Sunday, October 17, 2010

Wither FB

So, I don't know why so many smart people are on their knees praising Facebook. It's a glorified, trivial phone book. The only, the primary innovation is Facebook Connect, which creates a connecting foundation that many have tried to create for a dozen years. The reason it succeeded? The critical mass of the FB phone book.

Is there any particular value to FB itself? Google UX research lead Paul Adams shows on how we need to be able to segment our social networks in many ways, and structurally, FB doesn't allow for this. Thats why 90% of FB posts are trivial spam.

I am sure many are banging on Zuckerberg's door begging them to think this through but FB faces lock-in technically that reinforces the tactical nature of the engineering rather than wholesale strategic shifts.

So for the moment I am more a fan of Twitter, which is just an up-to-the-minute, personalized search engine, also following design ideas that have been around for over a dozen years.

A dozen years ago I was in a startup that used trained librarians to create MARC records of high-quality websites for academia. This human research was matched by a Java plugin allowing users to drag and drop web clips, complete with bibliographic references. Structurally the same as Twitter, except that now we rely on a chain of human editors retweeting links to some semi-original thought. "Snacks." This excremental word shows how few people are paying attention, and fewer still are actually thinking anything.

So what am I doing about it? What is next? Amazon's Mechanical Turk.

We are going to build a social media analysis method using commercial tools, homegrown spiders and Mechanical Turk. MT will help us add the human ability to discern qualitative values in posts from forums and boards on brands, conditions and other subjects. We create the right questions, the system scrapes the qualitative and quantitative evidence from all the API's and posts we can find.  Bingo, we have insight.

 I am sure Google's translation engine could  accomplish this with the right AI algorithms, but they haven't done it yet, so I might as well. I will keep you all posted.


No comments: