Thursday, November 5, 2009

Linked Open Data : Can we learn anything from failure of many B2B Exchanges?

"You can't stop an idea whose time has come!" Linked data seems to be that idea in the grand vision of Semantic Web! The growth of Linked data cloud always reminds you that we are in exponential times! The Semantic technology community, including me, believes that Linked Data is the best thing happened to the Semantic Web vision. It makes sense and it is the next step for web and can also contribute significantly, if done right, to the evolution of this civilization. The possibilities are endless! We also have one of the best brains behind this initiative.  Whenever I hear Metcalfe's law (by Bob Metcalfe) , in context of Linked data, which states that "the value of the network is proportional to the square of the connected users of the system" then it always reminds me of those early B2B days during late 90s when the world was about to change. Metcalfe law was very popular during those times also! B2B exchanges were considered the pillars of the new economy and their valuations made us lose our sleep. It seemed to all of us that everything was going to be re-defined and we wanted to be part of it. Well, it didn't happen exactly as we were led to believe! Despite the differences between B2B and Linkedata like one is trasactional and other is knowledge oriented, privately owned vs free, different technologies, different era etc., there are some similarities in both of them. Lets do some introspection so that we don't repeat some of those mistakes!

B2B (Business to Business) exchanges is an entity which brings multiple buyers and sellers to a  marketplace where all kinds of commodity, financial instruments, intellectual property and various other goods can be eletronically traded - web is used as the medium in most of the cases. It has following characterstics:
  • The perceived value followed Metcalfe's law
  • Enterprenuers could fundamentally re-invent how work gets done
  • No longer comparison between big-small and so on
  • Shift of power from producers to consumers who are in control of everything
  • Standardized marketplace and standardized contracts
  • Markets operated at fraction of physical word cost
  • Global reach and one stop shopping
  • Neutrality, transparency, self-regulation, market efficiency, confidentiality and anonymity were other virtues of this marketplace
  • The winner, of a particluar vetical B2B , takes all 
It did have some good principles but a very large number of the B2B exchanges failed within few years. There were few fundamental reasons for their failure:

  • The internet enterpreneurs didn't really understand their place in the overall marketplace
  • What they offered to business was something that already existed - at least in some form
  • Companies had always done business with other businesses and most of these businesses negotiate to lower the price. Adapting those existing processes to an internet format didn't really create anything new or different in the field. Unfortunately, many companies felt that switching to an internet-based sites controlled by third parties was a riskier bet than staying with the current partners and vendors
Few of these exchanges figured it out early and survived by either addressing these issues or by being creative about it - like some of them just focused on small businesses. ECN (Electronic Communication Network) are another success story in the world of B2B exchanges as they allowed a more efficient price discovery mechanism for stocks and currencies.  In the end, B2B exchanges were all about attracting buyers and sellers or producers and consumers which they couldn't do. In a similar way, the success of Linked data cloud will depend on creating a marketplace which should be able to attract producers and consumers or buyers or sellers. The technical design prinicples by Sir Tim Berners Lee are just great and well thought of. We are also aware of the existing issues in the Linked data Cloud like quality of data, what is out there, disambiguations issues, trust of the source, frequency of the update, how to get started, should we always publish in RDF, how do you erase inaccuracy and many others things like this. We had very similar issues, maybe in a different flavor, when Web 1.0 was developing and we have come a long way. We might say that HTML was much easier to adapt then RDF but the success or wide adaption of the original web was not only simplicity of HTML and HTTP protocol but because it was a great sales, marketing and ecommerce tool. In context of Linked data cloud, lets accept the fact that most of the businesses will not be  interested in "the future," they will always be interested in "their future." Their first responsibility will always be their share holders.

Companies need very strong incentives or business case to publish their data to this "Cloud" and write applications to consume the collective intelligence of the nodes in this cloud. Initiatives like exposing government data (UK and US at this point), dbpedia, scientific, map oriented data, music, individual research projects and so many other examples which can be good and interesting from exploration standpoint are already underway. We will also see more invovlement from non-profits and intelligence communities at some point. It is a great effort but still not enough! The Cloud will continue to grow and will add billions of more RDF triples but we need to proactively involve the corporate world. Technology enterpreneurs for Semantic Web or Linked data can't do it alone. They have to partner with the business community and the enterprise. The value proposition of Linked data needs to be articulated to the CXO community and their participation needs to be encouraged. We have to start having more conversations like the "benefits of publishing data to the cloud for the enterprise." Basically, much more on business development, business case analysis, education, marketing then technology alone - technology will always remain important and core. I had written an earlier blog about a new category name for Semantic technology which talks about some aspects. Once we have their attention and buy in then the ecosystem of new tools, applications, security infrastructure, new architecture, developers, APIs, SEO, creative sales and marketing ideas, consulting, outsourcing, selling and buying and many either good things will start maturing. It will be very important to do so in next few years to sustain this momentum of Linked data cloud. It will also attract significant investments in this effort! Don't we think that there are dollars to be spent on the Linked data cloud in the  worldwide IT spending of more than three trillion dollars in 2009. We should be thinking or advocating about the "guidelines" for  publishing to the Linked data Cloud like:

  • The benefits of Linked data and possible use cases.
  • How are they impacting their supply chain? 
  • What are the advantages in reciprocity of links? 
  • What should be the process of identifying the data which needs to be published
  • How will it help their revenue and help in their relationships with partners and customers?
  • What are the security issues? What kind of security tools or approaches which are out there?
  • How will it mitigate risks for the company?
  • What will be the possible compliance and legal issues for them?
  • What data access choices can they have - free access, partial access, sign up , qualification, payment?
  • What are the data packing options like - depth, breath, granularity, freshness etc..
  • Can they have handle on who is accessing this data, frequency of access, redistribution issues?
  • Can they create virtual networks of data clouds for their partners and business customers?
  • Can they promote their data?
  • Can they increase the findability of their data?
  • How can they do it in a phased manner?
  • If they use RDFa then are they part of Linked data?
  • Are they accountable for consistency, organization, correctness of their data?
I can think of so many more questions. A very clear policies and procedures aproach needs to be articulated which will lead to a governance model for publishing to Linked data cloud within an enterprise. Probably, we will have to learn more from the experience of present initiatives like publishing data from the UK government to the cloud. During the early days of web, if any company didn't have a web presence then it was perceived that they were missing the action, opportunity or future revenue. At some point, today's businesses should start looking at Linked data cloud with the same lens.
Reblog this post [with Zemanta]


  1. Priyank,

    Good post! There are many useful questions and hints in it, so I'm picking a more or less random one that caught my attention ;)

    You ask: 'If they use RDFa then are they part of Linked data?' Now, this is interesting one to me and here is my answer: this question is not specific to RDFa (same is true for other RDF serializations such as RDF/XML) - it depends on how you *use* it. If you *do* use HTTP URIs, if you *do* link your data to other data, in short, if you follow the LD principles, not matter which RDF serialization you use, you are in. That said, I can perfectly come up with RDFa examples that are definitely not linked data (and hence very much offer the same as some microformat). But on the other hand, RDF/XML has been used for years used without HTTP URIs, in data island, etc. - also, no linked data there, really.

    For examples on how to use RDFa for/in linked data, you might be interested in checking out [1]; we 're currently working on this tutorial and hoping to release it soon as a W3C Note ...



  2. Business is a Contact sport. Thus, and B2B exchange is really a Contact Network. Linked Data simply encapsulate all the potentially and possiblities of Networking into HTTP scheme based URIs. Ultimately, there will be no difference between a B2B Network, Facebook, LinkedIn, BlogRoll, Buddy List etc.. We'll just have LINKs, and these LINKs simply expose Data Networks that operate within the confines of the Data Observer's Context Halo.

    LINKs will simply deliver what most assumed they delivered, at the start of the global hypermedia phenomenon we know as the World Wide Web.

    Ironcially, lack of LINK granularity and flexibility is what broke the back of early B2B endeavors on the Web :-)