The current social web 'exists as a number of “closed silos” that limit their users to relationships with those who have accounts on the same site' (CfP). Discussions surrounding a Federated Social Web envision a better model where '[d]iverse social networking sites could federate using inter-operable standards to share social data' (ibid).

While this is a great step forward, it preserves a model that limits the autonomy of the users. In discussions of the federated model, sites run services to which humans apply for membership. Those sites then share their users' data, through federation protocols, to other similar services with a different set of users. While none of the following are strict requirements, the services are envisioned as:

• having membership
• hosting content
• transferring content to other services

These services are still silos, albeit considerably more open than their predecessors: Someone else hosts and effectively owns a representation of a person's identity and some segment of their content.

A better model would be a person hosting their own representation or representations of their identity or identities and all of the content they wished to share in diverse ways on the internet. Content would be provided, by URI, to services that act as engines of presentation and aggregation. In this model the relationship between traditional content providers (e.g. Flickr, Twitter, etc) and content creators (you and me) is inverted.

Unfortunately, expecting everyone to host their own content repository on the web is not reasonable for many reasons technical, economic and political. There is, however, another option: In these modern times people who are unable or do not want to have their own vault store their money in a bank. In return those banks provide a variety of services. Notably they are regulated to guarantee reasonable security and accessibility of funds.

An Internet Bank of Content could be used to invert the social web, placing control directly in the hands of the user while taking advantage of the simple principles of the open web.

The fundamental concept in an Internet Bank of Content (IBOC) is that rather than giving a piece (as binary data) of sociable content to a web service, a participant in the social web would upload or locate that content in their Bank and provide a URI to the service, called a presentation service. The bank holds the content in trust for the individual.

Why is a bank desirable? What would the benefits be for a user?

• All content remains the user's.
• If a presentation service goes out of business, the user's risk or loss is more manageable.
• Any publication of content can be immediately rescinded by changing access policies on the URI (modulo whatever caching policies have been chosen).
• All URIs could have access policies or capabilities statements.
• Anything which can be captured by a MIME type can be stored in the bank.
• Individuals can host their own bank if they want: It’s just HTTP.
• Content can be published to multiple places by reference instead of copy. When changes are made, they need only happen in one place.
• The user has the option of maintaining several content accounts for multiple internet personas.
• Banks could differentiate themselves by providing different levels of service: different ways of managing access control, service level agreements, data sharding and redundancy.
• A suitably well featured bank could provide persistent and versioned storage for the user's entire data, ever.

What would the benefits be for a presentation service?

• They do not need to worry about storage quite as much.
• They can concentrate specifically on the UI and UX of their service, distinguishing themselves through innovations in that area, rather than commodity-level concerns such as quantity of storage.

How does this relate to the Federated Social Web?

"Social interactions [have] become the central point of our communication using the Web" (charter) but those interactions are, out of necessity, mediated through the distribution of content between services. Each piece of content has the potential to be, and often is, archived for future reference and distribution. The content becomes an artifact on the web. Our identities on the social web emerge from the collection of artifacts we each leave behind. Therefore the effective distribution and management of content is critical to the success of a social web, especially one that is federated.

As currently implemented with OStatus and its use of PubSubHubbub, social content is sometimes federated by distributing copies of content to multiple subscribing sites. While this is useful to avoid the so-called thundering herd problem the technique is a) not particularly forward thinking, b) presents a huge challenge to a user who wishes to revoke whatever it is they published: they either need to go round to each site and delete the respective data or some protocol must be developed which enables the same thing. Both of these rely on substantial cooperation from the hosting sites. Better just to revoke access to the one canonical URI of the content and be done with it.

The main objections to this model are captured in the responses to two blog posts that introduced the concepts: Invert the Web and Internet Bank of Content.

Objection one is that the model implies a multi-step process to publish content: upload to the bank, then post the URI of the content to the one or more services that will present it. The counter is that such posting is exactly the kind of automatic handling computers can be trained to be good at.

Objection two is more complex: Why would a presentation service choose to allow their service to be dependent on content located elsewhere? This argument was discussed at some length on the Autonomo.us mailing list with no clear outcome. That people's content on the web is uploaded somewhere as bytes rather than as a URI is an accident of history, not a technical requirement. As such the pattern could be changed if sufficient benefits were the result.

The growth of the social web has caused a huge increase in concern about and awareness of the issues surrounding content ownership and distribution. The benefits of a URI-based model of distribution are now easier to explain as well as easier to implement.

The IBOC concept is of course not new, it is a rehash of ideas that we visit over and over again in information networks. It does however encapsulate some of those ideas nicely and provides a coherent metaphor from which to work. By comparing it with other similar concepts issues can be highlighted.

Jon Udell wrote about a his idea for a "hosted lifebits" service. He'd be "willing to pay for the service of consolidating all this stuff, syndicating it to wherever it’s needed, and guaranteeing its availability throughout — and indeed beyond — my lifetime." In his scenarios that "stuff" amounts to pretty much the entire breadth of his digital creations, be they blog posts or medical records. Jon's emphasis appears to be on permanence: the services would come with guarantees that data will persist. In contrast, while permanence could be provided, with an IBOC there is an equal or larger focus on the capabilities provided by using content by reference.

Vendor Relationship Management, notably Project VRM created by Doc Searls, has many of the same goals as an IBOC: Individual control over the distribution, use and reuse of personal information and content mediated through software tools. However, VRM is primarly concerned with economic relationships whereas an IBOC is more concerned with social and educational relationships (as mediated through content).

Dropbox with its "[a]lways have your stuff, wherever you are" tagline is helping to popularize some aspects of the IBOC concept: One place to put all your stuff and sharing ~URIs instead of files when you want to share with friends or colleagues. As yet no highly visible presentation services which operate with Dropbox URIs have shown up.

Locker "is an open source project that helps me collect all of my personal data, from wherever it is, into one place, and then lets me do really awesome stuff with it." That "awesome stuff" includes sending data to other services and synchronizing with its original source. This could indeed be awesome but is a few steps shy of the IBOC model because full datasets, not URIs, are being transferred. It also preserves the idea that the services are the natural home for pieces of content, rather than being solely for presentation.

The Mine! Project has very similar goals to an IBOC with an emphasis on using cryptographic techniques to ensure that URIs and content are not exposed to unwanted parties and access is strictly trackable and quickly revocable. The IBOC concept is more immediately concerned with maximizing distribution while maintaining good but not strict control.

Each of the above concepts is most concerned with the creation of a platform for effectively storing content. In some cases there are provisions for the creation of shareable URIs. What is left less defined, in the above and in the IBOC model, is how other services are to use the URIs. We have real-world experience showing that systems like PubSubHubbub can effectively distribute notifications of content changes, and there is nothing in the protocol which requires full content be used.

This means, therefore, that the next big step is enabling and encouraging more effective use of content by reference: by URI.

It is perhaps ironic that both Twitter and Facebook have made some inroads into using content by reference. Both are used as systems for annotating URIs and both, where possible, use oEmbed to display the content "behind" those URIs by reference.

The technical roadblocks to a presentation service, which just a few years ago would have seemed quite large, now seem quite small. The advent of technologies such as oEmbed, OAuth, UUIDs, WebFinger, PubSubHubbub, CORS and Activity Streams bring the development of full presentation services and applications, using content from multiple banks, within reach.

That leaves the social and economic roadblocks. As the Federated Social Web charter states, "In order for there to be massive uptake, the federated social web should offer a compelling experience for users." For Internet Banks of Content and similar services, this means creating a compelling experience at the level of the presentation service. The services must be convinced of the value of using content by reference. Users of the web have been enjoying the benefits of content by reference for years: it's time to invert the web and for services to start doing the same.

Thanks to Frederik Dohr for refining the ideas and text in the paper and to the several people who commented on the blog posts and email thread that introduced the IBOC concept.