MIT Technology Review, July 1996
The Web Maestro: An interview with Tim Berners-Lee
[Image] [Surf less. Know More. LeadStory.com] [AdFinder]
[Image] -----------------------------------------------------
[Image]
The man who wove
the first few strands in what has
grown into the World Wide Web - and who
ÊÊE.W. CD-ROM oversees the organization that coordinates
ÊÊReviews the Web's further development -
peers into the future he's helping to
ÊÊFortune Infotech create.
ÊÊSpecial
-------------------------------------------
ÊÊHome PC [Image]
The World Wide Web is giving Tim
ÊÊInformation Berners-Lee a problem. He is eager to
ÊÊWeek demonstrate a new feature he is working on,
which will make it much easier for users of
ÊÊInteractive Age the World Wide Web to forge connections
between documents, or "pages." But
ÊÊMIT Technology Berners-Lee's mouse clicks seem to fall on
ÊÊReview deaf silicon; supposed links appear not to
exist, and when connections do get made the
ÊÊNetGuide text and images flow onto the screen at
less than eye-popping speed. Since everyone
ÊÊQuain's Web who has used the Web has experienced this
ÊÊReview dozens of times, it is perversely
gratifying to see Berners-Lee suffer the
ÊÊTechdaily same frustration. Wouldn't we all love to
see the inventor of the VCR get hung up on
ÊÊTidBITS programming it to record a TV show?
"Usually, this is blindingly fast," he
ÊÊTIME Digital insists, and with persistence he prevails.
But the man who invented the World Wide Web
ÊÊTIME Magazine is, at least for the moment, trapped in his
own glorious creation.
ÊÊWindows
ÊÊMagazine The Web's key feature is information
connected through hypertext
"links"--clicking on a word or a picture
summons into the user's computer text,
pictures, sounds, software, or any of a
thousand and one gimmicks. So powerful is
the appeal of the graphics and hyperlinks
that newcomers to the online world might be
forgiven for thinking that the Web is the
Internet, rather than just an especially
powerful and convenient way to navigate
through the worldwide collection of
networked computers. Berners-Lee did not
set out to invent a contemporary cultural
phenomenon; rather, he says, "it was
something I needed in my work." He wanted
simply to solve a problem that was
hindering his efforts as a consulting
software engineer at CERN, the European
particle-physics laboratory in Geneva.
Mainly to become more efficient, he
developed a system that provided
easy-to-follow links between documents
stored on a number of different computer
systems at this international laboratory
and created by different groups.
Hypertext had been proposed as early as
1945 by Vannevar Bush, and rudimentary
hypertext software had been developed to
interlink material among different files on
individual PCs. Berners-Lee's innovation
was to apply the idea of hypertext to the
growing reality of networked computers. His
timing was just right. In the late 1980s
and early 1990s, the Internet was just
starting to blossom and achieve recognition
beyond a small cadre of military and
research institutions that had formed its
early clientele. As the number of
interconnected computers grew from dozens
into the tens of thousands, the Web offered
an ideal way to tap into the information
scattered among these machines. Berners-Lee
expanded the system he had devised at CERN
and made it available on the Internet in
the summer of 1991.
Unlike other computer-industry figures who
have become household names, he has stayed
in the shadows. Rather than spin off a
company to cash in on his ideas, the
British-born Berners-Lee became in 1994 the
first director of the World Wide Web
Consortium, a nonprofit organization with
more than 100 member organizations that
coordinates the development of Web software
and standards. When looking for a place to
locate the consortium, Berners-Lee chose
MIT's Laboratory for Computer Science. "MIT
has a reputation for doing the right thing
on technical standards," Berners-Lee
explains. Typically, he says, MIT will hold
the copyright on a standard but keep it for
the public good; "this isn't a place that
will turn around and slap on license
charges when you're five years down the
road."
Berners-Lee is not, it turns out, a typical
user of the World Wide Web. His use of the
Web is almost exclusively related to his
work on devising standards for it. Asked if
he surfs the Net for pleasure, he replies
that he doesn't have time for that kind of
thing; Henry Ford is too busy in the garage
to go out for a Sunday drive. In any case,
Berners-Lee tends to dismiss complaints
that the Web is too hard to search and that
the gems are hopelessly submerged in
gigabytes of drivel. "There's no
fundamental right for people to be able to
discover anything instantly," he maintains.
The glitch on the Web soon surrenders to
Berners-Lee's careful, persistent, and
knowledgeable ministrations, and the
illusion of the machine taking revenge on
its inventor recedes. Close call. With
electronic tools now acquiescent, a
confident Berners-Lee spoke with Technology
Review senior editor Herb Brody about how
he devised the Web, why its critics are off
base, and how he envisions it will change
and improve in the years ahead.
TR: Do you ever step back and marvel at how
rapidly your idea has taken root? The World
Wide Web has to have set some kind of
record in the speed with which it
progressed from unknown and esoteric to
fashionable and then to almost commonplace.
BERNERS-LEE: Yes, the Web's growth has been
exponential. For the first three years, the
load on the servers was always 10 times
what it was the year before. But after a
few years of that kind of growth, you get
used to it.
TR: Hypertext had been proposed many times
before, and implemented on a small scale.
Why do you think the concept caught fire
with the World Wide Web?
BERNERS-LEE: Earlier hypertext systems had
generally been limited to pointing to
documents within the same local file
system. Those systems often used a central
link database to keep track of all the
links. The advantage of this kind of
approach was that it ensured that a link
would never point to someplace that didn't
exist.
TR: And the disadvantage?
BERNERS-LEE: There was no way to scale up
such a system to allow outsiders to easily
contribute information to it.
TR: So the original concept of the Web
involved a trade-off favoring universality
over reliability.
BERNERS-LEE: Yes, I sacrificed that
consistency requirement to allow the Web to
work globally. What was really new with the
Web was the idea that you could code all
the information needed to find any document
on the network into a short string of
characters. These strings, originally
called universal document identifiers, are
now known as universal resource locators,
or URLs. The notion that all these tagged
documents from computers all over the world
could share a common naming and addressing
"space" was what made hypertext links so
much more powerful.
TR: What was your goal in designing the
World Wide Web?
BERNERS-LEE: It was something I needed in
my work. CERN is composed of a variety of
bright and creative people from institutes
in many countries. When they work together
on a project, the result can be a tangle of
complexity. Coming into this organization
as a software consultant, I found a
tremendous need to be able to find out what
was going on, particularly the
interdependencies--what work was related to
what. If I needed to modify some program
module, for instance, what else was that
change going to affect? I wrote a program
called Enquire, which had a little bit of
what we think of now as hypertext: at the
bottom of each document would be a list of
references that you could follow to
immediately jump to another piece of
information. I found this really useful
because it was so flexible--I used it to
keep track of everything I did. TR: But
this wasn't the Web as we now know it,
right? Weren't you interlinking things like
computer programs and their documentation?
BERNERS-LEE: Yes, mostly, but it wasn't
limited to that. You could put recipes in,
if you wanted, and link them back to your
ingredients--so that you could follow the
link from onion pies to onions, or whatever
you liked.
TR: Did Enquire solve your problem?
BERNERS-LEE: Not entirely. I had details of
my own work nicely organized in this
web-like fashion, but what I really needed
was to make links to other people's
documents. We needed a program that was so
easy to use that everybody would end up
putting their data into it. That way when
you wanted to collaborate with other people
you could easily share data, you could
point to things that they had written
before, rather than having to copy them. It
was also crucial to allow different people
to be able to start their own webs in
different places and later link them with
only incremental effort.
TR: How was this an advantage over what was
available to you at the time?
BERNERS-LEE: In a typical documentation
system, if you wanted to make a reference
from one document to another, you had to
merge the two computer databases that held
the information. That entailed moving all
the stuff onto the same computer and
arguing about who would keep it maintained.
That wasn't going to work.
TR: So the Web was born not as a "world
wide" system but as an internal computer
network--a closed universe.
BERNERS-LEE: The network was used mainly by
people working for CERN, but "closed" may
not be the best word. The people at CERN
did come from all over the world. And I was
working before, during, and after on other
projects with people from the Stanford
Linear Accelerator and FermiLab in the
United States and from Britain's Rutherford
Lab, to name just a few.
TR: Still, it seems like a pretty big leap
from a network for nuclear physicists to
the cultural phenomenon that the Web has
become. How did it get from there to here?
BERNERS-LEE: The first few years involved a
lot of persuasion--we had to convince
people to use the Web and to put
information up on it. But what really made
it go was the set of specifications we had
developed early on.
TR: You're referring to the alphabet soup
of Web standards?
BERNERS-LEE: Yes--there was hypertext
markup language (HTML) for creating the
documents with hypertext links, and
hypertext transfer protocol (HTTP) for
specifying how the network would respond
when a user clicked on a link. And above it
all was the system of URLs, which ensured
that every item put up on the Web had a
unique "address." For the Web to function
and to grow, everyone had to stick to these
specifications, and to agree on any changes
to them.
TR: How has the Web departed from your
early vision of it?
BERNERS-LEE: The original idea was that
anybody would very easily be able to write
documents that could be connected through
hypertext links. What has surprised me is
the way people have been prepared to put up
with manually encoding text. HTML was never
supposed to be something that you would
see--it was intended to be something
produced by an editor program. An analogy
is with word processors. Computer users
don't have to write in all kinds of codes
to format their document with fonts,
margins, and so on. So it staggers me that
people have actually put up with having to
write HTML by hand. Similarly, I had not
expected people to have to work out the
hypertext links by looking up and typing in
those long, complex codes for addressing.
URL syntax was never intended for human
consumption. It was intended for a machine.
TR: But ordinary users of the Web don't
need to know HTML--that's only for the
people who create content.
BERNERS-LEE: Yes, but the Web needs
information providers as well as readers.
And the fact that creating Web pages has
been difficult has directly influenced the
type of information made available on it;
content is produced only by those with
enough incentive to learn to write HTML.
TR: How had you envisioned it working?
BERNERS-LEE: In the prototype, you could
create a link without having to write any
code. You'd just browse around, find
something interesting, go back to the thing
you were writing, and then just make a
click on a hot key, and it would make a
link for you automatically. This ability is
now starting to become available--in a
couple of years, all the documents on the
Web will probably be created without the
direct use of HTML and URL syntax that is
now so much a part of the Web.
No Instant Gratification
TR: The Web has a reputation in some
quarters as more sizzle than steak--you
hear people complain that there's no way of
judging the authenticity or reliability of
the information they find there. What would
you do about this?
BERNERS-LEE: People will have to learn who
they can trust on the Web. One way to do
this is to put what I call an "Oh, yeah?"
button on the browser. Say you're going
into uncharted territory on the Web and you
find some piece of information that is
critical to the decision you're going to
make, but you're not confident that the
source of the information is who it is
claimed to be. You should be able to click
on "Oh, yeah?" and the browser program
would tell the server computer to get some
authentication--by comparing encrypted
digital signatures, for example--that the
document was in fact generated by its
claimed author. The server could then
present you with an argument as to why you
might believe this document or why you
might not.
TR: This would be particularly useful, I'd
think, in verifying orders or payments for
electronic commerce.
BERNERS-LEE: Yes--it would help if, for
example, you find a beautiful offer on the
Web for some product and you want to find
out if it's for real. But this kind of
verification is important for more than
just buying and selling things. Every
political candidate, for instance, seems to
have two or three "spoof" Web sites--they
look almost, but not quite, like the real
thing. When you visit the real White House
Web page, for example, you can click on an
icon of a cat and hear a meow. Then one day
you click on a White House link from
somebody's page and you click on the cat
and you hear some awful noise
instead--you've been spoofed. You're not
really at the White House--you're at
something like white-house.com instead of
the real thing, which is whitehouse.gov. So
you ought to be able to press "Oh, yeah?"
and the browser sends out a request to
cryptographically check the authenticity of
the site.
TR: Another common gripe is that the Web is
drowning in banal and useless material.
After awhile, some people get fed up and
stop bothering with it.
BERNERS-LEE: To people who complain that
they have been reading junk, I suggest they
think about how they got there. A link
implies things about quality. A link from a
quality source will generally be only to
other quality documents. A link to a
low-quality document reduces the effective
quality of the source document. The lesson
for people who create Web documents is that
the links are just as important as the
other content because that is how you give
quality to the people who read your
article. That's how paper publications
establish their credibility--they get their
information from credible sources. A
journal on the Web, for instance, needs to
have an editor who is paid to make sure
that the pointers lead to good stuff. You
don't go down the street, after all,
picking up every piece of paper blowing in
the breeze. If you find that a search
engine gives you garbage, don't use it. If
you don't like your local paper, don't buy
it. If you find that an article refers to
stupid articles, don't read it, and don't
quote it yourself. Pretty soon you'll have
some bookmarks on places you trust, and
your reading quality will increase. You may
find that the better sources have involved
considerable human effort, and so there
will be either advertising to read, a
subscription to pay, or a volunteer to
thank. Or did you want quality for nothing?
TR: While there are a number of tools to
help users find information on the Web,
they aren't terribly precise. Wouldn't it
be better if the creators of Web pages
labeled them with keywords that could be
searched for?
BERNERS-LEE: Yes, that would make searches
more productive, but there are two big
snags. The keywords, or the topic names,
form a rather centralized rigid point in
the system, which can never change fast
enough to keep up with the Web. And the
classification of material, like authorship
of material itself, can be subjective and
controversial, and quickly become obsolete.
The job of classifying all human output is
a never-ending one, and merges with the job
of creating it.
TR: But Web users nevertheless crave better
ways of searching through the sea of
information. The searching tools available
now give you a list of Web pages very
quickly, but many of these "hits" seem to
have little if any relevance to what the
person is looking for.
BERNERS-LEE: People have no fundamental
right to discover everything instantly.
Information providers want to be found
easily and will use standard ways of
registering themselves and their products,
and there will be tools to scan the Web so
that users may quickly find them. But other
providers will prefer to express themselves
any way they want and thus might be more
difficult to track down. They might not be
cataloged, and they won't mind. So just as
you have a right to scribble anything on
paper and not show it to anybody, you'll
also have the right to post things on the
Web and not make them easily found. The
Web, like paper, should be a universal
medium, in which it is possible for all
kinds of information to exist.
TR: People are developing high
expectations, though: they want to know
why, if we have all this information
online, it is so hard to find what we want.
BERNERS-LEE: This question goes back to the
early days, when I was really pushing this
idea of a World Wide Web uphill. My
response is: If there's a person in the
world who is apt to have the information
that you're after, go and persuade him or
her to put it on the Web for you and others
to find it. For instance, you can't as an
academic just sit back and say, "Why can't
I see a list of all the journals in my
field?" Go write that list! Somebody has to
put in the effort; the existence of a
network doesn't give you a well-sorted
catalog of all the information on all the
machines on that network for free. The
existence of paper doesn't give you a
library. You need a whole lot of librarians
working hard, plus a whole lot of journal
editors working hard, plus a whole lot of
academics working hard at writing and
supplying references. Together all these
people produce a library full of good
journals.
Looking Ahead
TR: Although begun as an academic project,
the Web stands to make a few companies a
lot of money. Do you see any dangers posed
by this commercialization?
BERNERS-LEE: Commercialization of the Web
is giving it a lot of momentum, helping it
expand, and bringing it a lot of new ideas.
It is true that some people feel that there
is a threat that a particular company will
try to take over the control of the
standard protocols that govern the Web's
operations.
TR: How might that happen?
BERNERS-LEE: A company could start by
releasing Web browser software and make it
available free or at very low cost, to
capture the vast majority of the market.
Later, that company decides to introduce a
feature in this product that can be taken
advantage of only if the designer of the
Web page deviates from accepted Web
standards in some fashion. A Web user would
then suddenly begin encountering pages that
read, "Sorry, you need software from
Company X to enter this site." Anyone who
slaps a "this page is best viewed with
Browser X" label on a Web page appears to
be yearning for the bad old days, before
the Web, when you had very little chance of
reading a document written on another
computer, another word processor, or
another network. And once a browser vendor
has established such a monopoly, it has an
incentive to continue to make arbitrary
changes to the de facto standard, forcing
potential competitors to play an endless
game of catch-up. All the other bright
ideas at all the other software companies
are stifled because they have to be
compatible with a "standard" that changes
at one company's whim.
TR: What forces might tend to prevent this?
BERNERS-LEE: If the competing companies
band together and move in a different
direction, then the monopoly company could
lose out badly for having introduced an
incompatibility. The World Wide Web
Consortium helps people to agree on
standards. We also sometimes write and
disseminate the programming code to give
people an idea of how to put these
standards in place. We recently did this,
for example, with style sheets, which are a
nice clean way to give Web pages consistent
and distinct layouts, fonts, and so on
without having to insert all the formatting
codes each time. Another good example is
Java--a programming language used, for
example, to create small applications
programs, or "applets," which can be put
into a Web page. When Java first came out,
three companies--Sun, Microsoft, and
Spyglass--introduced three different
applets for inserting an animation or video
file into a Web page. Since none of the
companies wants to be called the
incompatible one, each has to make sure it
supports the other two. This is an awful
lot of effort, which would be better spent
improving the products. The consortium got
everybody around a table, and we now have a
draft of a standard that is a compatible,
consistent way of doing this operation.
TR: In what ways do you think the Web is
being underutilized?
BERNERS-LEE: A lot more people can browse
the Web than can put up their own Web
pages. The Web is therefore not being used
so much the way I originally conceived
it--as a communications tool that would
enable small groups to work more
efficiently in teams.
TR: How do you envision that kind of use?
BERNERS-LEE: Say that you conduct a meeting
as a hypertext document. You start by
dragging in a video version of yourself,
with real-time sound. You remind those
invited to come by sending them a hypertext
e-mail with a pointer to the meeting. To
join, they just follow the link. They can
not only read this meeting/document, but
they also write to it. Some join by audio
and some drag their own video into the
document. People introduce points by
writing them into the minutes, making links
to background material. At one point in the
meeting three people realize they need to
discuss something separately, and with a
single keystroke one forks off a new
meeting document that they will catch up
with later. There is no rocket science
here, but an integration of group editing,
hypertext editing, and real-time
audio-video technologies. These
technologies all exist in crude forms, but
must mature and be standardized before
global hypertext teams can feel comfortable
using them.
TR: So the Web could be used for a kind of
videoconferencing?
BERNERS-LEE: Yes--participants would have
video cameras connected to their computers,
and they would all see on their screens a
picture of the meeting. This kind of
videoconference is possible right now, but
not everyone has a fast enough Internet
connection--that is, enough bandwidth--to
transmit all the data needed for
full-motion video. One option is to
represent people who don't have enough
bandwidth with flat, cutout shapes, which
could change when the person is talking or
indicates a desire to talk. Other people
will be present as a real-time video image.
All of these will be put into the virtual
space so that they all seem to be part of a
room.
TR: Wouldn't this require a leap forward in
graphics? Images on the Web now are pretty
two-dimensional.
BERNERS-LEE: Yes, but I expect
three-dimensional rendering and graphics to
become common. I mean, look at that screen
over there. You call that a "desktop"?
Maybe that's how a real desktop looks from
a camera flying at 10,000 feet.
TR: It's stylized, but it seems to work.
What's the benefit of 3-D, other than
razzle-dazzle?
BERNERS-LEE: It provides a better model of
the real world. In the physical world,
people's documents overlap each other and
stack up in piles. Imagine if you could
build "shelves" on your screen and you
could fly through them and find something
that you put somewhere. Maybe this way of
presenting information will click with how
people actually store and retrieve things.
TR: What other refinements in the Web are
you most eager for?
BERNERS-LEE: I hope that the notion of
having a separate piece of software called
a "browser" will disappear. A browser is
something that (a) only allows you to read
and not write, and (b) is a single window
on the world. Instead, your entire screen
should be a window on the information
world, with a small part of it representing
what's on your local "desktop." Browser and
operating-system interfaces will become so
interlinked that they will, for all
practical purposes, become one. Whether the
operating system swallows the browser or
the browser swallows the operating system,
there will be one interface. As with the
television and the home computer, the
question of which will "win" is really a
question about which companies will come
out on top; the resultant object in any
case will be both.
TR: What will using the Web be like in a
few years, assuming these developments
occur?
BERNERS-LEE: You won't see a browser, you
will see a document. You'll follow some
links and find other documents and these
documents will leave a trail of documents
across your desk. And then you might find
that one of them takes you to a store, and
in the store you find a shopping cart that
you can move around and put into it things
that you want to buy. And then at the end
of the day you can buy what's in the cart.
The code that makes this cart do what it
does won't be anything you've bought, but
when you first click on the cart icon that
software will be automatically transferred
through the Net to your computer.
TR: So software would be acquired on a
need-to-use basis?
BERNERS-LEE: Yes, as you wander around the
Web, your computer will become encrusted
with pieces of software necessary to allow
you to interact with and represent to you
the things that you're reading about. If
you happen to be an astronomer and you've
been looking at spectra, then
spectrum-analyzer software will allow you
to manipulate them. If you're a biology
student and you download some images of DNA
molecules, then the code to send this DNA
will come with a little bit of software
that allows you to spin it around and break
it up. And so your computer's software
ability will not depend on where you've
been shopping but just where you've been
reading --where you've been browsing on the
Web. The very idea of software will become
a bit more submerged. It will be seen less
as a discrete entity that you go out and
buy and more as a support to the objects
that are part of the information space. The
software will move on and off your machine
without your having to worry about it.
TR: What you're talking about sounds like a
world in which far more people write
software than do now.
BERNERS-LEE: Yes, but they won't think of
it as creating a program. They will just be
creating documents, but the software needed
to view and manipulate these documents will
be part of it. Tables of data will have
spreadsheet software built into them, for
example, but the person writing the table
certainly won't have to write a spreadsheet
program. Java is a step in this direction.
But an incredible amount of work needs to
be done to achieve the user interface that
I have rather glibly described. We also
have to establish a level of trust that
makes it possible for information to move
from the Net onto your computer and to do
work, possibly including writing files onto
your hard drive. You want to make sure that
it's not possible for a malicious person to
be able to send you something that will
look at your personal files and override
them, or broadcast their contents.
TR: Are there any other items on your World
Wide Web wish list?
BERNERS-LEE: I want better international
access, especially in developing countries.
And I'd like to see a more organized market
of Web server space, so that everybody with
an Internet connection could put
information out cheaply. I expect that
computers able to use the Web will become
fairly ubiquitous, about as pervasive as
televisions are now. In fact, the last
computer I bought can play video--when you
have a computer and good Web access, who
even needs a television? I don't think
everybody will want to post information on
what amounts to a global bulletin board,
but I certainly hope that every business
has a Web page. I would also like to see
deregulation of telecommunications globally
so that Internet access to the home becomes
cheaper. The United States is better than
Europe in this regard, but even here Net
access is not as cheap as it could be.
TR: Why do you think the Web has resonated
so strongly with today's culture?
BERNERS-LEE: The openness of the Web is a
powerful attraction. Everyone can not only
read what's on the Web but contribute to
it, and everybody is in a sense equal.
There's a sense of boundless opportunity.
[TECHNOLOGY REVIEW: MIT's National Magazine of Technology & Policy]