The interoperability of social networks

Google recently added a caustic warning message when users attempt to export their Google Contacts to Facebook:

Hold on a second. Are you super sure you want to import your contact information for your friends into a service that won’t let you get it out?

Facebook allows users to download their personal information (photos, profile info, etc) but has been fiercely protective of the social graph (you can’t download friends, etc). The downloaded data arrives in a .zip file – hardly a serious attempt to interoperate using modern APIs (update: Facebook employee corrects me/clarifies in comments here). In contrast, Google has taken an aggressively open posture with respect to the social graph, calling Facebook’s policy “data protectionism.”

The economic logic behind these positions is a straightforward application of Metcalf’s law, which states that the value of a network is the square of the number of nodes in the network*. A corollary to Metcalf’s law is that when two networks connect or interoperate the smaller network benefits more than the larger network does. If network A has 10 users then according to Metcalf’s law its “value” is 100 (10*10). If network B has 20 users than it’s value is 400 (20*20). If they interoperate, network A gains 400 in value but network B only gains 100 in value. Interoperating is generally good for end users, but assuming the two networks are directly competitive – one’s gain is the other’s loss – the larger network loses.

A similar network interoperability battle happened last decade among Instant Messaging networks. AIM was the dominant network for many years and refused to interoperate with other networks. Google Chat adopted open standards (Jabber) and MSN and Yahoo were much more open to interoperating. Eventually this battle ended in a whimper — AIM never generated much revenue, and capitulated to aggregators and openness. (Capitulating was probably a big mistake – they had the opportunity to be as financially successful as Skype or Tencent).

Google might very well genuinely believe in openness. But it is also strategically wise for them to be open in layers that are not strategic (mobile OS, social graph, Google docs) while remaining closed in layers that are strategic (search ranking algorithm, virtually all of their advertising services).

When Google releases their long-awaited new social network, Google Me, expect an emphasis on openness. This could create a rich ecosystem around their social platform that could put pressure on Facebook to interoperate. True interoperability would be great for startups, innovation, and – most importantly – end users.

* Metcalf’s law assumes that every node is connected to every node and each connection is equally valuable. Real world networks are normally not like this. In particular, social networks are much more clustered and therefore have somewhere between linear and exponential utility growth with each additional user.

http://twitter.com/MikeDuda MikeDuda

Chris, sorry if this is off topic, but thoughts on RockMelt as it relates to social networks embedded in a search-oriented provider?
http://twitter.com/MikeDuda MikeDuda

Chris, sorry if this is off topic, but thoughts on RockMelt as it relates to social networks embedded in a search-oriented provider?

http://www.cdixon.org chris dixon

I only tried RockMelt briefly. Seems cool. Organic growth for a browser in 2010 seems hard, but perhaps they have natural large allies like FB that will help them get distribution…?

http://buzzvolume.com/ marionogueira

Upcoming desktop app stores (like the one Apple is close to release) may have a positive impact on distribution for RockMelt.

Pingback: Tweets that mention cdixon.org – chris dixon's blog / The interoperability of social networks -- Topsy.com
http://twitter.com/teisenmann Tom Eisenmann

Chris: There’s some academic work that shows that networks are more likely to mutually agree to interoperate when: 1) a market has matured and growth has slowed, and 2) the networks are roughly equal in size. Examples include SMS, ATMs (Cirrus and Plus), and, as you point out, IM. In mature, slower growth markets, the “scale begets more scale” advantage that comes with strong network effects is attenuated. Likewise, if market shares are at parity, then neither side has to worry about share loss, post-interoperability. Social networks may be maturing, but no one comes close to Facebook in scale, so my bet is they’ll continue to resist interoperability vigorously. But I agree that Google will call for openness in this non-core market.

http://www.cdixon.org chris dixon

Thanks Professor, interesting that here the theory and empirical studies seem pretty aligned. Seems like the business model is also perhaps a factor: SMS and ATMs (as I understand them) get paid per transaction, so encouraging more transactions by interoperating is a countervailing force to protecting the network. Social networks currently make most of their money on advertising – as long as that remains the case I wonder if interoperating becomes less desirable..?

Anonymous

Interoperating makes more aspects of your life visible to the ad network, allowing better targeting. Most of the time they don’t need more refined data (exact age, rather then (5-year range) as much as more context (siblings, children).

I’m not sure about what articles Tom E. is referring to, but both results seem over-interpretation of some assumption. If by mature you mean with less innovation, or better understanding, then interoperability is easier; if you mean with more specialised conventions, then it’s both harder and less interesting or threatening. However, the likelihood that a network gets another one does not depend as much in their size as it depends on the interaction structure, switching costs and features difference between the two: one network would not agree to interoperate with a significantly better, similar-sized competitor if this makes switching easier—the most classic exemple of that are webmail operators not offering free forwarding all-e-mails.

http://twitter.com/aarlo Aarlo Stone Fish

Thanks Chris. I agree on the business case.I’d give some more credit to the emotional things -* Traditional Google culture – openness, “don’t be evil”, make good products people love, improve the world* Irrational resentment of Facebook. Ex: the talk of employees leaving.Especially given the snarky message.Enjoy reading as alwaysAarlo
http://how2startup.com/ Roy Rodenstein

Google Friend Connect vs. FB Connect is an interesting case study. Arguably Google did a lot of things right, such as providing greater security for users as apps were not allowed to directly access users’ data, as well as making implementation easier than FB through drop-in widgets. As I suspected early on, those benefits didn’t pan out because while protecting users’ data was a good idea, it severely restricted what the client sites could do with the rich social data. It seems Google Friend Connect also miscalculated aiming at the long tail of publishers that presumably were not sophisticated enough to do the heavier-weight development needed to implement FB Connect- what actually happened is that the ‘head’ aka premiere sites did do the work to implement FB and that became accepted as the cooler, trendier identity API to use. Of course you can’t discount that “sign in with the ID your real friends know” was also sexier than “sign in with the ID your business contacts and other random people know.”

In my view that was the real battleground, and this latest skirmish, unfortunately, is Google noticing that the horse left the barn. Unfortunate because, as you point out, they lost the leverage needed to push FB into also being more open.
Pingback: SortiPreneur
http://www.facebook.com/people/Luke-Shepard/2901279 Luke Shepard

I just want to correct this statement: “Facebook allows users to download their personal information (photos, profile info, etc) but has been fiercely protective of the social graph (you can’t download friends, etc). The downloaded data arrives in a .zip file – hardly a serious attempt to interoperate using modern APIs.”Actually, you can download friends, along with photos, profile info, and a bunch of other stuff, using the modern REST-ful Graph API. The “download your data” feature is designed for non-developers and it just uses the Graph API to package that info into a ZIP file for easy download.The social graph is open for any developer – you can download name, picture, unique ID, and lots of other information about a user’s friends. Google’s making this big hullaballo specifically just about your friends’ email addresses.

http://www.cdixon.org chris dixon

Ok didn’t know that (I went to read docs but couldn’t find that). I’ll link to your comment in the post. Thanks.

http://www.facebook.com/people/Luke-Shepard/2901279 Luke Shepard

Thanks Chris – and by the way I think the rest of your analysis is pretty insightful. You’re right – the whole point of the download feature is focused on letting users extract their own info, like making a backup. But I consider that to be a really minor feature when compared with the Graph API

@Gib – it’s just a question of what you mean by “download your friends”. Many services have figured out how to use the Graph API to let users find their friends on their service and do all sorts of useful things. This whole debate is really just over the specific field of contact email.

Here’s a list of all the things a developer can ask for from both the user and their friends: http://developers.facebook.com/docs/authentication/permissions

Goodmachin3

I had started writing something in response to this seeing that all you FB developers are towing the company line but Arrington covers most of what I wanted to say the other day

“Facebook is lying. Facebook’s statement today boiled down to this: “The most important principle for Facebook is that every person owns and controls her information. Each person owns her friends list, but not her friends’ information. A person has no more right to mass export all of her friends’ private email addresses than she does to mass export all of her friends’ private photo albums.”

That’s the same argument that they used two years ago with Scoble. But since then Facebook has been quite willing to allow “mass exports” of “friends’ private email addresses” if the terms are right. They did it with Microsoft, they’re doing it with Yahoo, and possibly other partners. Facebook violated their own privacy policy with the Microsoft relationship. The policy has since been updated.

The truth is Facebook doesn’t see this data as your friends’ private email addresses. They consider it their data. They own it. Literally. So when they say “A person has no more right to mass export all of her friends’ private email addresses than she does to mass export all of her friends’ private photo albums,” what they really need to do is add “, unless it’s with a partner that’s making it worth our while.”

The data export tool red herring. Last month Facebook started letting users download most of their Facebook data, and many people point to that as a sign they are opening up. Nonsense. All that data is nice, and it’s helpful to be able to download it. But it is not usable by third parties in any automated way, and it doesn’t include friend email addresses. You couldn’t use it to export all your photos to Flickr, for example. That means it’s a nice PR product, but it doesn’t address the core problem at all.”

http://techcrunch.com/2010/11/09/give-us-our-data-facebook/

http://twitter.com/Folletto Davide ‘Fol’ Casali

Still, open to use our own social graph is very different from interoperation between two social graphs.
http://www.BriefEpisode.com/ Gib Wallis

Actually, you can’t download friends. You can download a list of friends, but none of their contact information. If you’re on Yahoo or MSN, you can export your contacts to those services, but you’re not allowed to download them directly to your hard drive.

With so many names being common, downloading a list of names isn’t really downloading friends at all.
Anonymous

Are you saying that (rather then writing passive-agressive modals) Google should develop an application leveraging the Graph API to help me retrieve the Gmail adresses of my Facebook friends if, say, I want to share Google Calendars or large documents with them? Or implement Instant personnalisation, maybe.
That’s not a theoretical question: I would love to do precisely that.

http://www.accountinghelp.me accounting help

Google is annoyed at FB’s position. Last week they made if difficult for facebook as mentioned in the WSJ.com article titled
“Google and Facebook Escalate War of Words on ‘Openness’” (here is the full article http://blogs.wsj.com/digits/2010/11/09/google-and-facebook-escalate-war-of-words-on-openness/?KEYWORDS=google+facebook)

Facebook even reacted by creating a around the new Gmail blockage.
http://twitter.com/jfalbano John Albano

Interesting post. Something to keep in mind as we build twitter interoperability into http://www.plateside.com.
Bennet Yee

Size of A+B is 30 so value of inter-network is 900, resulting in gain of 800 by A and 500 by B.

http://www.cdixon.org chris dixon

yes, but i was assuming they are direct competitors so to some extent
it is zero sum game.

James

The whole point of interoperability is that it *isn’t* a zero sum game.

http://www.cdixon.org chris dixon

so in your view, google and fb aren’t competing directly for users,
page views, engagement etc. huh?

Anonymous

Metcalf’s law is inappropriate in that context. Most users care about their close relative more, so to make sense of this, you need to apply a similar law on a reasonable social network, ie. a weighted complex graph. This ames the absurdity of massive gains from network fusion—although not totally: some frontier users benefit significantly.

Anonymous

We should use units when quantifying value (you wouldn’t say the price of a candy bar is “800″)

According to Metcalfe, the units should be users^2. That’s kind of silly, so I propose the “FB” unit:

1 FB = (500M users)^2 = 2.5*10^16 users^2

So the value of twitter is about 40 milliFBs. The value of foursquare is 15 microFBs.

Anonymous

Anyone, starting by Metcalf himself, denounced the absurdity of using squares beyond a handful of users: the initial pitch was intended to sell printer-friendly, EtherNet standard networking cards in an office setting.
You should focus on connexion that make sense: you are interested in exchanging with all your desk-mates, some of your colleagues and the occasional random person. More recent studies of how interactions are distributed on social network have formalized this in a shape called a complex graph. It would seem to point at a total utility from more users that is close to square for a handful, but rapidly curbs, ie O( N*log(log(N)) ).

Anonymous

AIM was only dominant (and present) in the US. MSN had a much larger user base internationally, and teamed partially with Yahoo! based on internal studies on how to avoid Gmail Chat to over-take them.
Pingback: This Week In Brand Strategy & Content - PSFK
Pingback: iTunes Ping: An Opportunity Missed | Jeffrey Talajic
Anonymous

“That said, I think FB Connect and the Graph API is very good for developers and generally a great step toward openness & creating an ecosystem. ”

Chris, Any plans to make the Hunch API open to the public and independent developers?It seems a system like this could only be made stronger with more data, and would allow you to retain your competitive advantage with the right restrictions in place.The thing that scares me with API’s as a developer, is that it seems that any time one’s api key could be revoked at will. Although I would hope by playing fair and following the terms this wouldn’t happen, it’s always a bit of a latent danger.
http://technbiz.blogspot.com paramendra

But I hear Google Me is a hoax.
Malcolm Baker

Hi Chris. I came across your blog by chance, and thought I’d say hello, even though I don’t have anything to add to Metcalf’s law here. Hope all is well. I’d love to catch up. – Malcolm (Baker)
http://www.improffice.com EduardoF

This has been lots of fun watching from the outside. My startup is all about migrating Google Apps data (email, contacts…), so I thought it shouldn’t be too difficult coming up with a Facebook app to import Gmail contacts. It took me 48 hours:

http://apps.facebook.com/gmailcontacts

Wired wrote about it on Friday. With the release of FB Mail tomorrow, my app might turn into an easy way to import your Gmail contacts. I use Gmail’s and Facebook APIs to check whether your Gmail contacts are already Facebook friends. I’m working on making the tool password-less and add a feature to keep your contacts in sync.

I hope I don’t get burned now that I’m jumping in …
Pingback: Quora
Pingback: Quora
http://finance-tutor.com/finance-news/busines-valuation/ business valuation

On a different note, can you share your thoughts on Google’s rumored talks with Groupon with a $6 billion valuation. I’s be curious to see what you and this community think.

My back of the envelope calculation shows that Groupon is undervalued! Ive posted my valuation here http://finance-tutor.com/finance-news/busines-valuation/
http://twitter.com/bjornhendricks Bjorn Hendricks

Great post Chris! All of the fighting about interoperability and data is really based on the notion of WHAT is valuable. Facebook believes that the data it collected about its users is most valuable and important. So it wants to protect that at all costs from rival companies. I believe that the USERS themselves are more important. The data changes and even if it doesn’t, the “relevance of the data”, for Facebook to provide services and advertisements to the users, is only good if the user STAYS with Facebook. Of course there has been no viable alternative yet so Facebook hasn’t had much to worry about. What we are trying to build with “SpotCheq” may sound crazy at first but in the end, the USERS are more important to us so if THEY want to share their data, we will leave that decision to them!
http://positivetranceformations.com.au/hypnosis-gold-coast/ gold coast hypnosis

Collaboration can be fostered by using class wikis, blogs, learning management systems, and custom social networks. What would this kind of collaboration
http://www.escortsmeet.com/ adult services

Designed to contribute to improving the reliability of the Smart Grid, the interoperability standard helps utilities efficiently integrate and analyze event …
Pingback: Google and Metcalf’s Law

Chris Dixon

The interoperability of social networks