Wednesday, April 21, 2010

Oroborous: Facebook turns itself inside out, devours the Web

This superb article on GigaOm really puts into context what Facebook is doing. Google's search evolution was two stages:
1) Realize that the links that people create on the Web are a vote about the value and subject matter of websites ("hubs and authorities", aka PageRank)
2) Realize that the clicks that people execute on Google search pages are a vote about the value and subject matter of websites (machine learning on clicks)

and then organize search results according to the composite of the value indicated by those votes.

Facebook is now re-executing stage 2) and moving to stage 3):
2) Capture clicks via toolbar integration on websites and do Google-comparable analysis
3) Realize that "like" and other social actions (forward etc.) that people execute on the Web are a vote about the value and subject matter of websites

and then organize recommendations and exposure (in the Feed, etc.) according to the composite *personalized* value indicated by those votes.

Facebook is truly turning itself inside out with this move, and may re-shape the web around itself, in the same way that desire for "Google Juice" has re-shaped the web as we know it today. The beta phase of social networking has ended, and the real product now emerges. This is going to be an interesting ride.

Tuesday, March 16, 2010

Geeks of the world, please help Dilbert!

I just spent five minutes at the Dilbert cartoon site being frustrated by my utter inability to find a comic strip from a few years ago that I'd like to hang on my wall. It's the sequence where the new product is a box of twigs and nuts, which is integrated with the customer's network by running a CAT-5 cable through the box.

There's no text search function on Dilbert.com. No way that I can type in the relevant keywords and find the panels. Google and Bing are mute to the right answer. This is a tragedy. Given that approximately 100% of Dilbert's target audience is search-savvy, 10% of Dilbert's audience could configure a nice SOLR implementation between dinner and breakfast, and at least 1% of Dilbert's target audience could implement a crawler/scraper/parser combo to pull ASCII text out of the GIF panel archive, this is highly sub-optimal.

Geeks of the world, please help Dilbert!