-
Website
http://www.scobleizer.com/ -
Original page
http://scobleizer.com/2008/03/26/the-real-roadblocks-to-data-portability-on-social-networks/ -
Subscribe
All Comments -
Community
-
-
Top Commenters
-
Popular Threads
http://www.unodewaal.com/2008/03/26/microsoft-i...
LOLWUT? Why did Britain's Data Commissioner have to step in to slap down Facebook for it's efforts to prevent users getting their account fully deleted.
Dave Morin sounds like he's spreading FUD. There are existing implementations - like the ability to import your contacts into Dopplr from Twitter and from any page with hCards. That works fine, and nobody is bitching about the privacy concerns of this.
A lot of people are bringing up "policy" reasons why this kind of thing can't happen. The fact is that they are already happening. I'm sceptical that it'll be groups like DataPortability.org that'll push this - it's more likely it'll be hackers working together in an ad-hoc way will solve the problems. Then the vendors will jump in front of the parade and start preaching openness when it becomes inevitable.
In the business world you'd expect it to be a relatively straightforward task to decide on the 'standard schema' for a customer record to enable easy data interchange between systems (a long promised but not yet delivered facility) - but this has yet to happen and that is not on the scale of complexity that you are talking about here ...
(this is acutally why i love it that facebook and ms are getting cuddly. i use live services extensivly, and facebook too. they are together like 80% of my online social thingies)
I thought Plaxo was supposed to lead things off into the right direction but shifted focus to being a pseudo LinkedIn. I think that this will be much easier than the supposed Local Number Portability that was supposed to allow customers keep their phone number as they switched telcos in North America.
For me, too I have dozens of logins, usernames, nicks because of formatting, preference of nicks/handles over emails and such so how are some people to know it is me. Avatars help but have no ID or sniff test for people to say, "That's him!" or say, "This person is an imposter.".
I'd be interested to see if a solution for cross application interaction happens.
Robert, in your post you have defined what you think "Data Portability" should be. However, I think the industry has to define the requirements for what is "Data Portability" first and then deal with the politics of implementation.
To do anything of substance you have to have common requirements so everyone knows what's going on and can march to the same beat.
To answer your question, you could have some type of portability before the end of the year, depends on the requirements.
My take: unless I can control who can see what, I'm not in favor of "social data portability."
But if you look at LinkedIn, they have build their own Twitter for their system, instead of just using Twitter...those are the things my mind just cant seem to get..
I think for it to work it should be opt-in if that is even feasible.
The Google Social Graph API is this.
http://socialgraph-resources.googlecode.com/svn...
As soon as we can define standards for friends then I think the need for a social network has gone. And it has already been defined. The only way that data can become portable between social networks is if it doesn't matter which social network you are on you can still communicate with each other. i.e. Like EMAIL. But obviously with more standards it can be richer.
I wrote a post about this a while ago:
"If you could move your information about friends etc (your social graph) right now would you? Where would you move it to? Moving your social graph is not going to be useful unless all of your friends (or at least some of them) move too. In this way making data portable is not really going to change anything about the lock in with social networks. People are going to stay where people they know are. Similar arguments debunk the significance of Open Social. People don’t switch networks for applications or widgets. They move because of their friends.
What is needed instead is to make the rich experience of a social network standardised. I think Google knows this. The real future is ONE social network for everyone. I see social networking as an extension to email. From Gmail I can email people on Hotmail and they can email me back. There is a communication standard there. So what we need is a standard account type (that is OpenID) and then standards for feeds of activity (that’s RSS), and standards for everything else you do on a social network. Once the social networking “scape” is fully open in this way people can use their own bit of the network (essentially their network provider) like Facebook, or Google, or Bebo, or whatever. And it might have slightly different features just like the way that Gmail has different features to Hotmail but it would have all been the same network. So I can switch from Google to Facebook and I’ll see a different interface but can browse my friends on whichever network provider they’ve chosen.
OpenID and a standard social graph is the key. As far as I can tell there is no benefit in the “half-way” solution where I can move my data from Facebook to Orkut. My friends will still be on Facebook. Fingers crossed this can go all the way."
On the other hand, I don't understand where how you'd come to the conclusion that sharing content or making it portable (i.e. photos, videos, blog posts) is the hard part compared to sharing identities across different services. Don't we already have the former happening today?
Most of the scenarios you mention involve some kind of decision a service provider can't make on his own (mostly because it can't be automated or generalized). So what? Let the user decide which information to share with whom, when and for how long. We're the only ones really responsible for our data. Why don't they let us be?
You have raised a very valid point and I had twittered on that in reply. One thing that also comes into my mind is that in an open dataportable environment how will these applications trust each other. What is the qualifying parameter and checks that an application needs to do before talking to another application. Are systems like OpenID solution to that. In the portal environment 'Trust' will be the most critical factor and there needs to a common an open qualifier for 'Trust' too.
Thanks
Santosh Maharshi
Yes, are lots of big challenges around data portability. However, just because it's tough to solve the *whole* problem, doesn't mean it's not worth solving any part of it.
Lots of software developers fall into the trap of wanting to design the "big solution" for the "big problem". However, that's almost *never* the right answer as far as the end-users of a computer system are concerned.
http://thesocialnetworker.com/tsn/tsn.nsf/dx/wh...
Chris Saad and myself in a podcast on DataPortability.org
http://thesocialnetworker.com/tsn/tsn.nsf/dx/Th...
I don't believe portability can ever be done automatically by the networks. There are just too many use cases and privacy concerns and issues to get this right. IMO, external tools that help export and import data in and out of the networks is the right way. The user controls these tools and decide what and when to do data porting. Automation can be achieved by the user with the right tools. But they dont have to worry about the networks doing anything right or wrong.
Mary
Call it "Facilitating transfer of personal information" and you might get a different take on it. You write like it's all good, but it's going to mean that more people's personal data spreads to more and more companies -- and not just minor personal data, but your complete social network, likes, dislikes. The marketers are salivating.
Not that I don't want the "change your address once" convenience, but when it comes with the spamminess of Plaxo, I always wonder if it's worth it.
I think these issues are going to get solved from both ends. Top down defacto standards and bottom up innovation and evangelism. Let's hope they meet in the middle. The more people like Robert demand these sorts of functions and the more groups like DataPortability spread the word, the more likely it is to get done.
And as someone up there said, don't avoid the easy stuff because the big picture looks impossibly hard or some of the issues look intractable.
All of the use cases Robert pointed out seem to be solved by this sort of solution.
Oh, and for the totally paranoid, I'd even suggest having a browser plug-in that maintained your social networking rules.
I'm a fan of openness, but in a world where data is a commodity and Google is the 800lb gorilla, it's a good value proposition to differentiate yourself by building fences. It's like commodity PCs powered by Microsoft vs. Apple's walled garden.
I take social networks, at their core, to mean three related functions:
1. Some way of maintaining state about yourself: contact info, demographics, etc.
2. Some way to identify relationships - who and what kind of relationship.
3. Some way to control permissions of those related to you and their access to your state.
I see data portability as defining a standard set of policies governing the sharing and unsharing of these three things.
Done correctly, it means the end of social networks. The standard that it defines by default makes one gigantic social network, one that correctly mirrors human relationships off-line!
This is similar to Sam's thoughts above.
Vendor-led efforts are dead in the water. If data portability is accomplished, there are no social networks left. It becomes a public utility that people and businesses use, one that generates vast amounts of economic utility, but one that no private party is able to convert into fiscal value.
I second John's concerns, but they aren't a problem. This seems like an OO-purist's obsession: correctly establishing identity. If I want to lead a double-life, I can lead a double life. And if I bisect the AllThing's social graph with two nodes that represent me, oh well. I have that right.
Indeed, the idea of a universal social graph is flawed. It only benefits companies. If privacy is respected, then we each get a different social graph. Ralph may not want me to know that he knows Wanda, or in what capacity, so on my social graph, that link will be missing.
So my vision is also to have one social network which is the internet but not as one master server for everybody but following the OpenID model (I think this goes into the direction Sam was thinking). So this can be easily used to sync data. If you change your email address 3 times then the last one should be valid unless you state that you want to use another profile (with a different address) on some social application (if we have one social network the existing ones might then be called social applications making use of that network).
So everybody can actually choose their master server of choice (some might trust Facebook, some might trust their bank, some their own hosted thing) and you give permissions around.
Of course this is not fully thought through but it might be one idea to think further.
The thing is though that "Data Portability" can be sort of everything or nothing depending on who you ask. This email problem is more a problem of sync and sync is probably one of the most difficult to solve. I would already be happy if we would solve the export issue, defining technical standards and policy for exporting data (and policies which you are allowed to do with it).
I summarized some ideas of how to partition the problem space here: http://mrtopf.de/blog/web20/what-is-data-portab... and during the last DP meeting a similar idea seems to have been discussed in the form of DP labs. I think the main problem which stops the actual work in DP right now is really not having a defined scope (maybe narrow enough to being able to solve it).
Another thing I would like though would be that Dave Morin joins the conversation on one of the Data Portability Groups or even lists these use cases in the wiki so people can think about it. I mean FB joined this project so it would be good to know what the problems they are facing are. Both on the user and on the business side.
I simple "handshake" with a new website would allow them to see where else I share similar info, who I relate to and even (cringe) serve me relevant ads.
This could be expanded to all levels of customization or preferences for web use.
One config file to rule them all?
Facebook is not unique in having these problems, and the answers *are* being discussed - good answers.
I think Open ID is an important part of this. Open ID does not require every product I use to keep a copy of my ID and password, but whatever system I use as my authentication server must have a high degree of uptime and responsiveness to validate me to all the others.
Why can't my address list be housed on Google if I use that most for e-mail, or Yahoo if I use that, and available in real time to any service I've authorized to use it? You seem to be thinking in terms of every service copying all of my accumulated data when I first sign up. I'm quite sure the Facebook guy was thrilled to tell you how impractical that would be. I can just imagine him saying that RSS was impossible as it would require every player to maintain an entire copy of the Internet! Once people join these walled garden companies they can't think straight any more (if they ever could).
Oh a third thing... for data that DOES need to be copied from one service to another, all you need is a standard for defining a "delete transaction" or some more generic transaction mechanism to keep them all in sync without constant copying. And in the case that someone asked above of you changing your e-mail address to different things on several services (assuming you e-mail address was being used as the common linkage) then you would simply be breaking the connection at that point. I don't see anything wrong with that, if that's what you want to do. Maybe there could (and should) even be a way of linking two pre-existing accounts when new services hop on the bandwagon. I would hope so.
Facebook is dragging its heels for purely non-technical reasons, and they make themselves look progressively sillier with every new attempt.
The data portability working group have been looking at examples like that since the get go.
I touched on data ownership in http://shaidorsai.wordpress.com/2008/01/11/trus...
and said then that it was something we needed to get under. JP's talked about several times, refered to in that post.
Worried about your content? Stick a creative commons licence on it, if you want attribution and so on. That'll make your intentions *clear*. It won't stop thieves, but at least your wishes are explicit.
http://blogs.zdnet.com/microsoft/?p=1291 ?
[10:33:25] John Breslin says: even without comments, tags, etc. (if they "belong" to other people)
[10:34:06] John Breslin says: the stuff "belonging" to others could still be ported depending on a profile option for those users on the old/new sites
[10:34:40] John Breslin says: "Yes, let my friends port objects that connect us together from this site." / "Yes, let my friends port objects that connect us together to this site."
The actions are slower than the word, but still, one bite at a time.
My biggest worry in all of this is that we'll wind up with poorly thought-out hacks solving the immediate problem, without considering the bigger picture - the more general Web of Data (where the standards are already in place, the Semantic Web stack).
Fortunately the DataPortability folks seem relatively sane :-)
Perhaps the signup process will be more along the lines of "put checks next to all the data you want to be made public."
well, how about snaps?
What I dislike about this discussion is the horrid artificial problem created out of thin air, which then is mounted as a thing needing a "solution".
And that problem is an imaginary one, where tekkies who don't like typing their handles and passwords 10 times over are inconvenienced.
It's simply hard to see that as a problem. Who cares?! You can solve the problem by making your handle and password the same everywhere, which shouldn't matter on these dollar-a-dozen social networks that come and go, and aren't involving monetary transactions but only chat.
Surely some of them will die off and only a few stronger ones will survive and/or be bought out. So why fuss about your inability to port your friendz from Wikitikiwoo.com to Rinkydinki.net when both of them will become extinct by next year?
I also don't like the idea of having some social-mark-up that might inevitably enable some entity to have all your social data centralized somewhere under the guise of open-source and under the guise of you being to manage and apportion it out to this or that social network.
At some point, the places that exist on the Internet that you will be "on" aren't going to be the places anymore, such that you have to sign up for them.
The place you will be "on" will be your phone, with its unique number and password. Obviously you'll control what you want to be on that mobile phone yourself, linked to that number and password. You won't need any hand-holding from any opensourced wikitarian to do this.
And that phone will then access all the stuff out there and maybe not even bother to log in anymore -- why should it? it won't need to. The websites will be the connecting spaces between phones.
Yes, there will be multiple phone services that will not have interoperability. Good!
If you can, email me at matt at socialthing dot com...
Facebook has officially given the finger on Data Portabilty.
I've been emailing their corporate communications coordinator back and forth asking why/when/how the "export to CSV" feature was deleted. I asked if there was someone with institutional knowledge and expertise on the history of the system to know, and was told that there isn't such a person.
*cough*bullshit*cough*
Isn't it just easier to tell the truth?
Look at a really core example: person. If you look in a law book, the flesh and blood you will find that there are different concepts of person. You as a reader of this blog are a natural person or, at least for our purposes. A "legal person" can also be, for example, a corporation, a trustee, a partnership, etc.. So, a single natural person can have a wide variety of relationships and types of relationships with any number of legal persons. In fact, persons in general can have a wide variety of relationships with other persons; natural persons are just a special case. Data portability has to consider:
* Relationships between people and other people (natural or otherwise).
* Trust relationships between people.
* Transitivity of those relationships.
Now add back the reality of an online identity. Identity is not a person. It's the handle to interact with a system. Data portability is based on the idea that if "you" just had a portable identity, problems would be solved. The misguided assumption is that multiple identities are just an artifact of using computers, and that it's a problem that goes away in the real world. It doesn't. Identity is just a handle that ties some identifier such as "robertscoble" to some person, with some degree of confidence. So now the question is which person should it tie to? Robert Scoble? Scobleizer? Is Scobleizer more useful as an identity or as a "person" in it's own right? What relationship does either of those "people" have to the real Rober Scoble? Do you really want those tied to the same person as your birth certificate, passport and IRS records? Would you want that ability if, for example, somebody started publishing as "Scobleizer" on some other blog? If not, how could any of those identities legitimately be called yours? The boundary between you and your online presence is fuzzy. That adds to the data portability problem:
* Relationships between authentication techniques (passwords, etc.) and identities.
* Confidence in those relationships.
* Transitivity of the confidence across systems that are themselves authenticated to varying degrees of confidence.
* Relationships between identities and people.
* Trust between people and the sites that use the identities.
Whatever movement we get this year will have to be very careful about which part of the problem they try to solve. OpenId has done a really good job of making clear what pieces it solves and which it doesn't. It doesn't solve a lot of things, but at least acknowledges as much. I'm not so sure that DataPortability.org has yet established exactly where their boundaries are and should be.
I am VERY hopeful that the face to face opportunity to meet at the Data Sharing Summit will move things forward significantly. http://www.datasharingsummit.com - I hope you and other deeply concerned about getting this problem solved can come - there is nothing like a white board and and a real live face to face conversation to make progress on difficult issues.
The event is being run in the tradition of the Internet Identity Workshop that has been instrumental in progress in that realm. http://iiw.idcommons.net
Then you look at something like Twitter or Jaiku less as THE social network, and more as a social application. Your friends see your updates via FriendFeed, not via Twitter. They can comment directly on the FriendFeed site.
In such a scenario, switching from something like Twitter to Jaiku or Pownce means you don't lose the connections you have. They keep up with you via the lifestream cloud. Same idea could apply for moving from Flickr to SmugMug. People don't keep up with you via the social app - they follow you on the lifestream aggregator. A backdoor way to data (or maybe social network) portability.
More thoughts here: http://tinyurl.com/29cu6k
"We really need to push for an Open spec on social markup. A downloadable file (like OPML for feeds) marked up with microformats containing your social profile information... all you should have to do is import this file containing all info..."
That would be FOAF. It was around a long time before the current crop of social networks, and has had an awful lot of work put into it by some very smart people. I'm surprised nobody here has mentioned it yet, as it's an idea whose time has clearly come. See it at: http://www.foaf-project.org/
The Microsoft strategy appears to start with inviting your friends and connections to connect on Windows Live Messenger (not sounding a lot like portability here -- I am thinking "import from").
So I tried the only currently available option -- Facebook. A login to Facebook screen (with Windows Live logo but a Facebook URL) popped up, and the first try on login failed (hmmm, a phishing site?). But the next screen had the Facebook logo, and it logged me in just fine. I didn't however, see where I could add anyone to an invite list, so . . . I gave up and started blogging.
So we'll just keep beta testing while Microsoft keeps building . . .
http://carterfsmith.blogspot.com/2008/04/social...