The Barriers of Content and Context
[reposted from Darwin, January 2004, courtesy of the Wayback Machine]
Social networking is suffering the curse of all attractive innovations in the modern era: As even the most winning innovations rise into popular consciousness, the backlash against them begins instantaneously. The traditional lag between initial adoption by a small percentage of hip, connected “innovators” and the later contact with the “majoritarians” that comprise the overwhelming bulk of the market has been squashed to an almost immediate effect. Just as truckers’ caps begin to diffuse out to the average metrosexual a few weeks after becoming cool, the glitterati already declare them passé.
So we are confronted with the situation that social networking solutions — at a very early stage in their maturation — are being judged not just by the innovators who have a seemingly infinite hunger for the new and disruptive, but by those who are at best reluctant, if not actually averse, to adopting new technologies and the inevitable changes that they cause.
I am a strong advocate for social networking technologies. I believe that they are heading us in a positive direction, and that immediate business benefits can be realized from their application, such as accelerating the sales process (see Cracking the Social Code). Still, the reality is that many of the solutions in widespread use oversimplify the complexities of social networks and networking, and they will need to be augmented to reach their potential. At the same time, social networking as a whole is now being evaluated by the largest possible community — the world stage, is it were — based on the technologies now in use long before they have fully matured.
Social Content
Social networking involves people in the creation of all manner of content: bios, preferences, postings, requests for assistance and direct social networking information such as social relationship information. Today’s social networking solutions have moved toward a closed garden model, where much of this content is inaccessible except by direct navigation, which forms a serious barrier to use. This means that conventional search engines can’t spider the social network content. For example, the comments I make at Ryze or Tribe.net about instant messaging are not accessible to the same degree or through the same mechanisms as similar comments I make at my blog, or in this column.
Likewise, the content being created and “published” in these services is partitioned in a wide variety of idiosyncratic ways. None of the leading social technology solutions allow users to pull information in the form of RSS (Really Simple Syndication) feeds into a news aggregator. This is a model widely adopted by blogging solutions, and one that cries out to be integrated into social networking apps. Offering e-mail notification in a time when e-mail is being made inoperative by spam is simply insufficient, and the wrong approach anyway.
Perhaps most central to social networking content is the information regarding relationships. While a few of the offerings can import contact information from Outlook, this is intended as a starting point after which relationship information is managed within the social networking solution: again, as a closed system. And of course, without opportunities for interoperability between the various services, at least not yet.
The solution to this is likely to lie with a standard way to express relationship information, such as the FOAF (Friend of a Friend) XML approach (see The FOAF Project), which has been supported by a variety of social technologies, including MovableType, Ecademy and others. In this approach, individuals publish an XML document — either handcrafted or machine generated — that describes interpersonal relationships: who are my friends, how close are we, what are their e-mail addresses or blogs, and so on. There are certainly a lot of issues with publicly accessible XML files that represent relationship information, but some sort of hybridized approach will prove to be necessary if we are going to share social content across the gated communities that today’s solutions offer. For example, FOAF (or similar) representations could be encrypted with permission to access provided only to trusted software or trusted individuals.
Leaving aside the explicit relationships between people as found in FOAF, there is a huge amount of implicit relationship information available that more intelligent (and hard-working) social networking solutions could discover. For example, many of the enterprise social networking solutions (like VisiblePath and Spoke Software) can read e-mail folders to establish the strength of relationships. But few of the solutions are capable of determining relationships or their strength through blog networks, such as relationships that are implied by blog comments, blog trackbacks or “blogroll” inclusion (where one blog references another prominently, in a section like “Blogs I Read”). For those who live a full and active digital life, this information might be more useful than e-mail.
Other publicly available information can be mined to establish relationships. For example, the company LinkSV has amassed what I refer to as a “public” social network — a “yellow pages” for the Silicon Valley high-tech sector — that includes the known relationships between venture capitalists and high-tech management. This data has been amassed through mining public information available from venture capital and corporate information, such as websites, press releases, filings and so on. So the path from Mike to John to Betty can be established based on corporate boards and the history of employment of entrepreneurs. I expect that other companies will rapidly emerge that will present similar network models for other sectors, geographies and markets.
The immensity and complexity of converging and managing relationship content from private and public sources argues strongly for a federated and standardized representation of relationship, a la FOAF. My bet is that social networking services will resist standardization until they see the benefits of converging all sorts of private and public network information, and realize that no one company can create and manage all of it. At this point, in an immature and segmented marketplace, we are unlikely to hear anyone admit that they can’t do everything all by themselves, thank you very much. But at the point of market maturation, everyone will climb aboard that bandwagon.
Social Context
As problematic as the issues surrounding social content are, the issue of social context is significantly worse. The social context problem centers on a person’s heterogeneous social network; really a collection of independent networks, whose purpose and ethos are defined by context. I have one network of social software weenies, a second network of local pals and another collection of contacts from real-time collaboration circles.
Various tangents are being taken on social context. Ryze and Tribe.net support the formation of groups or tribes, which are thematic assemblages of people. Individuals simply discover such groups and join, or are invited to join. Such casual associations are not really much like the stronger affiliations that are implied by long-term personal interaction.
In principle, the context of my social networking could be analyzed. For example, as e-mail and other content is being analyzed to determine the strength of relationships, the context of those relationships could likewise be determined. Several of the social networking vendors state that they are looking at this as a future area of interest, but they have not yet attacked this angle.
On the other hand, more traditional knowledge management companies — notably Tacit — have refactored their expertise management technologies toward social context discovery. These solutions have been developed to perform linguistic analysis of content to determine expertise, under the assumption that someone who generates or reads a lot of content on hieroglyphics is likely to be an expert on the topic. Reapplication of these approaches toward the discovery or creation of social relationships for the purpose of collaboration seems a natural step, and that is what Tacit has done. Note that Tacit has had five or so years to develop the associated linguistic analysis technologies, so we should anticipate that it will take a few years before others begin to roll out similar approaches. Alternatively, technology to accomplish such analysis could be licensed from Tacit, or even conventional content management players like Convera, Verity or Documentum.
In the absence of a fine-grained control over the several subnetworks that comprise my total network, some attempts to leverage my network will simply not work well. Context shifts can diminish the value of social capital in unexpected ways, proving that relationship strength is not transitive across my different networks. A request for an introduction from one subnetwork to an individual in another starts to seem an imposition, while introducing two folks who are mad about downtempo lounge music seems natural.
As more and more of my distinct, context-bound networks are collapsed into an undifferentiated, context-free uber-network, the result will more and more approximate a random sample of people, and this will cancel out the social network effect. Relatively quickly, social software will need to provide a means to contextually denote relationships, so that I can gain or provide access to appropriate individuals or information easily.
As one example, I recently experimented with Eurekster, a search technology based on social network collaborative filtering. In principle, your social network of friends and colleagues share and shape your perspectives on what’s important, who is the most authoritative voice on a topic and so on. And clearly, I would like to know what people in my network think is “hot” today so I can keep abreast of breaking news. However, without the ability to partition the network — based on interests, affiliation and/or reputation — the answers will start to look like Google. The answer is that I really need to be able to pose social context-based questions: What are my social software buddies looking at today? What about my personal friends? What about people in the 20194 area? Until social networks attack this angle, we will be dealing with a very coarse-grained approximation for what is actually going on in social interactions.
Slouching Toward Context and Content
Social tools, i.e., software designed to intentionally shape culture, are going to become the cornerstone of a revolution in information technology. Social networking applications, one variant of social tools, will rapidly expand in functionality through the integration of conventional and innovative approaches to counter the context-and-content limitations of today’s applications.
I predict that 2004 will be a year when we’ll see the successful linkage between rich content systems — like salesforce automation applications, corporate knowledge management applications and blogs, as only the most obvious examples — with social network tools. This will certainly not be social tools nirvana. It will be years before interoperability and effective standards are in wide use, but we will see the initial hurdles of context and content surmounted. At least in part.
