But do you know why they chose to like something? With Coke, it's possibly a fair bet that it was because of a competition. But for a local restaurant, even if they've gone there and liked it, that could simply be because they saw a "get a 10% discount on your bill if you like us on Facebook",

Its a simple catch-22 - a "like" has no downside cost to me, so I will happily trade it in for 10% of my bill (Hell I wish more UK restaurants did this!)

However in the new graph search world (and indeed before this really) the downside cost was that I am seen as a like-whore trading my affections.

The more facebook becomes real the more the cost of like will increase.

There will be a virtual rate of exchange - as indeed there is today. Some of my friends tell everyone everything, some play their cards close to their chest. I judge their recommendations accordingly. Somehow social graphs need to do the same - and frankly that is pretty easy weighting really.

I'm still not sure how you (or Facebook) is meant to be able to tell the difference between a genuine like and a 10% discount like.

the point is that the real intent is somewhat independent and almost orthogonal to pressing the "like" button. that's the point the article is making about noisy and dirty data.