messages?
Even humans have a hard time determining the importance of other people's email. Here's a fun experiment: Go through your last 100 emails and write done which ones were important to you. Now have a friend go through the same emails and try to predict which ones you thought were important. You may be surprised at the difference.
Couldn’t agree more. Couple of weeks ago I almost lost my domain name registration, because Gmail had stuffed the Godaddy notifications under “Bulk” which I forget to check regularly.
That being said, there is a way to use AI to cope with email overload, without running that risk. We have built a tool (www.tagwolf.com) that reduces email filing to one mosue click. It uses AI technology to analyze each email and propose the most likely folder for it on a tagcloud. The user just needs to click on the folder name and Tagwolf files the email.