杏吧原创

The next big thing on the web is widely supposed to be computers learning to 鈥渞eason鈥�, as the long march to artificial intelligence begins in earnest. Park the hype, and the new 鈥渟emantic web鈥� is still revolutionary enough that your computer could use it to book a dental appointment at a practice near you 鈥� without being asked 鈥� or allow researchers based anywhere in the world to share and explore their data as they hunt cures for diseases. The man behind the web, however, is the unrevolutionary and private Tim Berners-Lee. Personal information about the man, appropriately enough, has to be gleaned from the web he proposed back in 1989. What really counts for him are ideas, ideals, global utopias, egalitarian communication and neutral open formats 鈥� all markers for a lack of ego in a cyberworld. Everything from being the son of British mathematicians who worked on one of the earliest computers to finding a faith (Unitarian Universalist) committed to being compatible with reason is grist to the mill. These days, Berners-Lee peers out an office window in MIT鈥檚 futuristic Ray and Maria Stata Center in Cambridge, Massachusetts, where he is director of the World Wide Web Consortium. Ivan Semeniuk dropped in to find out how the web is reshaping society

What is the difference between the web today and the semantic web? Will it bring us closer to artificial intelligence by automatically sifting through all the information for us?

Perhaps too much fuss has been made about automated systems and agents. Lots of things that people think are spooky and magical have come out of the web because of the wonderfulness of human creativity which has been applied to it. But the web is just about putting documents on line, which sounded boring at the time. The semantic web is about putting data online, which sounds boring now, but when we see the data connecting up it鈥檚 going to be powerful.

On an everyday level, what will it let me do I can鈥檛 do now?

Currently scientists 鈥� or science writers 鈥� can get someone鈥檚 paper from an online journal, but a frequent cry is: 鈥淚 want to see your data.鈥� The problem with journals is they can publish the paper but not the data. Often scientists have data in different 鈥渟tovepipes鈥� 鈥� vertical systems which don鈥檛 connect horizontally very well. You may have to go to one screen and write something on the back of an envelope and paste it somewhere else. This is like life before the web. The main thing about the semantic web is that we need the data on the web. We still want words and music and poetry 鈥� but we need data too.

鈥淲e want words, music, poetry on the web 鈥� but we need data too鈥�

How will it work?

Right now there鈥檚 no standard way to put data onto the web. If you go to a weather site, you can鈥檛 just pull off that data and drop it into a spreadsheet 鈥� well, you can write a program, but it鈥檚 horrible. An HTML table is no good because a computer won鈥檛 realise it can treat it as a table. We have created a mark-up language: just as HTML is a mark-up language for text, RDF is a mark-up language for data.

What does RDF let you do?

Put data on the web. It also allows you to say what data 鈥渕eans鈥�. Suppose you have two people who put data on the web. One has a column called 鈥減ostal code鈥�, the other a column that says 鈥減ostcode鈥�. The computer doesn鈥檛 know this is the same thing. So in RDF we write not only actual data but information about that data. When you make a few connections like that between the data, you can start to use all the data you have access to as one big 鈥渄atabase鈥�.

How will people use this giant 鈥渄atabase鈥�?

Well, computers could book you an appointment at the dentist or hairdresser near you 鈥� without even checking with you. But we鈥檙e really looking for serendipitous reuse 鈥� if you put it out there, then people you haven鈥檛 met will be able to do stuff with data that you haven鈥檛 even imagined. If you think about it, that鈥檚 always been the value-added of the web. People don鈥檛 yet realise how important this is with data. But they will.

Isn鈥檛 this all rather like what happened when the web first emerged?

In a way it鈥檚 very similar. There are some people who don鈥檛 get it and they won鈥檛 understand it until they are suddenly blown away when they find themselves with these very powerful data-handling tools. There are others who do get it and they will just start using it.

There are a lot of people who get it in the life sciences, because they have this data integration problem in spades. They have very urgent needs. There is so much hanging on the ability to make a new drug to cure a disease. There鈥檚 clearly a lot of money tied up in this as well. And they have huge amounts of data. So in the life sciences there鈥檚 been a very strong understanding of what鈥檚 possible.

Aren鈥檛 there also huge privacy issues?

People might naively think of the semantic web as just a pool of data, but in fact different parts of it come from different places and, legally speaking, can be used for different things. An awareness of this is very important. We can design systems that are aware of the policy aspects of data. There is some data you won鈥檛 necessarily want to share with anybody, but there is some data that you鈥檒l use in a business transaction and it鈥檚 not practical to write software which will prevent that data from being used by anybody else. Here at MIT, I鈥檓 working on this with a 鈥渄ecentralised information鈥� group, and the conclusion we鈥檝e come to is that you won鈥檛 really be able to stop someone from getting at your data, but if they are a government agency, for example, they will have to handle that data in a transparent way, so that you 鈥� or the judge 鈥� can find the data they used and whether they used it according to the laws.

How far is the web today from your vision?

What frustrates me is that the web was designed to be a creative play space where people could work together on the same piece of hypertext. They would not only use it as a form of communication, but leave a trail for people to see why they had done things a certain way. From that point of view, the web hasn鈥檛 met its potential by any means. In a way, blogs and wikis are both signs that this is changing, but they are still very crude in terms of the hypertext editing you can do. We need better tools for being creative on the web.

What stopped the web becoming the creative, collaborative, play space you describe?

One of the technical things that happened was that HTML became more powerful as a presentation medium than it had been at first. So while it was quite easy to create a browser, it became quite difficult to create a good editor. To create one that has the same features as HTML is really quite a challenge.

One of the social aspects of people really collaborating is whether there are social controls over the information 鈥� whether, for example, you guarantee the privacy of what they are saying when they are in a small group. Allowing people to write a page is not just giving them an editor, it鈥檚 allowing them to create groups and granting them read and write authority. But that is complicated, and we have to learn about how people like to make groups and learn about the social systems involved in collaborations as well as the technical side of things.

Is this notion of the common good, usability and 鈥渘eutrality鈥� of the web down to you?

Remember, I came along and designed the web 20 years after the internet had been invented. It more or less went without saying that the internet was designed not to care what was done with it. It just moved packets of information from one place to another: the fundamental properties that make the internet work could not be held to ransom.

The internet is all about division between layers. Because the people who built it made it universal and non-discriminatory, I could come along and invent HTTP. All these new things can happen on top of the web because it is also a universal medium, there is no central point of control. The web tries not to prefer one sort of information over another, it鈥檚 a flexibility point around which new applications can be built. Independent development of the layers 鈥� the internet layer, the web space layer and the components on top of that 鈥� is good engineering and it鈥檚 also good marketing.

So did this open, egalitarian attitude that pervades the web just 鈥渆merge鈥�?

The web needs to be the way it is to work. There were lots of systems that were invented to operate differently. There were all the dial-up information providers, which were very resistant to letting people break out and access information from different places. There were lots of centralised systems which just didn鈥檛 take off because they didn鈥檛 have the openness, the universality properties of the web.

Does being decentralised mean you have no choice but to tolerate inconsistencies, and cannot impose control?

Before the web, and even now, a lot of systems were being designed to be completely consistent. The way we鈥檝e traditionally done that is to make top-down hierarchical systems, whether in organisations or in programming. This has always been considered a good thing. The maxims of top-down, structured programming are 鈥渋nformation hiding鈥� so that modules don鈥檛 see into each other but are black boxes tied together at the edges. Having a place where someone could just jump in from outside is harmful in that kind of structure because it鈥檚 difficult to maintain this nice order if people keep doing that.

The maxim of the web, however, is if you have something important, give it a label and then people will link to it. So then we get all this mess of what happens when a link no longer works or when people disagree about what a concept means. But that is only the mess that humanity is in anyway, and by trying to constrain ourselves to use hierarchical systems, we鈥檝e reached the limit of scale.

So is the web closer to the way human beings actually interact?

Yes, except when humans are constrained to operate within a hierarchy, for example in a medium-sized company, when the cleanest way is for everyone to know where they are. It鈥檚 just that if you try to organise a large company or nation state that way it doesn鈥檛 work. We can鈥檛 manage hierarchies above a certain size.

The printing press changed society way beyond expectations when it was introduced. Are we in the middle of a 鈥淕utenberg moment鈥�?

The web is changing society dramatically, in ways that we have to be careful to track. The web is more connected than tree-like systems, so it can be more powerful as a way of connecting people, and more powerful as a leveller and as a divider. I think we have to be very aware of its power as we develop new web technologies so we make sure we鈥檙e using them to make a better world.

鈥淲e have to be very aware of the web鈥檚 power as we develop it鈥�

That said, we still have a limited number of things we can think about and do, a certain amount of enthusiasm, of time, energy and talent. The web allows you to choose what you鈥檙e involved in from a much larger palette. So if you do five things a week, two may be in different countries. If we learn as individuals to break out of hierarchies which in the past have been largely geographical, then the world ends up more connected. And when someone is crying out for a solution in one part of the world, maybe the number of clicks between them and the solution will be reduced.

Could the web also foster closed worlds, where extremists only talk to other extremists with narrow views?

This has been a worry from the beginning. The other worry is that when everyone is connected, diversity will vanish. I think what we need in society is a lot of different structures between these two extremes. If you look at your own life, you鈥檝e probably got interests in different places with groups of different sizes. You connect these groups and end up reducing the amount of inconsistency between them by making sure the local group thinks with the global mind, and hopefully that the global group thinks with the local mind. I鈥檓 basically an optimist. I don鈥檛 think humans have done very well at all times throughout history, but maybe connecting ourselves more efficiently and thoughtfully is a way that can help.

You recently started a blog. What鈥檚 that been like?

It鈥檚 fun. After my first blog I got over 500 replies from people saying 鈥渢hanks for inventing the web鈥�. I had to turn the comments off because it was overloading the thing. But now I鈥檓 part of a group who are blogging about the semantic web and so we鈥檙e sharing our excitement about its potential.