Facebook Open Graph and the Web of Trust
April 29, 2010
On April 12, 1961, Yuri Gagarin became the first man to go into orbit around the Earth. The very next day, the U.S. House Committee on Science and Astronautics held an emergency meeting and recommended to President Kennedy that the United States dedicate its resources to putting a man on the moon ahead of the Soviets. On May 25th, Kennedy announced his intention to fund the Apollo space program, which had the goal of “landing a man on the moon and returning him safely to earth” before the decade was out.
Last Wednesday, Facebook launched its “Open Graph” campaign to let Facebook members tag web pages and tell their friends what they like on sites anywhere around the web. Now that it’s Monday, I’m pretty much the last person in the blogosphere to have something to say about it. Or maybe one of the first to put some perspective on it.
First, it’s not the end of the world. Although there is plenty of complaining and Big Brother teeth gnashing, and the Open Graph isn’t really open, Facebook’s Open Graph is what my nemesis Martha Stewart would call “a good thing.” Did I just say that? Yes. Here’s why.
For starters, if you have a web site that provides value, you’ll get more traffic from having Facebook members tag your pages. You would be foolish not to do it, so expect the web to have your Facebook friends woven into millions of pages in short order. That’s not entirely fluff. It’s useful for some things (I’ll explain on Thursday), and it sets the pace for more metadata and more sharing, which could fire up more activity from people who have had this same dream for a long time but weren’t really getting any traction (the same with sign-in and authorization, which I’ll cover in another post). As Alex Iskold says:
“Publishers and websites finally have a strong incentive to mark things up and get return traffic from Facebook.”
Second, it uses an Open Web Foundation standard called the OpenGraph protocol, which anyone can use and does not belong to Facebook. (It’s open but it’s not completely open – it’s a nonprofit, but not anyone can join. We will have to see just how open it becomes.) It allows people to use the same protocol as Facebook without fear of being sued. As Jesse Stay explains:
“… it means that the new Open Graph Protocol announced by Facebook yesterday is under a completely open license agreement that other platform creators can adopt, use, and freely distribute without worry of patent.”
It’s true that the results of your “likes” will flow back to Facebook and can only be seen by your Facebook friends (and advertisers), but it’s also true that any company or group can implement the same exact thing. So if a large number of social networks align, say, with Google, they can implement a second “like” button, and when you click that one, its results will go to a central repository and be distributed everywhere from there.
As it turns out, a group of people already had that idea, so they started the OpenLike Project. Because the code and icons Facebook uses comes from the Open Web Foundation, this group decided to make it more general and allow people to own their own recommendations that go out more broadly than just Facebook. One of the people behind it is Chris Dixon, who has thought a lot about the open web and hopes it can catch up to Facebook.
Except that the central repository I just mentioned doesn’t exist. It could be your Google account, it could be your Ning account, or your LinkedIn account, or, as I mentioned in my book, someone may build the personal data locker, which would let you author, publish, and control all your personal data, including your ratings and reviews, and your social network connections. A company like Power.com could jump in here and try to get traction by using the “like” button to get more people to sign up for their service – all they have to do is host your metadata and distribute it. It could be a welcome service for all the little social networks out there (think of aSW, which is perennially cash strapped and has a hard time implementing any new features). As I say, so could Google. In fact, I’m sure some smart people could figure out how to combine this all into a single “like” button, so your rating goes out not just to Facebook, but also to your own platform in the cloud. This really could happen, because Facebook is playing nice and using an open, human-readable, non-proprietary way of putting the “like” buttons on web pages.
So if we have a dream of this open web, where people can share information about any web page with people they trust on any platform, where people own their own data and who sees it, the Facebook announcement takes us about halfway there in a matter of weeks. Goals of world domination aside, Facebook has played this card in a way that will encourage wide adoption. The one smelly fish in the mix is the Microsoft partnership, which simultaneously gives Facebook a bit more Googleness (docs) and Microsoft a bit more Facebookiness. It smells because Microsoft is an investor in Facebook, and we can now see the investment paying off for Microsoft without creating much value for Facebook users. Microsoft will take as much of the Facebook experience as they can get away with, and this is not the end of it.
And – ho! – here we see RDFa in the Open Graph! That’s right. Not only has Facebook chosen to go with an open format but with linked data! I hear a collective applause from the semantic web community, validating their principles. Linked data is important, because it lets us do things in the future we can’t imagine today. It’s flexible. It scales. It is human readable. It is the glue that makes the semantic web work. And now we’ll have tons of it.
How does the Open Graph fare on the semantic web acid test? Well, it’s semantic for sure. Is it on the web? I would have to say no. It’s distributed on the web, but everything feeds back to Facebook, the same way Amazon.com affiliate links all drain back to Amazon.com. That’s an extension of a silo, although it’s a pretty big extension. The OpenGraph is on the web, so let’s hope that one can catch up. And, finally, is it pull? Actually, it is. It’s pull (vs push) because you publish your “likes” in a semantic format that your friends can pull when they want it (e.g., from their mobile phones while walking around looking for a restaurant in your neighborhood). Okay, they have to use Facebook to do the pulling, but it’s still pull, and that makes it a good model. If we can extend it through the Open Web Foundation and OpenLike, we’ll cover the web in semantic recommendations that will add a lot of value and make using the web more effective and relevant for all of us.
Facebook’s Open Graph is a big step. It could provide the incentive for the Open Graph to get even bigger. It happened in the space race back in the 1960s. It could happen in the recommendation race online fifty years later. Let’s forget about whining and get the entire web on a semantic footing, at least for recommendations.
Part II
THREE DAYS IS A LONG TIME when everyone is talking about the Facebook Open Graph and billions of “likes” are heading back to Facebook servers to enrich our lives online. I said last week I hope this is the catalyst for a broader movement to build our identities, share our personal information, and establish networks online. I believe the Open Graph as it is today is a tiny fraction of what is to come. Here are a few observations:
Granularity counts. I have 500 friends on Facebook, but they represent many different kinds of relationships. I don’t have them subdivided into groups, so I’m careful what I share, because I know everyone sees everything. Some of these people are very good friends; others I’ve never met. It’s been great reconnecting with old friends on FB, but if the web site disappeared tomorrow, I’d get over it pretty quickly.
Ratings are practically meaningless. If I go to a web site and see that my friends like a story or a product or a movie, it isn’t generally going to move my needle very much. You can’t say you don’t like something, and you can’t say anything qualitative. Perhaps if I were thinking about going to a particular hotel and I saw that one of my friends had stayed there, I’d ask him or her about it. But ratings are quantitative, and the numbers are easy to manipulate. I think most site owners see the Open Graph as a new kind of SEO, to bring in friends from Facebook to their sites, and that’s what Facebook has in mind. But it isn’t really that big a deal.
Reviews are qualitative and potentially meaningful. If you come to New York, you could get some restaurant recommendations from me, but you could also just look at the Zagat guide and probably get more value out of it. Not only are the restaurants rated on four different criteria, you can also read the mini reviews, and they help a lot. The reviews on sites like Amazon.com are gold – they sell products. The best kinds of reviews share pros and cons, so you can see whether they line up with your values. I’d rather read the camera reviews from people I don’t know at DPReview.com than read camera reviews written by my friends (sorry, friends)> Like comments after blog posts, reviews can be well thought out or mindless rants or spam. Or even shills – people whom the author asked to make specific comments. Reviews can be well meaning but just not good enough for many people to make a decision on. This is where you might trust your friends, but you also trust authorities. On Amazon.com or epinions, you can pretty easily tell who the authorities are, and they don’t have to be your friends.
Ratings of reviews are valuable. At Amazon.com, the reviews are ranked by their ratings. Again, this can be valuable, but it can also be manipulated. One thing I’ve learned about Amazon.com is that the number of reviews is more important than the overall score given
A web of trust is not a web of “friends.” Let’s say you’re looking for a doctor or a lawyer or a babysitter or a mate online. You can find them on many web sites. You may trust a nanny referred by friend A but not a lawyer referred by that same person. You may read reviews written by people you don’t know and take their advice. You might read reviews by people you don’t know who let you contact them to ask more questions – now that’s a good review! One of the goals of the semantic web is to build a web of trust on open standards. The standards are coming, but how will we put them to practical use? Perhaps they will rid e on top of FOAF and become useful to millions. Perhaps Facebook or Google will implement something many people like. It’s not only difficult to set up a web of trust, but it also takes effort to maintain a web of trust. Things happen. Not all boyfriends go on to become husbands.
A web of trust will need discovery methods. Suppose you’re interested in hiring Tom as a project manager. Tom gives you his resume and three references. But Tom may not give you that fourth reference, Bill, the person you’d really like to talk to. Tom may have had a problem with Bill once. Tom still may be the right person for the job, but you would definitely like to get Bill on the phone and discuss it to see what you can learn and how you can benefit from that experience in better managing Tom. I don’t see Facebook providing the ability to find Bill any time soon.
Movies, songs, and books are good candidates for the “Like” button. In this case, you know your friends’ tastes, so having their “likes” at your fingertips is interesting, especially when you’re looking for a book to take to the beach. You certainly might want to read book reviews written by other people you’ve never met, but you also might be compelled to read Nurture Shock, by Po Bronson and Ashley Merryman. If you have a child under 15 years old, and you trust me even a tiny bit, please buy this book. You can thank me by buying mine while you’re at it.
If you’ve read this far, I’m going to leave you with a reward, a thought you may not have heard before. The Facebook “like” button has nothing to do with it, but I predict within three years we’ll see the emergence of Facebook cameras. Yes, you’ll buy a camera and it will have the Facebook logo on it. I wouldn’t be surprised if they were negotiating the deal now. A Facebook camera would not only send your photos straight to Facebook, it would use your list of friends to automatically identify the people in the photos, tag them, and notify them. You don’t really need to do much to the camera – you can do everything once the photos are inside Facebook. A Facebook phone will automatically suck your contacts in straight from Facebook, and whenever someone changes his/her phone number on his/her profile, that number will change in your phone. I note that Steve Rubel, the guy who is too busy to read my book, has already posted something about this, but he didn’t mention the part about the contacts automatically updating. Want to read more about these kinds of phones? It’s all in chapter 8 of my book, Pull.
