Previous  |  Next


Nov 4

Tim O'Reilly

Tim O'Reilly

Mark Cuban, Facebook, and OpenSocial

Mark Cuban is, as usual, insightful when he describes the potential of Facebook profiles to be used as the basis for smarter social applications outside of Facebook. His call for a truly open Facebook API is a must-read for anyone thinking about open social networks.

When you go to my Facebook profile, you get the real me. Thats not to say I answer every profile question. I don't. I'm not going to disclose everything about myself. However, the data that is available about me is the most comprehensive, self maintained database record about me on the internet or probably anywhere....

I happen to think that far better search and ad serving solutions can be developed around a combination of user published information and user activity It just seems to me that if Facebook were to give me an option of publishing a laundry list of relevant information about myself to external Facebook API applications, such as search and ad serving networks that those applications would serve me better results.... The options that would enable smarter use of the web are endless. Not everyone would avail themselves of what I call Personal Database Publishing to enhance the internet experience, but I believe enough would.

Of course any application can currently ask for this information and many do. But I dont want to have to publish and maintain a database for every application I want to use or happen to use. Nor do I want to have to maintain multiple social network accounts to make this information available. I recognize that this is the exact problem that Google wants to solve with their OpenSocial. But they are too late ... If Facebook opens their API up further and allows for its use outside the domain.

While I like the direction of Google OpenSocial, not only may Google be too late, as Mark argues, I don't think they go far enough. A framework and a set of Google Gadgets for building "social applications" misses the point. We don't want to build more applications that look like Facebook applications. It isn't about a social UI. It's about deeper re-use of social data to enliven any application. Some of those applications may have a minimal UI, like Google's breakthrough search app. OpenSocial doesn't give us any of that. Ajax widgets are a halfway house, an attempt to sandbox the kinds of applications that can be created. And that will be the downfall of OpenSocial. If all you can build are Facebook-like applications, Facebook wins.

We all want what Mark describes: a definitive place under our own control where we can describe who we are and what we care about so that applications can use that data to provide us with smarter services. We don't really care whether that repository is at Facebook or Google or any other site, or perhaps even if it's an aggregation of data from many places, but we do want it to become more useful to us. Not just more useful to the holder of our profile, but to every site we touch on the internet. Whichever company gets there first, to a truly open, user-empowering, internet-turbocharging social network platform, is going to be the net's next big winner.

One place where I disagree with Mark is in his assessment of Facebook as the only social network with data about real people. He opens his argument with this statement:

The beauty of Facebook, as opposed to Myspace and other social networks is that the people on there are for the most part who they say they are, and Facebook does their best to dismiss those who aren't. This simple differentiation makes the membership base of Facebook far more valuable than any other social network.

That comment is clearly wrong. My profile on LinkedIn, on dopplr, on jaiku, on twitter, are all really me. So is this blog, and my "museum" at As is my Flickr photo stream, and even Flickr's set of photos tagged timoreilly. For that matter, so is a substantial subset of everything that gets returned about me by a Google search. So is even more sensitive information like all the financial data I've got at Wesabe, or the one-click and purchase history information I have stored at Amazon. Not to mention my favorite dark horses in the social networking race: my email, IM, and phone accounts have even more social network data about me than my overtly "social" apps, as soon as the latent social graph in these applications is made available.

So, no, it isn't too late for OpenSocial. But it is too late if Google frames the problem too narrowly. Imagine a desktop "operating system" where the only APIs were those that allowed you to build desktop UI components, and gave you no access to deeper levels of the system. No one would take such a platform seriously. You couldn't develop real applications. Yet Google (and others) have been pushing the idea that APIs to Javascript widgets are sufficient. Google made a major wrong turn when they withdrew their SOAP APIs in favor of the gadget approach.

Don't get me wrong: I love the lightweight accessibility of gadgets. But it's clear that Google has fallen into the Microsoft trap once referred to as "the strategy tax." Google wants to keep too much control over what their developers can do. And that's the beginning of the end for them.

They need to reframe the problem. What would it take for me, as a user, to "authenticate" information about me that appears elsewhere on the web (Brad Fitzpatrick's original "" idea, now embodied as the SixApart Relationship Update Stream), and for applications to be able to follow that authentication stream? What would it take for me, as a user, to have fine grained control over that authentication, so that some applications could see all of it, and some could see only a little? What kind of system would make it easy for me to manage the data that appears about me, to reduce duplication of effort, yet to give me a single credential that I could proffer as a proxy for "the real me"?

I think Amazon is the only company that really understands that we're in the process of building an internet operating system, and is building services with the depth and power that will be required to create the next generation of computer applications. So far, they've focused on lower level services, but in one sense, isn't what we're looking for a kind of "one-click" that would use our stored social data to inform and activate new applications? (I take that back. I think SixApart gets it. Hmmm...Amazon ought to buy SixApart, purely for the social networking API play, and do it right....)

tags: facebook, mark_cuban, OpenSocial, web_2.0  | comments: 28   | Sphere It


0 TrackBacks

TrackBack URL for this entry:

Bob   [11.05.07 01:42 AM]

Actually, Apple is the king of the strategy tax. It put them 25 years behind MS.

Pascal Van Hecke   [11.05.07 02:06 AM]

Money quote:

"What would it take for me, as a user, to have fine grained control over that authentication, so that some applications could see all of it, and some could see only a little? What kind of system would make it easy for me to manage the data that appears about me, to reduce duplication of effort, yet to give me a single credential that I could proffer as a proxy for "the real me"?"

So far, the combination of OpenID and OAuth is the best bet...

Don Dodge   [11.05.07 03:31 AM]

Tim, Brilliant! You have avoided the hype and laser focused on the real issues.

Mark Cuban does have a good idea there...he has lots of them. Although in this case his angle is combining the Facebook API and Yahoo community traffic to better target advertising. I think it would work, but the benefits to users is marginal. Mark's idea is really about better advertising.

Google and OpenSocial think it is all about making it easier for developers to build more widgets. There are over 7,000 on Facebook now. Do we really need more?

You are absolutely right. The problem that needs to be solved is an easy way to re-use and control OUR social data across social applications. OpenSocial doesn't do that...yet.

Perhaps someone like Marc Andreessen (Ning) is in the best position to make this happen.

Tim O'Reilly   [11.05.07 03:47 AM]

Actually, Don, it strikes me that Microsoft is probably in a great position to make this happen. Think about the data locked up in outlook and exchange, for example? that's a great social network. What if Microsoft built me a kick-ass "software plus service" application, an iTunes-like application to help me manage all my contacts? The address book hasn't really changed in a decade.

Look at They are starting to do what Microsoft ought to do.

Now imagine that user console for managing my relationships, with a master profile for my own identity and the data about me that I might want to track or share, and the ability to publish from there data to other sites (so I don't need to re-enter it anywhere else.)

Address book 2.0 and the APIs to access it are the real killer app.

Oh, and of course, address book 2.0 could easily morph into wallet 2.0.

Chris   [11.05.07 04:05 AM]

Mark has a platform and he will get attention. Can anyone remember how he painted litigation and otherwise inevitable doomsday for anyone acquiring Youtube????

Don Dodge   [11.05.07 04:48 AM]

Tim, I agree on the Outlook and Exchange possibilities. I have already advocated those ideas to the Outlook team. Time will tell what they do with them.

Outlook Contacts already has all your "real friends" and they can be sorted by company colleagues, business friends, personal friends, or whatever. how hard would it be to add a picture to everyone's Outlook contact? How hard would it be to have LinkedIn like functionality within Outlook? How hard would it be to implement the ideas you suggest? Looks pretty simple and obvious to me. In a big company with millions of users it takes a lot longer to do things than it does in a startup. I have been in both situations and understand the strengths and weaknesses of both. It is what it is.

I just wrote another blog responding to your post and Mark Cuban's ideas. I like both but have two concerns; security and privacy. I know you understand those things, but to most of the consumer oriented web 2.0 developers it is an afterthought. More on that here

Man, you get up early! I am in Denver this week and it is now about 5:00AM. If you are in California it is around 4:00AM. Cool!

Tim O'Reilly   [11.05.07 05:07 AM]

Actually, Don, I'm in Berlin for Web 2.0 Expo... You'd know that if you followed my twitter feed :-)

David Thomson   [11.05.07 05:59 AM]

We are addressing this problem, but from a very different angle, and one that will take a bit longer to execute on if it works at all. However, it seems inevitable that someone will advance a strategy around what we call a "personal cloud". This is not a "personalized cloud", however. It's the idea to create a set of personal servers, most likely hosted virtual private/dedicated servers that are peered (for redundancy + CDN), with an interface that makes the location of the data seamless even though it's private to the user. We have an early version of what we call the "Explorer" interface, essentially a file/asset view of the information stored in the personal cloud.

We just released our entire platform as open source, and plan down the line to create an API where web applications can utilize the data in the personal cloud. Social networking applications abound, but we strongly feel the first step is to create a private space that will hopefully trigger more long-term innovation. Here is our vision document:

We have been in "incubation" for a while extremely busy delivering custom versions of our software to hedge funds and financial services companies, but hope that some of our ideas will take root out in the open source sphere.


manish   [11.05.07 06:28 AM]

Can you share your views on the impact of open social on orkut for which the api has already been relased.. and also the impact on other smaller competing social network which are not a part of open social.

Fabian Schonholz   [11.05.07 06:42 AM]

I think it is all good and dandy, but what is missing from the current social networks is REAL long term stickiness. So far social networks are nothing more than toys. They lack productivity suits that will provide real value to users and/or group of users.

Think about it, social networks are perfect platforms for cloud computing. And I agree, Amazon is doing it right.

manish   [11.05.07 07:31 AM]

Can you share your views on the impact of open social on orkut for which the api has already been relased.. and also the impact on other smaller competing social network which are not a part of open social.

Martin Lawrence   [11.05.07 07:42 AM]

Tim, thanks for bringing present Addressbook 2.0. Few people seem to see the treasuretrove of information digitally stored and tracked in our email and cellphones.

While a smart application could construe my "real" professional network from analyzing my Outlook conversations, I believe the greatest untapped potential lies within the cellphone addressbook; after all, this is where my private network is most accurately depicted.

Nokia should be in a strong position to exploit this (as of course is Google, once they get their foot into the mobile OS market).

Matt Peskett   [11.05.07 08:54 AM]

I keep wondering why everybody is overlooking Amazon in all this too - they've been collecting our buying behaviour data for years. Their ability to hit me with relevant marketing promotions that I actually convert on by far exceeds any other vendor.

I've always thought a quick win for them would be to serve relevant product listings to me when I log into Facebook? The value of my favourite books, music and films, plus an application for friends to see what would make a good gift for me (my wish list?) would seem priceless. Perhaps they already think they have that?

Maybe I'm too obsessed with online advertising to the extent that every ad I see in Facebook I question the intelligence behind serving it to me... and get frustrated at the lack of ingenuity to date.

Reedo   [11.05.07 10:43 AM]

Er...all this talk about Web-connected applications sharing information to enrich 1) each other and 2) the user experience reminds me of something...what was it...oh, right, the Semantic Web.

devin holloway   [11.05.07 11:19 AM]

3 words: Meta Social Network

When users get to "manage" their SoNets from one "dashboard" things are going to dramatically change. I think Google will announce their Socialstream soon, fresh on the heels of OpenSocial, and I *hope* they're able to use the OpenSocial API to tap into the power of the Meta SoNet concept.

Basically this means:

  • the Meta SoNet allows users to check the status of all of their SoNets from one place.
  • the Meta SoNet doesn't replace the other networks, just like RSS feeds don't replace blogs and websites.
  • the Meta SoNet and OpenSocial API help benefits consumers, advertisers, and network operators - targeted, opt-in ads made possible by linking semantic data across multiple networks. Consumers get the goods, advertisers of course find their consumers, and networks benefit from facilitating this process.

Eventually the Meta SoNet will be the actual Internet, but until that happens we'll just have to make do with the one-stop-shop that will evolve into the "dashboard."

Cheers - Devin

ahoving   [11.05.07 11:53 AM]

care for a, into which we could drag and drop all our networks?

Josh Patterson   [11.05.07 12:10 PM]

Our data on the web is doing a number of things, 2 of which are

getting stored in more places
getting more complex

- think about this for a minute: What would you do back in 1995 if a desktop computer application would not let you take your data off of that computer via a file?

One angle our team is taking is looking at how operating systems have grown over the years, and one thing strikes me time and time again: layers of abstraction. Do we program directly to physical memory anymore? no, we have vm addresses. Do we read (directly) individual disk blocks? no, we have inodes, and bmap, a lot of other layers of processes --- and so on.

I think we will end up with a "web data stack", ala/similar to the OSI Networking model, and javascript (ironically enough) will become the "system api" of the internet. You will simply say "give me all pictures of animals that are owned by openID X" and you will get possibly a SIOC-RDF data cluster of xml from inside your widget *on any site* that contains the aggregate data of multiple image sites, just like how querying a filesystem gets you an inode that points to many disk blocks. The internet will simply become a single logical disk/database.

Via a single "internet login", all applications are "inherently installed", and you jump right into a tutorial with an application that already has discovered all of your data.

And yes, I think Tim Oreilly is right --- OpenSocial is not really that "open", and Amazon probably gets this more than anyone right now, they just dont get the hype for it.

Dominik   [11.05.07 12:44 PM]

As fascinating as the idea of Amazon, ebay, facebook etc uniting their knowledge is, I think it is not going to happen. Companies like these put so much effort in following TimĀ“s advise of creating unique data that they will not share it.

rektide   [11.05.07 03:49 PM]

Brad's Thoughts on the Social Web is what you are looking for. "Goals:
Ultimately make the social graph a community asset," his bold & mine. But of course, you already know that.

I too thought OpenSocial was going to provide forms of authentication and information control & disemination. After snapping up Brad Fitzpatrick I had hope Google'd get right to it and build the distributed systems networking API's needed to do the job right. A closed source sandbox and three more Atom+APP API's was a disappointment I could never have forseen.

One of the last things Brad mentions in the linked post is Social Network Portability Google Group. The OpenSocial thread is one of the few places I've seen attitudes mirroring your own and mine. Baudrillard would've crapped himself to see hyper reality like this, watching the hyperlinks fly is amazing, but I cannot figure out over what.


Amazon Fan   [11.05.07 09:17 PM]

Tim says: "I think Amazon is the only company that really understands that we're in the process of building an internet operating system, and is building services with the depth and power that will be required to create the next generation of computer applications."

I agree. Furthermore, Amazon know about the $5,000 worth of stuff I bought there, and is targeting me. And I don't care. Because they are not trying to sell my profile. They are just trying to make my experience better, privately. And that I trust and don't mind. Amazon is already the Facebook everyone is looking for. So while Yahoo and Google are trying to intermediate with advertising, Amazon is actually delivering the goods, literally. Amazon is going to be the ultimate winner.

David Geller   [11.05.07 10:47 PM]

It strikes me as odd that anyone would be thinking any of these majors players is too late to the game. How can that possibly be true? There's so much yet to be done and, still, so many opportunities - not only for MySpace and other Open Social players, but Microsoft, Amazon and Facebook. While Facebook may have, in the eyes of many, a commanding lead in exposing and hosting a robust API-driven social network, it's still only a few months old!

Lauren Cooney   [11.05.07 10:54 PM]


Sorry I couldn't join you in Berlin this time around... definitely missing it!

That said, what do you think about the access Google has to the data of those folks that are building with certain APIs? Specifically, how the "preferred approach" to be authenticated to the API is to use individuals' google accts to log in?

I posted about it on my blog. Mostly, I focus on the community part and being a "good community citizen" but would love to hear your take.


Claudius   [11.06.07 01:12 AM]

I think it's going to be very interesting what's going to happen with the social networks now.

But I do have to agree with Mark Cuban, that on Facebook I do use my real details, on MySpace it was more 'just having fun' and most people just used nick names.

Dzejms   [11.06.07 03:29 AM]

Don't forget about setting your personal EULA preferences so you never have to click "I agree" again!

Jegan   [11.07.07 08:03 AM]

I can only partially agree with your point regarding MS Outlook's potential to become 'The Social App'. Outlook may not have most of my friends' details. What outlook has is all of my collegues details.

Another missing part - my personal email and in turn my friends who are not my collegues - is with Gmail. If Google starts to use that more effectively, it will be a significant impact.

Jon Grande   [11.09.07 05:16 PM]

Tim - I actually think the answer lies in two things ...

1) Based on the communal power of the internet as an aggregator - I would guess that there won't be a single company/player who provides the answer. If it were just about $ and vision, Microsoft would have found a way to make it's failed Passport initiative work (in combination with Hotmail, Outlook, IE, etc) >.

IMHO it'll need to be some form of standards driven consortium.

2) With the pace of evolution in the online space increasing geometrically (a $15B valuation for Facebook in 3 years - OMFG!) I would theorize that the most likely path to success will be for companies who engage broad sets of users in interesting ways (Amazon, Google, etc) to find ways to syndicate their data out in trusted ways.

I don't want to give all my info to one uber-service, who'll likley turn around and sell themselves (witness "" in the MMO space) and I don't want to have to tell each service/site that I encounter than they can/can't have baseline info about me. I'd rather see the major sites I use find ways to syndicate info about me so that it's just "there" when I encounter other services - with varying levels of approval or credentials based based on how "trusted" that new site or service is by the sites I've already agreed to trust.

Rytis Sileika   [11.24.07 11:17 AM]

With regards to "...It's about deeper re-use of social data to enliven any application..." Isn't that a security concern? Facebook API allows accessing only peer data, similarly OpenSocial, but to enrich an application (even to implement Mark's suggested Facebook+Yahoo pair) whole user data should become acessible, shouldn't it?

Tim O'Reilly   [11.24.07 11:24 AM]

Rytis, obviously, you need to put the user into the driver's seat regarding what data gets re-used, and by whom.

See my post and the comment I made about halfway down the comment stream, about how putting the user in charge of the data cuts the gordian knot of security.

Post A Comment:

 (please be patient, comments may take awhile to post)

Remember Me?

Subscribe to this Site

Radar RSS feed