incrediBILL - 4:36 pm on Sep 13, 2012 (gmt 0)
Likes and Links have a lot in common, they're both a vote, they should both be taken into account, they both can be gamed and paid.
Then there are more complex forms of 'likes' like reviews, 4sq logins, even references in Tweets and email are votes or 'likes'. Why do you think the big players all host email so the automation, the same automation that shows you relevant ads, can also show more relevant searches. It's another form of counting votes for sites or topics.
Sites that get a ton of reviews should probably outrank sites that only get a couple. However, if the site with only a few reviews has a higher rating should it outweigh the massively review site? Massively more popular doesn't necessarily mean better, only that it gets a lot more traffic and maybe it shouldn't.
Checking in with Four Square is another form of like and even your raw GPS tracking is a form of LIKE as they translate your coordinates to the real world B&M and then look up the related site. It's a lot of data to process to get real world preferences, but far more reliable than depending on your reviews or even those 4sq checking because if you keep going back to that location you really like it or need it and it should rank better.
Clicks on the links at the SE itself are a basic 'LIKE+' and bounce rates should be counted as 'LIKE-', actual user preferences, which would be a far better indicator IMO of what people are actually using and deserve to be ranked higher than any simplistic social ranking.
If you start adding it all up, traditional SEO could be (may already be) meaningless except for initially positioning the site in the search engine as real world usage will trend it into it's final position.
Cool stuff will float and the sucky will sink. The whole thing will sort itself out organically and there will be so many signals taken into account that gaming the system will be nearly impossible unless you lots of human resources.