subscribeLooking Back, Looking Forward Tim Berners-Lee
The Web: Looking Back, Looking Forward' Tim Berners-Lee, director of the World Wide Web Consortium. 2007 Tim Berners-Lee was awarded with the Lovelace Medal. The Lovelace Medal was established by the BCS in 1998. Ada Lovelace was a mathematician and scientist who worked with, and was an inspiration to, the computer pioneer Charles Babbage.
Here the lecture notes kindly made available by Dave.
These lecture notes are informal, and although they cover the talk in
detail, should not be taken as a transcription and may contain many
errors, misattributions and so on. Verbatim copying and redistribution
of this entire article are permitted provided this notice is
preserved. Dave Crossland
Nigel Shadbolt Introduction:
This award is named after Ada Lovelace, who worked as an assistant to
Charles Babbage. Its for major advancements in computer systems. Sir
Tim Berners-Lee is the latest recipient of the award.
Since the last award, the BCS increased its visibility and impact as a
professional body and learned society. We have record membership at
over 60,000 and more and more of our members are taking our exams and
qualifications. Our better finances are being invested in our
infrastructure thanks to a great executive team headed by David
Clarke.
In the next year, BCS is challenging preconceptions held by the
public: We are not geeks in basements. Our profession has changed the
world, from our cruise missile controls to our hospital technologies,
and all the communications technology in between.
Tim Berners-Lee's parents were both mathematicians, and were involved
in building the Manchester Mark I, one of the first computers in the
UK. He studies "Experimental Philosophy" at Oxford, or Physics as its
now known, and was involved in computers the whole time, advocating
electronic communication to advance the state of the art in 1980.
The first website was created at CERN in August 1991, and in 1994 Tim
founded the W3C, comprised of organisations of all sizes, working
together to improve the web with standards. He has staffed at the
CSAIL lab at MIT, and in 2001 moved to Southampton where he's working
on a new discipline, "Web Science."
The Web: Looking Back, looking Forward
TBL uses a Mac, with a webpage rendered in Safari with no address bar,
but status, to present with - No fullscreen leetness. The type is Gill
Sans and I guess the diagrams are done with OmniGraffle.
My parents showed me that with a computer you can do anything, if you
can take on the programming challenges, and challenges to the
imagination. You can do whatever you can imagine. Developing the web,
and seeing everything that's been done with it, has been tremendously
exciting for me. So I'm looking back at the process of making it, and
what its like to design things that become large systems, and then how
to build large systems on top of large systems.
Web Science is about when you design something that doesn't run on
just one computer, but on the web, and the web is made of people.
As just mentioned, I studied 'experimental philosophy' as physics was
known back then, and so I'll start with the experimental philosophy of
the web.
Experimental philosophy is finding out about the world, about life,
the universe and everything, through experiment. The semantic web is
building new worlds through experiment. Rules about how computers
communicate across the Internet. Defining new logic, worlds in which
certain things are made possible, where meaning can be assigned to
things. Not just doing experiments, but creating and building things.
The semantic web is building philosophical space. Designing the way
two computers, and then two people, interact.
There is a curious connection between the microscopic and macroscopic.
Web science examines this. There are two kinds of science: Science in
the social sciences sense, and then in the normal analytic sense. But
Computer Science is largely engineering, and people complain its not
really science because its so much about building things.
Its a complicated affair though, building things with computers. I
mean, the web is so big. The worlds population is 10^10, and the
number of web pages is in the same order. The number of neurons in a
brain is about 10 times bigger, though. And these are growing, the
world population is growing dramatically, and the web is growing
exponentially. But the neurons in our head are going down, haha
So ideas design technical protocols, that make up systems. Also, ideas
make social protocols. Such systems are microscopic small things. To
physics this is like figuring that gasses are like a table of billiard
balls, and that explains how they interact to an extent. So you can do
maths with such an idea, what if the table is smaller. The idea is
that when you make a microscopic model, you can try to predict
macroscopic effects.
The web is similar, with some magic thing that happens from
microscopic to macroscopic, and so perhaps sometimes you can model and
predict it. Emergent phenomenon happen when the microscopic system has
a lot of complexity. So we analyse the macroscopic and find issues,
and we have various people's values on those issues. Issues lead
through creativity to ideas, and the cycle starts over. Creativity is
another kind of magical process, and its a very important part of
engineering.
Sometimes you need to go walk, take a bath, do something else, to
solve a creative problem. Art, music, and engineering all have this in
common. And the other magic is collaborative complexity, people
working together, not in a team of 10 or so, but all the people in a
community, at a kind of global scale. Thats the best magic from
microscopic to macroscopic systems.
The first website had instructions about how to make more websites,
and thats how it started growing so fast. Growth on the first
webserver went up by a factor of 10 for 3 years. From 100 hits to
1,000 to 10,000. Another interesting thing was the ratio of visits
from weekends to weekdays. Every person who went to the site, they
would probably install a browser, and perhaps make their own site, and
people would see them webbrowsing, and see their sites, and start
browsing and publishing themselves.
Magic is just stuff you don't yet understand, as I told my kids when I
played magic tricks for them. They're now amazed they ever fell for my
magic tricks, haha
So when you have a creative moment and idea, and develop it into a
system, you hope it will have this emergent property. There's a lot of
second guessing in this.
What was one of the earliest internet technologies, after file
transfer? It started with an issue, the need to communicate to people.
So the concept of 'internet messages' developed, with SMTP technically
and a friendly academic community socially: Email. An interconnected
academia, and then an interconnected public, created a phase shift,
and we had spam. This happened a while before the web.
For the web, the issue was 'cant access information'. I was at CERN in
the 1980s, with lots of computer systems, lots of different operating
systems and documentation systems, used by people from all over the
world with different home institutions and funding providers. So there
would never be a unified platform there. That's why it was great fun
to work there with those people, of course.
All the information was on hard disks, not paper, and it was
impossible to access. So you used to find people in the corridors, at
the coffee intersections, who could access what you needed to read. So
you'd go there and just wait, until the people responsible for what
you wanted to do turned up, and then you'd buy them a coffee and
croissant, and get them into what you were doing.
So coming from a technical background, I found this social engineering
a bit strange. I thought it would be nice to have access to all the
documentation immediately, without asking anyone. And I could imagine
a virtual system that published already existing data. So that's the
idea of hypertext and the net.
So we had URIs, HTTP and HTML on the technical side, and links and the
incentives to make good, valuable links, on the social side. These two
things made the user-client, server/publisher system of the web
possible. This had an emergent magic, but many other things I did
lacked this. With this magic, there was this 'web explosion.'
People felt that because they could read anything, they wanted to read
everything! People got a kick out of just seeing what was out there,
burning the midnight oil, surfing the web. So now the new issue wasn't
one of access, but of finding the right information.
When I designed the Web, I knew it had to be a universal space. I'd
seen people design 'the documentation system everyone should use.'
First the mainframe people, but no one logged into mainframes any
more, we all used unix workstations. Next the unix people wanted
things in SGML, but we used word processors on PCs and Macs by then.
So I wanted a universal system in which you could put anything. That
meant you couldn't ask anything else from people - to use a particular
file format, or to pay tuppence per link, haha
I wanted something that could be a scribbed note like those we make
over coffee, and change it into a working document, and then into a
formal specification, all without having to change how that document
was accessed.
The universality was the most important thing about the web; it didn't
dictate anything else.
Another essential was the layering. Things are designed independently
of application. Foundation technologies instead of ceiling
technologies. Online music stores are good examples of ceiling
technologies, there aren't hooks all over them to build new things on
top of. Its important to build things as foundation technologies,
that's what I learned with the web. Open, royalty free standards,
clean interfaces that make no assumptions, and flexibility.
So the web problem became finding things. Google staggered people when
it was released, because it was a huge problem, not finding stuff. The
idea of indexing things... We did early experiments that showed that
it was impossible to index the web, due to exponential growth. So the
index was superexponential, and superimpossible. But actually, it was
possible, with an eigenvector algorithm (technically) and the links
socially – Google style search tools rely on people who make
hyperlinks, that socially people are finding good pages and linking to
them. So you had the Google site, as a microscopic system, and the
Google phenomenon as a macroscopic thing, that people started linking
to valuable pages more. So then the new issue became web spam:
websites that look to googlebot as much as possible like a real
community site, to spam the search results.
Also, Wiki webs. The original web model was that everyone could create
links and edit everything. I never imagined that people would be happy
to just read the web. Wrong again, haha
But some people weren't happy, and the issue for them was 'can't write
stuff'. So the idea was to use forms to edit content, and form-based
editors was technically on the technical side, and the social side was
the idea that literally anyone can edit a page with respect. So that's
the microscopic wiki system. And the Wikipedia is the macroscopic
example. And this leads to wikiwars, and Wikmedia Foundation is
creating a "wiki process" out of them, finding a meritocratic way to
make us all smarter than just one of us. Thats an ongoing process, but
its exciting and something to 'watch this space.'
Blogs, with their trackbacks and such, are also important to mention,
but of course I'd now like to focus on the semantic web.
The issue is, 'can't reuse data.' The idea is data sharing.
Technically: RDF, OWL for ontologies, SPARQL as a query language -
that's just being finalised as a standard now - and RIF for writing
rules like your mail filtering rules and calendar alert rules - the
standardisation process has started on this.
Socially, we use URIs, make links, and share ontologies. So the
semantic web is about data, but its also a web, that allows people to
pull together and see things that are related. Leads to microscopic
semantic web systems. The macroscopic systems are things like FOAF,
and Life Science applications. Who knows what the next issue will be?
Basic tenets of the semantic web?
Everything has a URI. Don't say "colour," say
don't like other peoples precise definitions, you can make your own.
There's a concept of 'definitive meaning', its not like words, you can
actually look up what something means.
So we have a big stack, from URIs and Unicode up through XML, up
through RDF, SPARQL, RDF-S, OWL, RIF, up to Unifying Logic, up to
Proofs and then up to Trust. And Crypto covers the stack from XML to
Trust, of course. (Nice diagram of this on the slides)
What the progress in shapes of data? We start from lines (tapes and
cards) through matrix (databases) and into trees. SGML, XML, top down
structured design, Object Orientation programming. These all worked
well for even large companies, but they don't scale forever.
Eventually you get into nets: Internet, WWW.
There's an old chestnut about the net, that you can cut bits out and
things will still get from A to B. Don't try this in your house, of
course. You are not meant to connect to your neighbour. Now you might
connect to your neighbours wireless network, subject to the terms and
conditions of their service provider and prevaling legal conditions,
of course! haha
So the shape of the web. Societies include communities on many scales,
and a universal WWW must also include communities on many scales.
Home, work, play. Society is a complicated and overlapping set of
communities. On the semantic web, there's a direct connection between
a set of people and the shared terms they use. (There's a neat image
here like a TFL tube map)
Remember fractals? When they were cool and interesting, and cool to
have them as a poster on your wall? They pop up everywhere, in nature
and on the web. The total cost of ontologies is fractal in nature. As
the number of people in a community sharing an ontology go up, the
costs of maintaining that ontology go down logarithmically. This is
true if the community is made up of individual people, or teams within
a company, or companies within a market.
So when I'm in a given group, the amount of work I put in goes down,
the larger the group is. If you do your bit, and others do theirs, the
costs of doing things go from infinite to finite to obtainable. If you
get bogged down, as sometimes happens, find another group, haha
So things have been done by geeks. Great, nice, lovely people, solving
a problem right at that moment. But the common wisdom is that you're
meant to find out what your customers want, and solve the problems
that they want solved. But sometimes people go out and design things
however they want. So if we just have a list of what's not working
right, we miss out on what we can imagine. Its those things that are
the most disruptive and world challenging things, and we shouldn't
lose sight of them.
So I want to blow spreadsheet programs away. How do you give people
total access to all the data that's relevant to them?
Huge issues in doing this: Privacy, Identity, Appropriate Use – not
such a big deal in the UK, but this is about data collected for one
use, and actually used for another – a big deal in the USA, where the
government are collecting lots of data in emergencies, but no one
checks how that data is used after the emergencies.
There must also be Resilience. Resiliance to social breakage,
phishing, and technical breakage, slashdotting, and web breakage,
Error 404s. And there are people out there trying to break things. So
we need to respond to these kinds of threats.
The cost of pixels is going down. New devices is a big challenge.
Great diversity of things, portable things, and things in developing
countries. Go to times square and spot the neon sign – its now all
pixels. Instead of light shows, most of the stages are now pixels. The
main arena in Boston is now all pixels for ads. Rock concerts are
programmed. So when pixels are everywhere, what will happen then?
Perhaps when I'm in the town center, my phone can negotiate via
Bluetooth to use the screen space on a nearby shop screen to show me a
map of the area, to use the mass of pixels in front of me instead of
the small sample in my hand. Is this possible? Its a lot of work,
sure.
And the OLPC and cheap phones are coming to developing countries,
which changes things a lot. Check out the W3C Mobile Web for
Developing Countries working group!
"Intercreativity," a word for both kinds of magic at the top and
bottom of my diagram, the interaction of microscopic things and the
creativity of turning issues into ideas.
I want to get creative in my religious community group. With my group
at MIT. With my family to plan our holiday. Fractal groups, from our
new geography-free connectivity. So we need more intuitive interfaces,
new forms of democracy, and to connect half-formed ideas. If I have a
half formed idea, how can I leave a trace of it, that someone else
with the other half can find?
So, web science is going multidisciplinary, and we are working on all
of this between Southampton and MIT, even getting students travelling
between the places, and doing it under the banner of the Web Science
Research Initiative – www.webscience.org
20:42 Q&A
Q: I came a long way and happy that I did. Its like 600 years ago,
going to see a lecture by Gutenburg, and I say that in all
seriousness. You mentioned that the royalty free decision was made at
an early stage. Given the value of what you created, do you ever
regret not establishing an intellectual property right, to control
those who abuse the web? I'm thinking especially of the Chinese
government.
A: No. (Its nice to answer with a yes or no at these things, isn't it?
haha) No, I never had the slightest regret about my decision. If there
was a hint of patents, it just wouldn't work. Like the gopher system,
which the University of Wisconsin hinted, not for academics, not for
current uses, not now, just maybe, perhaps in the future, about
charging. As soon as there was a hint there, everyone dropped it very
fast. It was critical that I not do anything like that. A web based in
competition would not work at all, people in their garages would never
put any effort in like they did, if they felt they were working for
someone else. So yes, we could patent our semantic web technology, but
its a lot of trouble, I don't think it would work anyway. And with the
Chinese government, they are not going to take any notice of our own
intellectual property ideas, they are a sovereign nation. Someone said
that the Internet treats censorship as a fault and routes around it,
and I think that applies to humanity as a whole. Its unwise for a
country to change too fast, and I think change will change steadily.
Q: Andrew Hardy, director of democratic society. I have a long and
sustained interest in the use of web technology in small democratic
forms. Can you explain more about the new web developments aid
democratic developments?
A: eBay has buyers and sellers rating each other, so people in the
community become respected. So meritocratic systems happen, and they
aren't exactly democratic, but they involve the wisdom of the masses.
Early on, I had a student in the first year of the web who made 'web
interactive talk' - a form of discussion, that was different to a blog
comment in that you said if you were supporting or opposing the main
post. So you can colour the posts, and make what was a riotous flame
war into a sort of order and allow the real issues to come out. This
never caught on, though. So one dream is that when politicians say
things, its on the web, and their aids can post the supporting
documentation that proves what they said. So someone who disagrees can
see this, and discuss it properly. So the truth will out, and people
will apologies if they use statistics inappropriately. Can we make it
happen? We are trying. I encourage us all to play with experiments in
ways of doing things differently.
Q Karel Dudesek: I'd like to know your thoughts about the futuristic
scenario of people and the machine. The machine gets more intelligent,
learning, the software is learning. So will a human get obsolete by
the machine?
A: Ray Kurzweil wrote good books on this. What happens when the
transistors on a chip outnumber the neurons in a human brain? So I
hope we can build things that think, although we too easily
underestimate whats involved. And I don't think they will revolt when
things reach a critical mass. I don't think the machines will take
over, but we will get a situation where we as humans can do crazy
things. So our tendencies to polarise into cults, something happening
throughout the ages that's destroyed previous great civilisations,
becomes important. As technology gets faster and faster, we don't get
less and less prone to that stuff. So a responsibility of web science
is to think of the web as connected humanity, and care for it, and
make sure people don't get sick.
Q: Someone Norman. I first cut code 49 years ago. 40 years ago I
joined the BCS. You put up a slide, "Semantic Web: Technical," and
Trust was at the top. And I waited for a slide "Semantic Web: Social,"
and to see at the top, Trust. Where will the social trust come from,
especially since many of the people who want to control it, we don't
trust.
A: The systems that enable social trust are many and varied. We make
laws to define social systems. I was before a US congress subcommittee
the other day, and it was interesting to see the people who define
social protocols. This is an old and established way of doing of
things. The web won't change that immediately. There are first order
and second order interactions between the web and the law. Its
important that we are able to get lawmakers to really understand the
web though.
21:00
Vote of Thanks. From Professor Brian Collins, from Department of Transport.
Its rare to hear a lecture from a creator of something that's had
global impact, like printing, or the internal combustion engine. Its
rarer still to hear a discussion of the future of that thing within 2
decades. We've had both today.
The words I would use to describe this evening are, visionary,
pervasive, and exciting. This lecture is being webcast. Thanks again!
