More Questions Than Answers About OpenSocial
The launch of the OpenSocial Foundation yesterday was probably more about preempting lawsuits than it is about anything else, but getting Yahoo!, Google and Newscorp all in the same room raised some eyebrows. It also raises some big questions. In fact, there may be more questions on the table about OpenSocial than answers.
Is OpenSocial for real? It it going to make a big impact on the web? Before asking all of that, here's some thoughts on the nitty gritty as the platform moves forward.
What will Yahoo OpenSocial?
The definition of a social network gets looser every day, but what exactly is Yahoo! going to open up to OpenSocial? Maybe we'll see OpenSocial apps jumping from MySpace to Yahoo! Answers and Yahoo! Finance, but more useful would be pulling data from Yahoo! properties for use on other networks. This would have been a great question to ask in yesterday's conference call about the announcement.
Yahoo's participation may mean a lot more smoke than fire. Hopefully they won't pull a LinkedIn, launching a platform, promising openness but months later nothing but a closed development environment and... What is LinkedIn doing these days other than rolling out some features in-house?
Interoperability or Another Splintered Landscape?
Only a handful of OpenSocial "platforms," or supporting social networks, have gone live so far and already each is "customizing" in ways that aren't interoperable. Some of these networks have user data fields that just aren't available on other networks, other are extending OpenSocial in ways that could be widely supported but aren't part of the platform right now.
Company representatives said yesterday that porting an app from one network to another is today "a matter of hours instead of weeks or months" like it would be if it weren't for OpenSocial. As more and more containers go live, though, it could end up taking weeks to make something truly interoperable. OpenSocial could very likely just end up being a network of sites that happen to give people permission to build apps instead of offering meaningful advantages for developers compared to stand-alone sites. Cross site-data portability would be the one remaining advantage, but we're not seeing a lot of that implemented yet.
Who Will Use Data Portability?
If there are apps that allow users to draw their friend and activity data from one OpenSocial container to another, we haven't seen them yet. That's a big part of the value proposition for users, but despite the rhetoric it's still a struggle for vendors. Likewise, the most successful apps so far have been little more than lightweight games - nothing that developers seem incentivized to make very sophisticated.
When do the Good Apps Show Up?
The MySpace rep yesterday said that good apps are on their way, it's not just going to be a bunch of toys. That's right, it was the MySpace rep that said that. We'll see. So far most of the best apps are just games that aren't annoying. Data portability has world changing potential, but it's unclear that OpenSocial is going to be where we see it.
How does Microsoft feel about this?
Microsoft made it's own interoperability announcement yesterday. Some people think that Microsoft probably isn't very happy about Yahoo! teaming with Google and promising further openness. Microsoft could shut down or pull out from OpenSocial if it buys Yahoo!
Who's More Open, Microsoft's Deal or OpenSocial?
According to Josh Catone here at RWW, this is what Microsoft's announcement yesterday means: "users on Facebook and Bebo will be able to add friends via their Windows Live address book. The functionality will be coming to Hi5, Tagged and LinkedIn in the next few months. Microsoft is calling this a "two-way street" and has launched a new site, invite2messenger.net, where users can invite friends from those networks to chat on Live Messenger."
Google has been downright creepy in some of its openness strategies lately, but some people think Microsoft is emerging as more genuinely open. Microsoft isn't participating in OpenSocial, but its defacto network of networks could end up providing more meaningful openness for users.
At What Point is Facebook Obliged to Join?
OpenSocial was ostensibly created to fight back against Facebook's momentum. Facebook doesn't have much incentive to join a fledgeling app network today, but at what point might they?
OpenSocial should be able to trounce the closed Facebook platform, but that's only if it becomes a meaningful network of networks. Facebook is also focused on privacy controls, something that would take on a new level of complexity if it jumped into OpenSocial. Surely though there's some breaking point whereupon Facebook would have to join everyone else. Perhaps they'll go with ad-partner Microsoft instead.
What Happened to oAuth?
When OpenSocial was announced, we asked why the companies weren't just using existing and emerging data portability standards and protocols. We were encouraged to learn that OpenSocial does offer support for the user authentication protocol oAuth. OAuth lets an app access your data on another network, with your permission but without itself asking you for your password. It's really exciting - but we're not seeing any OpenSocial app developers use it.
Will data portability standards get used in OpenSocial or is this network of networks just going to do its own thing? The vision of free flowing user information, with sophisticated and powerful granular controls, data-centric features and free love for all seems less central to OpenSocial every day.
It's easy to feel cynical about OpenSocial, with so many giant, proprietary cooks in the kitchen. Do the big networks finally "get it" or are they just faking it and actually dragging their feet? The potential and promise are huge. So are the obstacles and remaining questions, though.
Marshall Kirkpatrick will be a guest on G4's Attack of the Show tonight, discussing OpenSocial. If you're a TV watcher, check it out and let us know what you think.
Share
Comments
Subscribe to comments for this post OR Subscribe to comments for all ReadWriteWeb posts
IMHO the only people who feel cynical about OpenSocial are those who don't understand what it is in the first place. Unfortunately, Google did such a poor job of defining OpenSocial that the blogosphere somehow assumed OpenSocial had something to do with social graph portability (hint: it doesn’t).
OpenSocial is simply a broadly defined API for connecting social “containers” (e.g. platforms that manage a user’s activity stream, friends and profile information) with social “applications” (e.g. web apps that manage this data and render inside containers).
In that sense, OpenSocial is exactly what it is designed to be: an “open” standard for social applications (vs. the proprietary standard defined by Facebook).
In short, SN inteoperability and SN social graph portability was never promised in OpenSocial. Of course, it *can* be done through OpenSocial if someone develops an abstraction that thoroughly maps the proprietary data from one container to the next (OpenSocial identifies the container domain so that can be used to drive a mapping system).
Is OpenSocial for real? Yeah. Is it what people want? Probably not. Is it was developers want? Almost and, well, it's good enough.
1. I could see Yahoo using OpenSocial for My Yahoo... that's the only current property where it makes much sense to me. OpenSocial gadgets are not too far removed from iGoogle gadgets, best I can tell. The question would then be what social graph to use... perhaps by Yahoo Mail address book?
2. To re-emphasize: OpenSocial has nothing to do with data portability. It never has. Anyone feel free to correct me if I'm wrong, but my understanding all along has been that OpenSocial adds a layer on top of a social networking site (host) that applications can then plug into... which is where interoperability comes in - ideally the layer is the same on all hosts. I'm not surprised that there's some splintering (once again, a generic platform can't offer tight integration), but I don't think it will be a huge issue. If the basics are the same, Google is right that modifications will take hours instead of weeks. Slight variations on the spec are far better for developers than completely different specs. The problem will be how hosts distinguish themselves to users, which may be behind some of the platform extensions. This is almost starting to sound like browser wars...
3. I highly doubt we'll see OpenSocial apps offering complete portability. The social graph remains at the host layer, so even if an Orkut app exported friend data, how would a MySpace app import it into a user's MySpace friends? Besides, this kind of data transfer would most likely violate the TOS of any social networking site. Data portability raises so many issues of privacy, security, content ownership, etc., that I would say it has to happen at the host layer. Once again, OpenSocial has nothing to do with data portability.
4. I continue to say that the best Facebook apps have yet to be written... same for OpenSocial. These kind of applications have pretty much existed for less than a year, and it will probably take more time before developers find real killer apps for social networking.
5. Possibly, though I don't think Microsoft would care too much. With the new OpenSocial setup, Yahoo isn't really teaming with Google any more. Microsoft has already joined DataPortability.org, which Google is also a member of... OpenSocial is now another consortium of companies, and Microsoft has no major competing product.
6. I viewed Microsoft's announcement as similar to the Google Contacts API - a very important step in that it stops users from having to enter their e-mail password on a third-party site, but not exactly a huge announcement otherwise. And comparing OpenSocial is a bit like apples and oranges - they address two different things (app development vs. contact import/export). But "open" is simply a buzzword with positive vibes at the moment. What do you mean by "open"? For instance, no social networking site has yet implemented very much of DP.org's vision for "open."
7. How exactly is Facebook's platform closed? If you mean that it's controlled by one for-profit company instead of a non-profit consortium, fine - but how does that make one better or more certain of success? I don't see how OpenSocial will "trounce" Facebook. If a developer only writes for OpenSocial, they're ignoring a huge set of potential users (Facebook's). Facebook has begun licensing their platform - it's already available on Bebo. Facebook's Platform is a very proven setup that offers advantages OpenSocial can't match (tight integration, iframes and script not necessary, mobile version, etc.). I maintain that Facebook has no reason to join OpenSocial and that no reason will appear in the near future. It will take a very different landscape for Facebook to start supporting OpenSocial.
8. Marshall, your statements on this last question seem to confirm what I've been thinking as I've read - that you're misunderstanding OpenSocial. I can't blame you because the initial hype could lead one to think that OpenSocial would walk your dog, but now much has become clear. I can't recall any indication that OpenSocial has ever been about data portability. As I've argued before, OpenSocial is a "cloud" for developers - it remains a layer on top of disparate social networking sites. OpenSocial is not a "network of networks" - it's the Java of social networking sites. Rather than writing an application in OrkutScript, MySpaceScript, and BeboScript, you can write an application that uses OpenSocial APIs and deploy on multiple platforms. The data used by the application, though, depends on the host - when running on Orkut, your app gets Orkut data, etc. While a developer could theoretically build an application that lets you "friend" people within the application, thus building a network on top of the other networks, OpenSocial is not made for that and I think it would end up being way to complicated (and not integrated enough) to be useful - if it were even possible given TOS restrictions.
In conclusion, I want to reiterate that OpenSocial is a good thing - it helps solve a problem. But it's only solving one problem, not problems like data portability. You have to keep OpenSocial in perspective.
btw, Josh Catone hit some of these same points in January: http://www.readwriteweb.com/archives/opensocial_or_opengadget.php
"So if OpenSocial is really not about data portability and interoperability between networks (except as far as widget creation is concerned), we'll have to look elsewhere for that. But that's not to say that OpenSocial is a total wash - widget standardization isn't such a bad idea."
you mean I still have to walk my own dog? :) all fair enough theharmonyguy, though I don't think we can so easily say the two are unrelated. Conversation after this post has got me convinced I need to talk to some more OpenSocial folks. I'm far from the only one who is unclear on this.
Marshall, the commenters above seem to have grasped a lot of what OpenSocial is about. The idea of 'Data Potability' is a good one, and one that I support, but it is built on a model of every application having its own website and its own member and friend databases. This means each application has to get you to re-enter your personal information and re-connect to your friends on the site, for 'friend' functionality to be useful. This adds overhead for each developer, and each user, and is clearly a limiting factor on adoption.
One answer to this is 'data portability' - being able to bring information you have already entered in one site to a new one. Where this works well is with profile information, if we can define interoperable profile standards, but for 'friend' information it is trickier, as you then need to bring other people's information from one site to another, which means getting them to sign up too. The Social Graph API is a way to do this with publicly expressed 'me' and 'friend' connections on the web.
However, many applications would take advantage of friend data if it was easier and more natural for the developer and the user. OpenSocial gives a way for web applications to discover the friend and personal information that the user has already shared with he container context the application is running in. This frees the developers to be more creative on the core application code, and the users from having to re-enter the information and reconnect to their friends, and the container sites gain new applications for their users that they don't have to write themselves.
Itsa great to see some of these questions asked.
But all you have ot do is dive into what OpenSocial is today and you'll see that all that Dataportability and Interoperability is not possible.
So the real questions to ask are:
- when will OpenSoical2 appear?
- are these folks who are claiming to support OpenSocial- really supporting it?
- do they offer the same level of interop that MS is doing right now?
- and what about all those intrusions into people's data rights that Google did - connecting GMail contacts to Reader accounts? Why should we trust Google?
- and what about the Soical Graph APIs?
My understanding is that OpenSocial is 3 things.
1) A Gadget API that gadget developers can work to.
2) A Container API that website owners can add so that gadgets can run in their web site as well as Orkut.
3) A set of RESTful Data APIs that can be used to build mashups.
The Gadget API (1) is coming along nicely. There's a solid bunch of people busy writing gadgets. The Container API (2) is following behind with, as you mentioned, early adopters unilaterally changing the spec. But this is at least party because the spec is still in flux. The OpenSocial Foundation should help to bring everyone back in line. The Data APIs are on hold. And there's some serious overlap to be resolved between the Data APIs and Google's GData APIs such as Contacts.
Set this against MS Live Contacts. It's not clear if anyone other than MS will implement as a supplier of data. So for the moment it's a way of getting at Hotmail contact data in a properly architected way. The demo of Facebook -> Hotmail may or may not actually use it. Facebook may or may not let other people use it. And all of this is in competition with Google's Contacts API, OpenSocial People API, Plaxo's Sync API and no doubt others out there as well.
It's all good but at the moment, as ever, it's all deeply flawed as well. And anyway, anything that pumps up the volume for DataPortability is a good thing.
Excellent interview on G4, Marshall!
-Amit
I had some of the same confusion about OpenSocial, especially after that initial video of pairs of containers seeming to access each other's events graphs. OpenSocial is
- an API for exposing social graphs _within_ a container
- this is of interest to app developers
- and to advertisers
OpenSocial does not seem to address use cases such as "give me a list of all my friends on all my networks, and please strip out the duplicates". BUT - the locus for this is not the container out there (FB or whatever) - it's ME. More specifically, this is what my Identity Provider (my OpenID Provider) should be able to offer. This also puts privacy concerns in a different light - they are my issue, not that vendor's issue: I own my identity on all these networks, and I should be able to control it and its rights, and be able to weave my own cross-network views. Killer app here, folks!
- @szpak
I think Marc nails it well.
watch abc tv online