Building Smart Lists from an Inbox

Photo by koalazymonkey on Flickr used under Creative Commons

Saw an interesting quote attributed to Mark Zuckerberg yesterday:

But guess what? Nobody wants to make lists.

A bit out of context, but Zuckerberg was expanding on the best way to share things with a network of people. He’s absolutely right about lists.

They are simultaneously the right way to share and too cumbersome to manage. Lists are a lot like email filters and folders and other methods for controlling information: easy to configure, tough to maintain.

Setting up lists isn’t terribly difficult, but as your network grows, they become a chore to manage, e.g. a new connection may not fit nicely into an existing list. Does the new person warrant a new list, or should s/he be crammed into an existing list? If the latter, are their other connections that should also be changed?

Curation is work. This is exactly my problem with organizing email.

Clayton (@cdonley) makes this very point about enterprises. LDAP stores are perfect for creating these lists, and IT manages them, making this a better solution. Inside the firewall, your relationships are known (and protected), and LDAP can use backoffice systems to create smart lists based on this information.

But what if you don’t have these benefits?

Paul (@ppedrazzi) and I had an ad hoc brainstorming session at SXSW about this. As with most software problems, the answer is an algorithm. In this case, the data source for the algorithm is your inbox.

Email is the standard for sharing things. You have to tell it with whom to share and at what level, e.g. to vs. cc vs. bcc. And email already has lists.

Indexing an inbox could tell an algorithm a lot about the relationship between the sender and recipient, which in turn, would help build smart lists.

Facebook already offers email and is rumored to be moving toward offering web mail, e.g. @facebook.com addresses. Google already indexes GMail and could easily add a smart list capability to their rumored Google Me network.

The major downside is the freaky factor, as with most algorithms. Mining someone’s inbox to create smart lists could easily feel like an invasion of privacy.

Anyway, there’s a lot of useful information locked away in an inbox that could be used to create lists, among other things. Once created, the network could maintain the lists in a relatively hands-off manner, or leave it to the user.

What do you think? Do you mind curating lists? How would you automate list creation and management?

Find the comments.