Understanding the Open Graph Protocol

All likes lead to Facebook

I attended Facebook’s F8 conference yesterday (missed the keynote IRL, but you can catch it online) and came away pondering the Open Graph Protocol.

In they keynote Zuck said (as Luke Shepard calls him):

Today the web exists mostly as a series of unstructured links between pages. This has been a powerful model, but it’s really just the start. The open graph puts people at the center of the web. It means that the web can become a set of personally and semantically meaningful connections between people and things.

While I agree that the web is transmogrifying from a web of documents to a web of people, I have deep misgivings about what the Open Graph Protocol — along with Facebook’s new Like button — means for the open web.

There are three elements of Facebook’s announcements that seem to conspire against the web:

  • A new format
  • Convenient to implement
  • Facebook account required

First, to support the Open Graph Protocol, all you need to do is add some RDFa-formatted metatags to the HEAD of your HTML pages (as this example demonstrates, from IMDB):

Simple right? Indeed.

And from the looks of it, pretty innocuous. Structured data is good for the web, and I’d never argue to the contrary. I’m skeptical about calling this format “open” — because it smells more like openwashing from here, but I’m willing to give it the benefit of the doubt for now. (Similarly, XAuth still has to prove its openness cred, so I understand how these things can come together quickly behind closed doors and then adopt a more open footing over time.)

So, rather than using data that’s already on the web, everyone that wants to play Facebook’s game needs to go and retrofit their pages to include these new metadata types. While they’re busy with that (it should take a few minutes at most, really), won’t they also implement support for Facebook’s Like button? Isn’t that the motivation for supporting the Open Graph Protocol in the first place?

Why yes, yes it is.

And that’s the carrot to convince site publishers to support the Open Graph Protocol.

Here’s the rub though: those Like buttons only work for Facebook. I can’t just be signed in to any social web provider… it’s got to be Facebook. And on top of that, whenever I “like” something, I’m sending a signal back to Facebook that gets recorded on both my profile, and in my activity stream.

Ok, not a big deal, but think laterally: how about this? What if Larry and Sergey wanted to recreate PageRank today?

You know what I bet they wish they could have done? Forced anyone who wanted to add a page to the web to authenticate with them first. It sure would have kept out all those pesky spammers! Oh, and anyone that wanted to be part of the Google index, well they’d have to add additional metadata to their pages so that the content graph would be spic and span. Then add in the “like” button to track user engagement and then use that data to determine which pages and content to recommend to people based on their social connections (also stored on their server) and you’ve got a pretty compelling, centralized service. All those other pages from the long tail? Well, they’re just not that interesting anyway, right?

This sounds a lot to me like “Authenticated PageRank” — where everyone that wants to be listed in the index would have to get a Google account first. Sounds kind of smart, right? Except — shucks — there’s just one problem with this model: it’s evil!

When all likes lead to Facebook, and liking requires a Facebook account, and Facebook gets to hoard all of the metadata and likes around the interactions between people and content, it depletes the ecosystem of potential and chaos — those attributes which make the technology industry so interesting and competitive. It’s one thing for semantic and identity layers to emerge on the web, but it’s something else entirely for the all of the interactions on those layers to be piped through a single provider (and not just because that provider becomes a single point of failure).

I give Facebook credit for launching a compelling product, but it’s dishonest to think that the Facebook Open Graph Protocol benefits anyone more than Facebook — as it exists in its current incarnation, with Facebook accounts as the only valid participants.

As I and others have said before, your identity is too important to be owned by any one company.

Thus I’m looking forward to what efforts like OpenLike might do to tip back the scales, and bring the potential and value of such simple and meaningful interactions to other social identity providers across the web.


Please note that this post only represents my views and opinions as an independent citizen of the web, and not that of my employer.

51 Comments

  1. at 5pm on Apr 22nd # |

    Couldn’t agree more Chris – my full thoughts here: http://bit.ly/bnZo0j

    Facebook is bringing to reality a lot of the functionality we hope for in the social web (My friends select a favorite restaurant on OpenTable – Hey! Now Yelp is informed by these Likes), but at the cost of service lock-in and a unidimensional social presence.

    Lock-in is anathema to innovation and a unidimensional social presence is anathema to meaningful human interaction. When Zuck talks about “Building The Social Web Together” he needs to recognize that he can’t do it alone!

  2. at 6pm on Apr 22nd # |

    Landed here via openlike – well put post. In a way, respect to Facebook for launching this – it’s a no-brainer in a lot of ways – FB can later correlate all those likes back to it’s profile data (which is even more value than just knowing about “likes” on aggregate. But, as well as being smart for FB to do, the model really is “evil” as you point out.

    It’s not just that the provision of the liking service goes through a single provider (thus driving signups maybe and increasing stickiness) but also that the valuable data about correlations is only available in full to one provider. FB might choose to expose that data but it will always be more limited than what’s available internally. THAT data is potentially the “social” parallel to links that the current web has.

  3. at 6pm on Apr 22nd # |

    I posted the below comment on Jesse Stay’s blog. I am interested to hear your thoughts Chris:

    I like to see the web as a repository of information where user,like me, can get the most relevant information for my needs. Facebook effort is like a double edge sword.On one hand it allows developers to give user more relevant information. Others view that FB is trying to be the authority or even worst single point of failure on the net.

    For me I like to have more than 1 options in my hand on how to get things from the web. There is nothing more painful than just having only 1 tool that can do the job for you. If that tool is broken, I need to wait for the service company to come over at my place to fix it. I want to have options or other alternative tool to do my task.

    Let take a look at Open Graph by facebook. It is great ! But it will be even better if we can have more options for the social web. There are other alternative to Open Graph, e.g: Google’s Social Graph, that leverages open standard such as XFN or FOAF. I look forward for other open standard adopters to push the popularity of Google’s Social Graph. Having more option can be handy when 1 option fail us. What do you think about Social Graph ?

  4. max engel said
    at 7pm on Apr 22nd # |

    i’m actually excited about the Open Graph Protocol. while i do wish it had embraced more existing community work, i am glad to see it leverage hCard/PoCo for location.

    ultimately, i think this broadened the discussion about the importance of structured data, and hopefully more people will be motivated to help make this schema better.

    theoretically, any XAuth-participating sharing site could choose to read the schema, so a Buzz button (be it Yahoo! or Google ;-) ) could parse the same info.

    however, it certainly is unfortunate that this was developed completely in a vacuum. i agree that it isn’t the same degree of openness to say, “we built this great thing and you can use it too,” as it is to say, “we want to build this great thing and we should all do it together.”

    i’m hopeful that this is a true step in the right direction towards open collaboration and standardization for Facebook.

  5. Monica said
    at 8pm on Apr 22nd # |

    Chris,
    Are you saying that Facebook chose new meta tags to make people edit their web pages so they find it convenient to add a Like button in the process ?
    I actually think the reverse is going to happen. People will want to add Like buttons and in the process of doing so they will enrich the web with semantics. Publishing compelling information about your site to a large audience is really good motivation.

    Any site can read the meta tags. I am sure Google will and MySpace and anyone else wanting to present a richer experience for end users in an ocean of links.

    I don’t think Facebook is opposed to using microformats either. There are many microformat lovers inside FB who will be championing this – so far adoption is not too widespread

    Also I definitely think its important for people to NOT present this as a standard. Its a just a proposed protocol and we are encouraging a transparent process now that our contribution is announced.. this is the opportunity to work together and enrich the web.

  6. at 9pm on Apr 22nd # |

    At the beginning of the essay, you said you’re willing to give the Open Graph Protocol the benefit of the doubt, but by the end you concluded it was evil.

    We are really discussing two things here:
    1) the Open Graph Protocol, an open technology which any content website can implement and which any identity provider or crawler can consume.
    2) the Facebook Like button, a product which consumes Open Graph data and publishes connections back to Facebook.

    You’re correct that Facebook Like acts as the carrot to persuade site publishers to join the Open Graph. What’s wrong with a compelling carrot? Once this graph exists, it is a graph that all participants of the open web can benefit from. As you said, structured data is good for everyone, and I think it’s awesome that we finally have a product that is encouraging people and companies to structure their data in an open way.

    Facebook Like is just the first consumer of the Open Graph; it is just one product illustrating the possibilities of pervasive semantic data. I hope (and expect) that Google’s Buzz button, Twitter’s Retweet button, the OpenLike effort, and others will all build similar tools that parse and consume the same Open Graph. I’m also looking forward to see what new and interesting ideas sprout up to take advantage of all this semantic data.

  7. Paul Walker said
    at 11pm on Apr 22nd # |

    wow, couldn’t disagree more.

    It’s a product. It’s no different than asking sites to put a tag for adwords that made Google and your current salary what it is today. Did you complain that the JS wasn’t an open standard that could communicate to any ad provider?

    ~Paul

  8. at 11pm on Apr 22nd # |

    Chirs
    Great post…..I agree with all that you have said…anyone thinking that the FB move is not evil are lambs begin ready for slaughter… and any developers working inside of FB or as third parties to build extend its services are complicit with closing off the internet and shredding to bits notions of users rights…..We can no longer focus of the evils of the “Facebook”….Zuck came up with the evil ideas but developers that know better implemented them….

    http://www.factoetum.com/factoetum/List_of_Technology_Icons

  9. Ken said
    at 12am on Apr 23rd # |

    You’re right it is _not_ open open but you know open is the new closed.

  10. jda said
    at 4am on Apr 23rd # |

    I’m not sure I buy into the idea that this is Facebook domination, especially with respect to likes. The fact is that someone has to do it first – and that someone was Facebook. When I look at openlike, I see a Nascar issue. If anything, I think this all ends up as a net positive, after more services start offering similar buttons, which leads to Nascar interfaces, which leads – hopefully – to Xauth implementations. The only reason that this all has to go through Facebook right now, is because they have an early advantage. Do you think in your analogy you could have just said, if Larry and Serge were to redo pagerank they just shouldn’t because someone else is already indexing web content… And as far as retrofitting pages, the same goes for anyone that wants to use microformats, or activitystrea.ms or salmon. I can’t buy into that argument.

  11. at 5am on Apr 23rd # |

    I’m not so keen on putting the new RDFa tags into HEAD section of the page. In general it is best to have the metadata inside the actual page contents so it also remains human-readable, just like it is now possible with both microformats and RDFa.

    By putting them to HTML HEAD you make people write same info separately for humans and computers, and that kind of duplication is never good.

  12. at 5am on Apr 23rd # |

    Great summary. Thanks!

  13. JP Kab said
    at 6am on Apr 23rd # |

    Monica,

    You are exactly on point. The fact that the RDFa vocabulary used for the tags is a creation of Facebook is irrelevant. The linked data standards designed by the W3C (RDF/OWL) are meant to accomodate merging multiple vocabularies, no matter what silo they were built in. Any, and I do mean ANY, incentive to include structured data mark up in a web page is good for the web. The same tags used to power Facebook’s like button can be used by Drupal’s upcoming 7.0 release to populate content, or can be indexed by structured data applications/search engines.

    The W3C had no illusions about large groups of people getting together and creating open vocabularies of tags for the greater good. They designed RDF/OWL to allow silos of structured data to be connected, so that if you call it a lorry and I call it a truck, an inference engine driven application can use an OWL vocabulary to treat them the same.

  14. at 6am on Apr 23rd # |

    Is the overall effort really so different from Google’s own?

  15. at 6am on Apr 23rd # |

    It is, at least in one important way: there should be more than one identity provider for people to choose from. Facebook has said that they want to be the best identity provider on the web, and I’m fine with that aspirational objective. But the approach with auto-logins, in concert with the Like buttons feels starkly different than how Google might approach this problem.

  16. at 7am on Apr 23rd # |

    While I share your discomfort at the Facebook worldview, I’m also uncomfortable with your dismissal of Open Graph as “evil”. If you look at it with a bit perspective, it’s more a step away from the evil of a closed social graph facebookverse designed to suck in the internet and towards the creation of a self-sustaining graph-organism.

  17. at 7am on Apr 23rd # |

    +1: http://blog.woobling.org/2010/04/where-are-open-edges.html

  18. Ali said
    at 7am on Apr 23rd # |

    “And on top of that, whenever I “like” something, I’m sending a signal back to Facebook that gets recorded on both my profile, and in my activity stream”

    you don’t need to like it, *any* page you visit while still holding a facebook cookie is being reported back to facebook (since it’s an iframe) – they don’t need to get your like, your visit alone is worth a lot.

    ps. any page = any page using one of those new fancy facebook widgets that is

  19. James said
    at 7am on Apr 23rd # |

    Interesting take. Hadn’t thought of it like this. I like the article. If you had a facebook “like” button, I would push it. :)

  20. at 8am on Apr 23rd # |

    I agree that Facebook’s goals are ultimately damaging to openness, but in my opinion the work they are doing will ultimately benefit everyone in the long run. OK, so it would be nice if the Open Graph format didn’t require adding meta to the head, and it would also be nice if it weren’t likely to come with a Facebook-exclusive Like button.

    However, the fact that sites have a Facebook carrot encouraging them to utilize more structured data is going to be good for everyone, and Facebook will lose its monopoly once the browsers themselves start getting into the identity management and semantic parsing game.

    Full thoughts here: http://intridea.com/2010/4/23/facebook-is-the-private-beta-of-the-semantic-web

  21. Prasanna said
    at 8am on Apr 23rd # |

    Nice Article. You should put a ‘like’ button on this one. I am sure it will get clicked a lot.. ;)

  22. srw said
    at 9am on Apr 23rd # |

    Why use the ‘open’ word when the real word is ‘closed’? Except for search and some details, current FB API is better from the development point of view (versus FQL) but you can’t even access the number of members of a group.

  23. Tom said
    at 9am on Apr 23rd # |

    One way to counter this might be to ‘like’ everything. Then there would mo signal, only noise.

  24. Edwin Aoki said
    at 10am on Apr 23rd # |

    The arguments for and against openness and a company hijacking the web have been around almost since the first days of the web itself. Whether it was Netscape’s tag or Microsoft’s Channel Description Format, the dominant players have always looked for ways to encourage the adoption of protocols and formats that would advantage their position. In modern times, as a broader ecosystem has developed, it’s increasingly the case that these companies have also acted in ways that would also be good for the web. Open (or at least published) APIs and services, participation with other groups in standardization efforts (admittedly some after the fact) and greater transparency have the potential to open up these efforts for the greater good. I think it’s generally positive that companies are incented to consider the ecosystem when developing products. But we should be clear – most companies don’t do something altruistic that they think will ultimately harm them. And companies with market dominance will use that dominance to advance their goals. Take the recent announcement that PageRank would take page load performance into account. That’s probably a good thing for the web, but it’s certainly the case that it creates an artificial market pressure on sites to improve performance – if you can’t be found on Google, you might as well not be on the web. Or Apple’s recent stance against Flash. Arguably, encouraging the adoption of HTML5 over proprietary protocols is a good thing and Apple’s leadership (or dominance, depending on how you look at it) in this area will ultimately be positive. But it’s certainly not (just) because Apple is all about standards.

    Today, Facebook dominantes the Open Graph (which is probably unfortunately named, at least for now). Google dominates Android, a nominally open environment. There are similar analogies for a number of open (again, or at least published) protocols. The good news is that now, there are advocates for openness at each of these companies, and that these conversations are happening, not just behind closed doors in standards bodies, but here where we can all participate and help shape the discussion.

  25. Ed Borasky said
    at 10am on Apr 23rd # |

    Well – I’m deleting my Facebook account. And probably my Google account and my SpamCop account as well. That will leave me with Yahoo, LinkedIn and Twitter only among the major players, and I’m guessing I could live without Yahoo / Delicious / Flickr as well. I need a Microsoft account to get updates for my Zunes or I’d delete that one too.

    I don’t trust Google. I don’t trust Facebook. SpamCop doesn’t scale and they don’t have an 800 number I can call when there servers are down.

  26. at 12pm on Apr 23rd # |

    Chris,

    I appreciate your wise voice of caution regarding Facebook’s walled-garden approach to their Open Graph Protocol implementation of RDFa. Sure, it could even be rhetorically–termed “evil.” But, despite FB’s prominent position in participatory social graphing, I, like Fred Wilson and Albert Wenger, don’t think FB’s version has much chance to be the “One Graph To Rule Them All” http://bit.ly/9El9Yi

    Emergence of a Linked Data Meme.

    Ironically, the attention afforded the metadata standards of Open Graph is likely to raise the prominence and engagement of new people in open web, Linked Data, FoAF and Open ID implementations. Interestingly, after deciding to leave such standardizations to the marketplace, there is evidence the Twiter development team was inspired by the Open Graph announcement to look to at OG and other structured data approaches for their new annotations feature. Yes, another semi-closed garden approach; I raise the instance only as a small example of what I see happening in the last week: an emergence of a Linked Data meme.

  27. at 1pm on Apr 23rd # |

    Good points there. I guess ‘open’ would generally apply to being ‘all embracing’ amongst others on the Web but Facebook has a different version to the meaning of Open and that is ‘Domination’ and one that is absolute.

    I remember your keynote in Finland [Mindtrek 2009] and agree that there should be a choice for people to choose from various identity providers. And that is common sense.

    Would anyone put all their money in one place – irrespective of how trustworthy it claims to be or already is?

  28. Andryo said
    at 3pm on Apr 23rd # |

    Well to be fair, not just facebook that can have advantages over this. Small websites/brands can have it as well.

    Most of the readers here already have facebook account anyways, right? So what’s the different with tweeting in which you need a twitter account too…

  29. Erik Petersen said
    at 4pm on Apr 23rd # |

    Interesting discussion. I’m a little on the fence right now. There are certainly big players duking it out for big chunks of a fat pie but I find it hard to be critical of this Facebook effort while at the same time ignoring the very *same* efforts on the part of Google (and even Microsoft if they had a clue.) Apple is doing it (trying) with iTunes and the list goes on.

    What is “the web” exactly? Seems different people have different ideological ideas about that. What I do know is the web is a true free market. Look at MySpace. Hero to zero in 1 year flat; not because they were out-spent on advertising or shelf space but people simply didn’t like what they were doing anymore and a more attractive solution (Facebook) came along as a viable alternative.

    If Facebook succeeds in pissing off their customers and treating them like cattle they too will circle the drain faster than you can blink. If on the other hand people like this new idea of directly linking the greater web into their profiles and Facebook treads lightly and respects their users they will reign supreme (for a while anyway).

    The so-called too-big-to-fail titans of the web were all preceded by same. AOL, Yahoo, MySpace where are you now?

  30. Dan Birchall said
    at 7pm on Apr 23rd # |

    Like a lot of other people who’re likely to be commenting here, I’m a techie-slash-creative sort (not slash in the fanfic sense, slash in the -cum- sense, no, not cum in the fanfic sense…).

    Why do I use social networks at all? Well, they’re a little less work than running my own site, worrying about code, worrying about host uptime and all that stuff, and since nobody pays me to run my own site, I’ve gotten lazy over the last several years.

    But in the back of my mind, there has always been a little voice saying that as soon as I can find the time, I really do need to set up a proper site of my own again, because it really is about ownership of my own data, my own code, and my own creations.

    And every time someone – Facebook or not – pulls some dumb stunt like this, that voice gets a little bit louder.

    Once I listen to it, will I close or deactivate my Facebook account? Of course not! I’ll still be on Facebook – probably with, as Danny Sullivan suggested elsewhere, a fan page (I’ve had mine reserved for ages, but haven’t launched it yet). I’ll still maintain some kind of token presence on all kinds of other social networks as well. I might even get back onto Twitter.

    Why? Because it’s not like those things are going to simply go away, and astoundingly, there are people out there who don’t have the sense to just google me. :)

    But as far as giving Facebook my content, my thoughts, my consciousness, my personal info, and the private social structure of my life? Yeah, I think I’m just about done with that.

  31. at 11pm on Apr 23rd # |

    Considering Facebook is the most visited site in the US, businesses with a marketing budget for social media should be jumping all over this.

  32. at 3am on Apr 24th # |

    Where is the FB like button on this page? I really like it and wanted to share it and make it part of the algo to get relevant news for my profile :)

    I think the open graph is a good thing (period). Its another player running along side of rdf, microdata and microformats. Any search engine or app can crawl pages and index that meta data. That is good (but can also get spammy!).

    What I dont get from your post is how the FB like button is different from Google Buzz button or Orkut share button (and plenty other buttons out there). They both have the same goal in mind and Google can also use it (if it doesnt already) to understand what is relevant content for me.

    Also, FB API allows you to get the data from all likes of a user but only if the user gives an app permition and that is what I believe people, me included, are worried about. But I share the same worries with all the other share, like, retweet, etc buttons.

    I believe the big issues here is the lack of a standard both for meta data and share buttons.

  33. at 11am on Apr 24th # |

    The open graph data is almost certainly a good thing as long as we don’t end up with tons of different formats doing the same thing and we end up having to fill HTML header with tons of shrapnel – I agree with Bret Goldman’s split between that and the like button and personalisation tools.

    What’s happening here is the true emergence of an identity layer over the Web which Facebook is beginning to leverage for more and more value add for it’s users and itself. This makes perfect sense, but it’s naturally worrying if it leads to a web were large amounts of crucial data are only available to one party. As I said in an earlier post I think that facebook’s advantage is not subsequently getting information back from the like buttons, but in what it is able to tie that information too – all those user profiles – something they’ve built up by having a very compelling service.

    I really like the idea behind http://openlike.org/ and hope it will break through – this would be awesome information for new and old apps alike. However, I think the like issue is just symptom of what will happen as “identity providers” get big and flex their muscles.

    Facebook is certainly taking it’s turn to move the web forward which is a great thing – Facebook sat up and said the web would be better for us if this meta-data was there, likely it will be better for everybody else too. The evilness is in the increasing capture of more and more data which isn’t visible / exploitable by others – on the other hand plenty of other companies would do this given half a chance – Facebook would be dumb not too.

  34. at 4pm on Apr 24th # |

    > Facebook gets to hoard all of the metadata and likes around the interactions between
    > people and content, it depletes the ecosystem of potential and chaos

    Chris, it seems like you’re saying Facebook should release all the data about who likes what and what sites they visit. I think it’s worth pointing out other services that have equally vast amounts of data that aren’t shared publicly.

    For instance, Google knows what things people search for, what they click on when they search, which sites they visit (through AdWords) and how long they stay on those sites. Undoubtedly that data would be invaluable for the next search startup, and by keeping it to itself Google depletes the ecosystem of potential.

    I assume that you’d encourage Google to release all the data that it currently “hoards”?

  35. at 6pm on Apr 24th # |

    Luke, that’s a good point, and I’m glad you pointed it out. It’s a little bit of an indirection, but it’s a fair question.

    To answer it, I would say yes, actually, I do think that Google should make such metadata available to the owners of that data. Through efforts like the Data Liberation Front we’re making progress but haven’t done quite enough I’d say.

    I think the data that I’d be very interested in having access to — from a large number of providers — would be the weighting data that is used to infer proximity and relevance of social ties.

    For example, many people (with whom I’ve spoken recently) suggest that Facebook is where they connect with all their 3rd grade friends — and that it’s not really where their real friends are. But surely Facebook is able to detect this, and is also able to look at how often people interact and then weigh those interactions so that only recent or relevant people show up throughout Facebook’s interface.

    That kind of weighting-over-time is what I’m talking about, and why I think that the Like button is so powerful when attached to a Facebook account. When I talk about the metadata — I’m not just talking about the static attributes that define an object — I’m talking about the weighting values that determine how often and which objects and elements should be shown and promoted to a given user.

  36. at 11pm on Apr 24th # |

    Thanks for the response, Chris. I don’t think that my point is a misdirection – if the core of your argument says that Facebook should release more data, or somehow it’s bad to release an open standard that benefits Facebook – then I think it’s fair to point out that your employer does much of the same thing when it comes to search data.

    You raises one of the most interest questions of our time – who owns the data about an edge? If I say I like a CNN article, then who owns that? Is it the user, or CNN, or Facebook?

    If it’s the user, then as it turns out, I DO have access to that – it shows up on my profile, and any platform app that I install can grab that info and display it to me or use it to inform their own results. With an OAuth access token, try accessing https://graph.facebook.com/me/likes?access_token=…. to see what I mean. (grab an access token from any link on this page: http://developers.facebook.com/docs/api )

    The individual weighting of edges in the social graph (the EdgeRank, as Techcrunch called it), well, that’s a little more complex. The analogy at Google would be letting site owners view their own PageRank – something which Google has refused to do (as far as I know).

    Also, it’s worth pointing out that webmasters have been changing their HTML tags for years to accomodate big companies – it’s just that for most of the last decade, Google has been the one to please. But it’s not a bad thing – when Google tells web owners to use text instead of images, it benefits Google – but also has nice side benefits, like increased accessibility. Similarly, any markup added to match the Open Graph Protocol will benefit Facebook (obviously) but also any other service that chooses to scrape their site. for instance, I hope that Google uses the Open Graph markup to improve search results.

    In any case, I appreciate your post and agree with the general sentiment of openness, but just wanted to keep the debate honest.

  37. Rick said
    at 11pm on Apr 24th # |

    awesome summary, i can’t wait to see what happens next…

  38. Paul Walker said
    at 12am on Apr 25th # |

    Watching Google and FB argue over openness and privacy could put this blog in history.

    Profit motive is good boys, keep it up. Both companies further our lives and I am happy for it.

  39. at 10am on Apr 26th # |

    GReat post Chris. I wrote about Facebook’s announcement with excitement recently and several people have pointed me to your post as an alternative view. I think there’s good and bad in Facebook’s announcements, and if a better social web is the goal, then we might have to wait and see, but I’m glad to see so many great thoughts about the future of the web (including Facebook and Google) such as yours.

  40. at 2pm on Apr 27th # |

    Evil? Really?

    You don’t have to use it. Nobody is forcing anybody to use it… If you don’t think your site can benefit from it. Don’t use it…

    To think that it’s evil.. is pure comedy, or it it just simple jealousy???

  41. at 3pm on Apr 27th # |

    As I said in my follow up post, what’s evil is a social web where there is only one identity provider.

    Facebook isn’t evil; they’re just acting in their own self-interest, as any business does. The Open Graph Protocol combined with “instant personalization” and like buttons benefit them, but there’s nothing evil about that. I just want the option to choose a different identity provider to serve these kinds of features and functionality.

  42. at 4pm on Apr 27th # |

    But Chris…. I think you assume too much power to them… They are not an identity provider at all, only a channel for personal expression… It’s the individual human who provides the identity, not Facebook.

    For example…. I know my FB friends are not my REAL friends.

    Real friendship and real identities can never be trivialized into one organization’s website… assuming so, is just silly.

  43. at 4pm on Apr 27th # |

    Perhaps I’m too deep in this area, but unless you control what goes on your public profile, and can edit the HTML of that page, and make it *actually* express something about you, I’m going to disagree.

    While it’s true that your “real friends” exist beyond Facebook, many people increasingly rely on these kinds of services to connect and stay up to date with friends, family, and colleagues. That’s not a bad thing either, it’s one of the successes of the social web.

    The problem is that centralizing so many digital profiles in the hands of a single provider creates opportunities for abuse, and a lack of meaningful choice in the marketplace. I understand how and why things get to be this way, but I don’t think they must be this way.

    In comparison, email is a distributed social communication channel. Social networks are not, but could be. If your email account was deleted or hacked, you could get another one and continue interacting with all your email friends. If your Facebook account were to be deleted, you would be prohibited from interacting with your Facebook contacts.

    Providing more choice of “identity providers” in the marketplace would potentially act as a buffer against the negative consequences of such a situation.

  44. Don Park said
    at 1pm on Apr 28th # |

    The open graph is getting mixed up with the like button. Open graph on its own seems harmless enough. As has been commented already its the wrong approach. A microformat can describe the relevant bits of content in the html body where it belongs. how many pages represent a single object like movie? Seems more likely a page will have a dozen objects in it.

    I wont add a facebook like button to my pages because it is too centralized, I agree with the post there. Openlike should be called multilike because its not any more open, it gives more closed choices.

    What I see is a button that does a POST to any destination with an activitystream snippit that says . Use the activity stream format for writing as well as reading and authenticate with OAUTH.

  45. Don Park said
    at 1pm on Apr 28th # |

    (correction on last post)
    activitystream snippit that says I LIKE (Object URL).
    (thats what i get for putting words into angle brackets)

  46. at 12am on Apr 29th # |

    Nice post Chris – I think this is bad news all round and would also point out that Facebook is filling any gaps by auto-creating pages for almost any ‘liked’ term and importing the Wikipedia data under Creative Commons License. As far as I can see there are adverts on those pages, which must be a breach of the license. At the very least Jimmy Wales must be spitting feathers for such an abuse of his altruistic model… I wrote a blog –

    ‘Your Privacy Got Raped, Did You ‘LIKE’ it?

    here – http://blog.famebook.com/famebook/2010/04/your-privacy-got-raped-did-you-like-it.html

  47. Tim said
    at 1am on May 4th # |

    What a very strange article. Less of the paranoia please.

    So Facebook would like it if you used RDFa in your webpages? That is a good thing. The more RDFa in use the better, and I’m very glad to see it being adopted by a “major player” such as FB.

    What I do not understand is why they’ve gone against best semweb/linked-data principles and created their own schema which merely duplicates predicates from existing ones. We already have well-known dc, foaf, geo, geonames and other schemas. This is wasteful. The protocol also uses predicates that have a range of literal values (text strings, floats) rather than URIs, which denies the opportunity to disambiguate meaning[0] or link between entities.

    [0] E.g. if you’re interested in “photography” and I’m interested in “Photography” then that’s a mess; if we’re both interested in http://dbpedia.org/resource/Photography, identified as a URI, then we all know exactly what we mean.

  48. Mark Sigal said
    at 11pm on May 5th # |

    Great piece. I would assert that even if Facebook turns this into an open platform across all social services and content sources, it still forces the larger question of whether consumers want their micro-moves tracked if they can’t manage it in a globally informed and brain-dead simple fashion.

    As such, it raises Privacy Rights and Policy Policy questions.

    Specific to Facebook, they have been pretty clear through their actions (Beacon, ScamVille, Privacy Putsch) that they think that consumers should “get over” their privacy concerns. Plus, their culture is to act first and then backtrack later, if necessary, something that I blogged about in:

    Is Facebook a Brand that You Can Trust?
    http://oreil.ly/8xySuW

    At the same time, you have to give these guys their props. They have executed on their platform about as well as Apple has theirs.

    Check out the post, if interested.

    Cheers,

    Mark

  49. Matt said
    at 11am on May 7th # |

    Facebook has made some very questionable decisions are privacy and default settings. In December with the defaults switch-a-roo (surprise!) and now with the new connect (still limited to 3 trusted partners so let’s not get too freaked out yet…)

    But the like button is not evil.

    Your analogy to PageRank is flawed – they are different scenarios and mechanics. PageRank is based on automated crawl, whereas “like” is user-initiated syndication. Pull vs. Push. PageRank has no need for user authentication – the “entity” is the page itself, which is already known by its URL. The “like” scenario necessitates an authenticated state to tie the action to a user. (Yes, it does not necessarily need to be Facebook authentication, but do I really need to be logged into Google to use Buzz, etc?)

    Yes, the data then becomes Facebook’s to build “recommended lists” and “top pages” and all that. But last time I checked, Google wasn’t making any of its crawl or click data public either.

    I empathize and agree with your sentiment that “identity is too important to be owned by one company.” But so are your search results.

  50. at 1am on May 10th # |

    I wrote a simple example (~30 lines of PHP) showing how to get a user’s likes via the Graph API: http://gist.github.com/395844.

  51. at 3pm on Jul 27th # |

    @ Jan: “Facebook is filling any gaps by auto-creating pages for almost any ‘liked’ term and importing the Wikipedia data under Creative Commons License.”

    I know some authors who are tremendously upset by this practice because it infringes on their rights by creating “authority” pages about their characters without their input. In this way, they can earn advertising revenue by leveraging someone else’s intellectual property. As always, the law is lagging behind actual internet practices. It will be interesting to see whether Facebook eventually gets sued over this. I wouldn’t be surprised if they did–deep pockets.

24 Trackbacks

  1. […] via FactoryCity » Understanding the Open Graph Protocol. […]

  2. […] FactoryCity » Understanding the Open Graph Protocol […]

  3. […] who is now working for the search company. Messina mentioned the Open Like effort approvingly in a post about Facebook’s social graph, in which he says: When all likes lead to Facebook, and liking requires a Facebook account, and […]

  4. Facebook Like on Apr 23rd at 4pm

    […] partilhar essa informação via Facebook…Por outro lado… há também quem levante as preocupações sobre o “poder” que isto dá ao Facebook. Questões pertinentes que poderão preocupar alguns, e ser completamente ignoradas por outros […]

  5. […] drugiej strony pojawiaj? si? pytania o prywatno?? i otwarto?? projektu. Jak podkre?la znany bloger Chris Messina, jest rzecz? bardzo niebezpieczn?, gdy ogromna ilo?? danych staje […]

  6. […] Le principe est de permettre à qui veut de mettre un bouton ‘I Like’ sur chaque page du web. Chris Messina explique en détail le fonctionnement sur son blog. […]

  7. […] for Palm -sources – Reuters For Web’s New Wave, Sharing Details Is the Point – New York Times Understanding the Open Graph Protocol – FactoryCity CONFIRMED: Fusion Garage SOLD ONLY 64 JooJoo Devices – UNEASYsilence Dear […]

  8. […] Specifically, most of them are targeting the new Like button that is appearing everywhere on the web (including on TechCrunch). It’s an obvious target as it’s the most visible part (at least so far) of the Open Graph protocol. Investor/Hunch co-founder Chris Dixon is leading the effort for a new OpenLike button (though he wants someone else to be in charge). And Google’s Open Web Advocate, Chris Messina, has already ripped apart Facebook’s Like button in a blog post. […]

  9. […] 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 […]

  10. […] who now works at Google — about how the company’s “open” protocol and API weren’t really open. The bottom line, Messina said, is: [I]t’s dishonest to think that the Facebook Open Graph […]

  11. […] Chris Messina points out in this post, the new Facebook global Like button and the Open Graph Protocol benefit Facebook more than anyone. […]

  12. […] already happening around the subject – not just the Open Graph and the issues related to it (open web ramifications, privacy, etc), but how we can indeed take advantage of it. […]

  13. […] of these detractors, like open source activist Chris Messina, make a compelling point. Facebook claims that it is “opening” the social graph, the […]

  14. […] Chris Messina also chimes in with this excellent post identifying the problem of a single point of failure for web content. Categories: Uncategorized […]

  15. […] Understanding the Open Graph Protocol (tags: facebook identity google standards social data API privacy rdf semanticweb future critique socialmedia) […]

  16. […] Messina: http://factoryjoe.com/blog/2010/04/22/understanding-the-open-graph-protocol/ But, here’s the rub: rather than using data that’s already on the web, everyone that wants to […]

  17. […] all, Facebook technically owns the protocol, the data, the access. But on the other hand, they’re giving the web a gift—a new understanding of the […]

  18. […] the (in)valid “property” attribute of the meta tag, and of course many opinions on Facebook’s real reasoning for the protocol. Though, since it is backed by Facebook, I don’t see it going […]

  19. […] commercial gain, and there is questioning by those who know much more about this than I do about  how “open” Open Graph really is,  but Facebook has gone a long way to illustrating to the mainstream the concept of the semantic […]

  20. […] part of their recently announced Open Graph Protocol, Facebook launched a number of new plugins to make integrating with their site easier. […]

  21. […] 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 […]

  22. […] already happening around the subject – not just the Open Graph and the issues related to it (open web ramifications, privacy, etc), but how we can indeed take advantage of it. Traffic In a recent article we […]

  23. […] from integrating Twitter results — although since Facebook’s Open Graph protocol is theoretically an open standard, there is the potential for Google to use that to pull in the network’s results in the same […]

  24. […] WebMonkey’s Michael Calore reports, even as it sucks the data into a closed community. But those standards are used almost exclusively by Facebook, and ignore the work that’s been done by others to create universally understandable […]