Select Page

I have always asserted that Facebook’s most valuable asset was its event stream (which you can see by clicking ‘home’ from within Facebook). It shows you what’s happening in your network. The other day I complained about having to block more than 100 apps within Facebook to keep their spam out of my daily flow.  A pain, for sure, but this has had substantial effect in making Facebook more usable to me day-to-day.

I haven’t ever really thought of Facebook as a productivity tool (more the opposite) until lately.  I have tried using many feedreaders, including Google Reader, Firefox, YahOo, NewsGator, and more over the years.  More recently I have abandoned many of those in favour of a desktop client like Nambu or  Tweetie in respect of the fact that I think my friends (or at least those that I choose to follow on Twitter) are the best filter imaginable — and far better than any search engine algorithm could ever produce.  Therefore, Twitter has quickly usurped my consumption of RSS as mediated through most aggregation tools — simply because no search engine I’ve found is good enough at understanding me and filtering the crap.

facebook-feedreaderHowever, as more and more of my friends are adding the Twitter tool to their Facebook accounts, and thereby syndicating their tweets to the Facebook status updates, I am turning to the desktop clients less and less and spending more time watching the event stream in Facebook.

I still post 100% of the time from within a twitter client on my Macbook Pro or my iPhone (fanboi) however the time I spend reading stuff that comes to me via those clients is decelerating pretty rapidly. In fact, my biggest gripe about Nambu for the iPhone is that it insists on loading up my event stream before I can post a new tweet (try that on EDGE).

Facebook’s event stream has one key advantage that Twitter doesn’t.  If I find someone on Twitter is annoying me with their posts, the relationship is fairly binary: I either follow them or I don’t.  However, with Facebook this is quite nuanced.  Specifically, I can Hide updates from people I’m friends with who post garbage.  This is an improvement over Twitter, but is again too binary (and is punitive to Facebook’s parity-centric follow model).

What I think both Facebook and Twitter users would benefit from are two nuanced approaches to promoting or demoting content in the event stream.  There are computational effects here that are not trivial, but this is the kind of stuff we were working on at Something Simpler.  I want a thumbs up / thumbs down on both users and content.  In the user context, a thumbs up promotes the user in the priority tree, thumbs down demotes.  In the context of the content item (a tweet or otherwise) I am training a bayesian filter that pulls keywords out of the content and promotes or demotes similar future items depending on which I selected.  The engine must then score the content based on who it came from and what the extracted keywords are.  It can also look at my OWN event stream to train the Bayesian filter and surmise that the things I post about are likely to be similar to the things that I want to read.  These embrace the peer relationship while increasing the quality of my event stream.

I think this is a rejection of the hypothesis that I held when we were originally offered the chance to find a place for the PubSub assets:  that machines could judge this on their own, without our help.  The reality is that’s expensive, inefficient, takes too long to train, and doesn’t return enough immediate near-term benefit to the user.  However the methods described above are very relevant, embrace the social nature of online networks, and displace the heavy lifting of filtering on to my friend networks, which Twitter et al already accomplish.

The first tool to implement these features gets my vote to be my primary feedreader.  Nothing’s more relevant to me than my friends.  Neither Facebook nor Twitter is doing enough to make their knowledge of who my friends are useful day-to-day.  That’s all they’ve got versus other tools today.