Right, and especially if it was not just saying, "Here's who we think you are," but also at a basic level, "Here's what we're doing with that." And "Here's the domain of the screen that is using that information and here's some stuff that gets you outside of that information." I talked to YouTube yesterday and had a great conversation with people there about, "What would you actually do about this?" And the idea that comes up again and again, kind of silly, is an "If you like this, you might not like that" function. The thing I love about it is it's this kind of psychological dare. It's betting you that you're a bigger, better person than the typecasting would suggest you are. In some ways you're playing off of the Librarything.com "unsuggester," which you talk about in the book, and is sort of everyone's favorite response to over-personalization. [Ed. note: It works like this: type in books you've read, and it will return personalized suggestions for whatnot to read.] I happened to go to Librarything before this, typed in books I liked, and I immediately got back a ton of Bible commentaries. I don't know how three books turn me into a sinner, but it was actually very hard to figure out where I might go from there. Can we do better than "You like this, here's something you don't like?"
Yeah, I think the question is, How do you bring other dimensions to play? In a way, you're taking this incredibly complex world and trying to flatten it into two dimensional space that you're mapping on a map, right? And you're saying this is near this, and this is far from that. And really what you want to do is have a bunch of different dimensions at play, so you can say, "This article is something that is far from you in political beliefs but people feel is of really high quality." I would totally be interested in that. I wouldn't necessarily agree with it, but my problem when I got to read right wing blogs is that I don't know where to go to find the thing that is really going to speak to me that way. You can mix in, I think, other signals which starts with just capturing other signal aside from "thumbs up, this was great!" It's sort of interesting that tagging culture and the explosion of folksonomies somehow hasn't totally merged with the personalization culture. People are learning to describe things in all these wonderful ways but it's not really being drawn onto inform how data is then presented very well.
I think you put your finger on it: "Who's acting?" Right? What you're basically saying is that the people who are personalizing don't care so much about giving you great results as they do about getting you to buy – or getting you to stick around. And if they can give you something slightly better and that's gonna do slightly better, they're willing to do that. The goal is to have it be very broad and very pervasive. Things like tagging are pretty small, at this point: people being able to tag things on Twitter or Flickr and so on. The domain is still pretty small even if the accuracy is pretty high. So maybe what this really comes down to is not asking the Googles and the Facebooks to solve it, but asking people to start building tools to solve it. Are you running into anyone that is helping you punch through bubbles in an effective way?
, Technology, Facebook, Harvard Book Store, More