W3C to Help the NSA Snoop Social Networks?
Here’s an interesting article on the unintended consequences of social networking. Basically, it is reported that the NSA is snooping social networking sites (with the juicy twist that it plans to do so using Semantic Web technology – more on that later). This seems to fit into the category of “examples of why it’s important to have some kind of user-controlled trust / privacy layer in the fabric of the Web.” Who should be able to see information you put online (including your links to others and the nature of these links) and who shouldn’t? P3P addressed some of these issues but it was never widely adopted. Liberty Alliance has built some interesting technology standards around federated identity, but they are not user-centric, they are provider-centric and they do not really cover privacy. An interesting effort called Dix seems to blend the two approaches, but after a quick read of some of their use cases, it doesn’t seem that they cover “prevent the government from snooping my network.”
Or is laying ourselves open to government surveillance the price we pay for living more of our lives in the digital realm?
Discuss!
By the way, on the whole Semantic Web issue, I think the link they are drawing in this article is tenuous at best, but it is true that the Semantic Web architecture is likewise lacking a coherent identity and trust mechanism.
Liked this post? Follow this blog to get more.
Hi Dan, what does ‘user-centric’ mean for you? You assert that Liberty specs aren’t user-centric, but instead ‘provider-centric’. Is it because the specs were defined in part by companies interested in eventually acting as identity providers? I’m unaware of any community of end-users actively defining identity architectures.
Liberty specs can be deployed in a variety of models – from enterprise scenarios where the employee has no control over their identity beyond that initial employment contract signature, to e-commerce applications where the user can choose to explicitly control the sharing of each piece of their identity.
regards
paul
Users control their phone (land and cellular) lines don’t they? Strangely enough, governments are still able to snoop (but only with help).
I guess Liberty is, first of all, about attribute sharing between users and service providers, and among service providers, in order to give users better service and (some) control over what attributes are shared between which parties. With these goals in mind, I can’t think of a way of being more user-centric.
However, one could indeed imagine doing a bit better with privacy (unlinkability) and tracking of released personal data. See e.g. the ideas of the EU PRIME project (start with the White Paper, then go to the Architecture and Framework).
Peter
Comments on my Blog! I can’t believe it… Identity must be a touchy subject.
So first of all, I in no way intended to denigrate Liberty. Liberty provides a really great “platform” and set of specifications for rolling out federated identity solutions that can work across implementations and providers. Of that, there is no doubt. But the scenarios that Liberty is built to support are operator and provider centric.
One example of how Dix might be more user-centric is the idea that a user might be able to switch identity providers and seamlessly migrate their data and preferences from the old provider to the new provider.
And thanks, Peter — I will take a look at the EU Prime project.
I don’t know about ‘touchy’, but it’s certainly a hot topic… particularly when the phrase ‘user centric’ comes into play.
I think one problem is that ‘user centric’ is used to mean so many different things… it’s pretty natural, given how new a term it is – but there’s just no consensus about what this neologism means in either theoretical or practical terms.
The analogy I tend to use is retail banking: as a client, you trust your bank (bear with me here… ;^) to maintain the integrity and security of your bank balance. You don’t insist on keeping all your cash under the mattress – which would be far more ‘user centric’, but far less convenient and more risky.
Similarly, having a model in which service providers and identity providers exchange data about the user is not necessarily incompatible with ‘user centricity’.
Actually the Liberty specs, particularly the User Interaction parts, have always included the option for the user to confirm consent to any attribute exchange between third parties. That option is built into the message flow, and it seems to me to go directly to your points about user centricity and also privacy.
However, the specs are not the same thing as the ‘use cases’, and ‘use cases’ are not necessarily the same as what ends up being deployed. So in practice, many service providers and identity providers prefer (for their own commercial reasons) to address this in other ways – for instance, by including a ‘consent’ clause in some initial agreement with the user, under which subsequent attribute exchanges can take place without asking the user each time.
However, one should be clear that that’s a commercial decision about which parts of the specification to implement, and not a shortcoming of either the specs themselves or Liberty’s general approach to privacy.
Having participated in Liberty’s Public Policy and ID Theft groups, I think its approach to privacy and user-centricity is actually pretty robust. Incidentally, so does the Article 29 Working Group of the European Commission…
Thanks, Robin — lots of good information there. It’s important to separate specifications from actual implementation and as I also said it’s important to understand all the value that Liberty does bring to the table.
I think this all doesn’t really address the general issue I was raising, however, which was to do with identity as part of the fabric of the Web itself.
If you look at how people are interacting with Media these days (for example, this very conversation) it is in a distributed fashion across different services, different providers, different technologies. How can the provenance of blog comments, FOAF files or information found on Wikipedia become more transparent and therefore trustworthy? In the same way that we have a Web of content and services now and that W3C aims to bring us a “Web of Meaning,” do we also need a “Web of Identity” – that is to say, a unified approach to identity across the Web which does not depend on a single, or even a set of identity providers? Social software improves with use and becomes more useful the more users are using it. Do we likewise need a bottom-up approach to identity? If so, this does not obviate the need for the top-down approaches – one simply has to imagine an interface at which these two approaches meet.
Hi Dan, regarding the statement “the scenarios that Liberty is built to support are operator and provider centric”, I agree with Robin. The information could be stored in the operator’s infrastructure or anywhere else (for instance in infrastructure hosted and supported by the user himself).
In the first case (information stored in the operator’s infrastructure), this doesn’t mean that the operators can do whatever they want with it, on the contrary, one of the very needs to deploy identity infrastructure is to protect the user’s privacy and comply with legislation regarding data protection.
Secondly, you also mentioned the need to protect the information related with the user’s social network. One of the features included in ID-WSF2.0, People Service, specifically caters for that: different applications can share information corresponding to the user’s social network, whilst being such information and the conditions under which it is released always under the control of the user. Bur let’s continue discussing if still concerned. Cheers, carolina.
Thanks, Carolina. I stand corrected. Looks like the reference to that is here just for completeness.