Yesterday evening (March 14) Sir Tim Berners-Lee spoke on The Future of the Web to an audience of over 200 people at the Martin Wood Lecture Theatre in Oxford's Physics Department.
Berners-Lee focused on the Semantic Web, a subject he has addressed many times in his writing and speaking over the last five years. However, as 'right' as the Semantic Web concepts appears to be, it is hard to detect real progress in his mission. (See The Lord of the Webs Washington Post, 01/30/03 [link may be out-of-date]. Recalling how hard it was for people to understand what the Web was when he crafted it in 1989, Berners-Lee said he's having difficulty again explaining the Semantic Web, for the same reason: "There's this mental leap involved.")
He began by noting that he was interested in the gap between the microscopic and macroscopic rules, and elaborated the rules behind email, the Web, Wikis and Weblogs. The rules for email are store and forward, with no trust infrastructure. He noted that these rules allow for scaling in the academic environment (except with lists) but not commercially (vis junk mail). Rules are social as well as technical, he noted. The former seems harder when you are a geek. A technology that is going to happen is going to need social rules.
The Semantic Web can be described by another set of rules, he continued. But at this point his talk became overly technical and unmediated. Berners-Lee understandably loves the concept of the Semantic Web. Despite his acknowledgement of the importance of social dynamics he is focused on its technical side. He did usefully assert the maxims: 'It is a good idea to serve useful stuff, and make useful links' and (on the challenge of building ontologies) 'Do your bit. Others will do theirs', and noted the problem that the Semantic Web community has tended to only convert data as it needs it, and has a habit of putting data into OWL and RDF but "not leaving it out there".
He also urged us when we got home to create a URI where people could look up information about us. But it still wasn't clear for the semi-lay person how this whole schema works, how to work with it, and how it might grow.
There were some interesting individual insights on: The relationship of Subject > Property > Value. Semantic Web data "Won't go back in a database" and has to be a graph [not clear on this]. "I have a hunch we should be building a fractal system." The Semantic Web as a set of building blocks for all kinds of trusted systems.
He also showed an intriguing map on the subject of applications connected by concepts (slide 19/37 of his presentation), that more than anything explained a key Semantic Web concept. He illustrated this by talking about adding GPS information to photograhs, and showed how some ontologies could be restricted to (and defined within) specific systems.
Discussing why the growth of the Semantic Web has not been instant he noted that the Semantic Web 'wave' takes a long time. Going back to the Web he noted that the first memo was published in 1989, the first code created in 1990, the Web was on the Net in 1991, and in the press (The Economist) in 1992. It used to be hard to explain what the Web was. "People were scared about being lost in hyperspace, or that it should be hierachical" he said. After the Web it is hard to explain this! The value of your bit is dependent on the value of the other bits out there, he noted. But he revealed his technical bent when he said "I tend to write the programme first and figure out what it is afterwards!". It is not clear if this rate of progress indicates that the Semantic Web is ahead of or behind in its growth.
Of course, the phenomena are not directly comparable. "We have the eyes of the world on the Semantic Web" he said. With the Web, we were just working with high energy physicists, and they needed to share data. They were "a good petri dish". Also, he noted, "data isn't as interesting as Web pages!". "There is much less 'Woohoo!, I have got my spreadsheet on the Semantic Web'."
On the 'good news' he noted that If you make the data browsable certain things will come out. But the examples he gave came from life sciences and geo-spatial systems. They used to be all about intelligent travel booking. He really needs better scenarios.
Among the good news was the recognition that user interface matters, at least in communicating the concept. "I used to argue that the Semantic Web wasn't about [presentation]" he said. He then demoed an 'Ajax' browser application, which was moderately compelling. "For a long time people thought they needed to display data using circles and arrows, but this takes up a lot of space" he insightfully observed. He noted that he wanted to create a system that is as powerful as iTunes, iPhoto, Quicken for people to parse, access and view their data. "You should be able to find all photos from a particular date, where you were, bank statement entries from that date, and just fly though it" he said. This sounds more compelling!
My concerns are that we still haven't mastered the Web. People still don't understand sharing information on the Web and what its affordance are (or its use isn't supported by their organisations): information is in one place, up-to-date, editable. We still email things to people that should be posted centrally. The characteristics of the Web are: the blank page (which he talked about in his book Weaving the Web, Texere Publishing, 2000), no central database/hierachy (as he noted in this lecture), and simple markup. Compared to this, is the Semantic Web setting the barrier too high, and demanding too much collaboration? Is what we have now good enough and in fact a barrier to progress to the Semantic Web. More generally, do we as a society have the ambition to make the Semantic Web?
If the travel scenarios Berners-Lee used to use were so compelling, can he tell us to what extent the travel industry has adopted the Semantic Web philosophy, and if it hasn't tell us why?
And if social dynamics are so important (which they are), he had suprisingly little to say about the blogosphere, which incorporates elements or characteristics of the Semantic Web and is very successful. It would be useful to ask what we can we learn and incorporate from this phenomenon.Other reports:
Sitting at the feet of Tim Berners-Lee, Laura Petersen, New Media Awards 2006 Weblog. A similarly skeptical report, including comments from the talk that I hadn't noted, and also usefully interviews a number of attendees.
Sir Tim Berners-Lee is Director of the World Wide Web Consortium, Senior Researcher at MIT's CSAIL, and Professor of Computer Science at Southampton ECS.
The lecture was organised by the e-Horizons Institute in collaboration with the Oxford Internet Institute, the Oxford e-Research Centre and the Electronics and Computer Science Department of the University of Southhampton, with sponsorship from the British Computer Society.
For the Roger Needham Lecture on Ontologies and the Semantic Web, given by Professor Ian Horrocks at the The Royal Society last year the goal of Semantic Web research was defined as being "to transform the Web from a linked document repository into a distributed knowledge base and application platform. Key to the realisation of this goal is the use of ontologies to capture knowledge that will help automated processes to better understand and exploit Web content".