Future of Web Apps (FOWA)
Session: The future of your online presence
Note by Phil Wilson:
Brett was at google and wrote google maps :)
how to aggregate and sort information
disintermediation of traditional news sources, from AP and Reuters syndication to the whole of the States to blogs and other UGC.
technical term for the amount of content people subscribe to is "a crapload".
no way to read all this stuff.
people's sharing habits are different, so the challenge really is how to sift and discover relevance. reminiscent of some of Kevin Rose's.
public data has reading trends of traditional media. different for private data, and mixing the two just doesn't work.
"filters, clustering, ranking" for helping users with the reading of content
clustering groups related content from multiple users together, so you can still get it all by expanding on that item you're interested in whilst still seeing the rest of the content.
detail on using the social graph for ranking, and how people use "like" to promote things in the rank.
mysql instances+memcached
how to best reduce the dimensionality of the data whilst keeping the relationships in place. currently most clustering is link-cluster based.
monetization may be hard, but they have lots of data about what people are interested in so there's lots of targeted stuff to be explored.
integration with other apps (i.e. other types of content being published in friendfeed) through an API is unlikely because of the potential for spam.
why did FF succeed? timing, user interface, lots of factors combined into one.
can't really believe FF has only been around for a year!
how to aggregate and sort information
disintermediation of traditional news sources, from AP and Reuters syndication to the whole of the States to blogs and other UGC.
technical term for the amount of content people subscribe to is "a crapload".
no way to read all this stuff.
people's sharing habits are different, so the challenge really is how to sift and discover relevance. reminiscent of some of Kevin Rose's.
public data has reading trends of traditional media. different for private data, and mixing the two just doesn't work.
"filters, clustering, ranking" for helping users with the reading of content
clustering groups related content from multiple users together, so you can still get it all by expanding on that item you're interested in whilst still seeing the rest of the content.
detail on using the social graph for ranking, and how people use "like" to promote things in the rank.
mysql instances+memcached
how to best reduce the dimensionality of the data whilst keeping the relationships in place. currently most clustering is link-cluster based.
monetization may be hard, but they have lots of data about what people are interested in so there's lots of targeted stuff to be explored.
integration with other apps (i.e. other types of content being published in friendfeed) through an API is unlikely because of the potential for spam.
why did FF succeed? timing, user interface, lots of factors combined into one.
can't really believe FF has only been around for a year!
