I have not said much about this. Maybe it's appropriate now. For
this, I need to relate my view of the critical path to the current
state of the industry. Your mileage may vary.
I have tried to get this put into wikipedia but it was removed on a
few occasions. The rational was that it could not have occurred
because it is not on the web. Later, I appealed to Jimbo, sending
him 25 pages or documentation and early history on Silversmith, and
he added a brief account into the history of the web browser and
locked the page.
(unknown CALS browser)
probably a beltway vendor, 1988
4. Viola browser, 1993
Silversmith was the first modern browser based on tags. It was
created as an a
ccess method for ontologies and large
documents. Contrary to popular opinion, TimBL's browser was not the
first. As indicated below, he has noted that it was the first
"graphical web browser". I had been inspired by "Literary Machines"
by Ted Nelson and his Xanadu project. I later asked him when was the
first time he saw the term "web" applied to documents and he
indicated it was earlier than his work, perhaps from Nolan Bushnell.
At the time we were doing standards work for CD-ROM's, and were the
first company to put images on a CD. I was chairman of the Rockridge
committee, and I also sat on the ANSI CD-ROM committee. We had a
booth at a Microsoft conference when Dr. James Mason made an
announcement about the preliminary approval of the ISO standard for
SGML developed by Charles Goldfarb. (Since CD's are silver, the tool
for them became "Silversmith".)
That was in early 1986, and I believe (the late) Bill Tunnicliffe
was chairman of that standards work. Bill was the person who
conceived of the separation of data and control for documents. That
is now an idea that is undergoing acceptance in other areas as
developers prepare to move to smarter systems. Bill is one of the
unsung heroes of present systems in that his insight now provides a
key ingredient in computer systems on the order of $300B - $400B per
year, in addition to the commerce they enable.
James made the announcement at a CD-ROM conference at Microsoft Hq
in Washington. As I listened to him make the announcement I couldn't
understand why it was needed. Slowly, as he spoke it occurred to me
what he was describing and my head went into overdrive. It was the
key that I had been looking for for some time. I was interested in
rapid access for large documents, and this was the missing part. I
didn't know if I should shout or keep my mouth shut. (I decided to
act quietly and when I returned to Boston I started reading the
After James made the announcement we talked, and he mentioned that
an AAP committee was developing the first tag set for documents
using SGML, and they needed assistance. That work was done by the
Electronic Manuscript Project committee of the AAP. The
committee, an ad hoc project, aimed to
develop publishing standards so that publishers could have access to
printing sources that produced identical content layout. The
committee developed tag sets for book, article and serial
(magazine). It also developed layouts for tables and math equations
in addition to what is now CSS (Cascading Style Sheets).
Silversmith is a text-based, full-featured browser that could link
between images, sound and text. This was in the days of the PC-AT
and a robust page composition system was not yet available. Still,
Silversmith had a number of interesting features. It had a
replaceable font system, and we delivered a special version for Dr.
Mason that performed searches using runes for Old Norse literature.
Bill lived not far from me, and he came to my lab on a couple of
occasions. I demonstrated Silversmith for him and his response was,
"Yes, that is interesting. But I don't understand why you would want
to do that".
Silversmith could do regular _expression_ searches on text. It could
search on text or tags, including semantic tags within a restrictive
context. By clicking on an image or a highlighted (blue) text you
would like to the destination. We had to add a "Back" button to
return to the earlier linked text or entity. It does linear searches
on documents that have not been indexed. It also permits a four (4)
level library system that allows searches of documents across
document, shelf, stack and department. Imagine today if you had all
web documents in a Dewey or MARC system that could be searched by
topic or title.
Semantic tags in Silversmith allowed more sophisticated searches.
One system that was delivered allowed users to search for tools
identified by tags. Faceted searches could also be done. The system
was delivered with an indexing tool and an automatic tagging
application. It was introduced at a TechDoc conference in Austria in
1986 and was demonstrated at about a dozen conferences in Europe,
Canada and the U.S. In 1988, Dr. Ken Baklawski and I sponsored the
first conference on SGML applications and Yuri Rabinski discussed
his SGML editor and I discussed the Silversmith text retrieval
system. There are other advanced search features in Silversmith that
have not yet been released.
One delivered system allowed searches such as:
"find all 'top-secret' paragraphs in the section 'Navy
I met with Charles at the first conference where Silversmith was
demonstrated, and he looked at it with interest. I was particularly
interested in finding out whether there was a similar product at IBM
or that he had knowledge of. He was not aware of such a tool so I
assumed for the time that the introduction would preclude anyone
from a subsequent patent or from stopping our development. His
comment was that this would be of interesting for certain court
cases. Later, Charles told me that he had served as an expert
witness on a number of patent cases.
Because I sat on the committee that developed the tags I had been
keeping an eye out for anyone beyond our group that were early
adopters. In fact, the committee work was restricted to the AAP
participants who were, for the most part, the large book publishers
in Manhattan and Boston.
Silversmith was successful in that it repaid its initial investment,
but we could not demonstrate to a VC a significant market potential.
One VC told me, "If the market needed one, they would already have
I believe that later systems experienced the same kinds of response.
Ultimately, we found that we were spending a lot of time in
education mode, marketing and creating new web pages. All this in
addition to developing new versions of the browser. We also found it
very difficult to get into conferences to discuss Silversmith
because we were not one of the "Big 3" companies so our development
could not be that significant. Other projects offered more revenue,
so Silversmith was eventually put on the back burner. The result was
a delay of about 9 years in having access to a browser, and the one
that was eventually successful was "of European origin", even though
the technology was developed in the U.S.
There were several myths that were brought up when we discussed
Silversmith in public. Some brought on head-shaking disagreements,
and in one case the Apple Evangelists tried to shout me off the dais
at a conference. In general, the myths followed one of these lines.
a. "All text retrievals should use SQL. You need to put each
paragraph in a table row, and then have SQL search for the desired
(That would have been very slow for large documents and did not
provide a migration to semantic information. In fact, Silversmith
was built around a B-tree which gave us a response time of about 1/3
second on the largest documents we could find, about 500 Megabytes
(a CD-ROM's worth).
b. "The AAP tags were not intended for use on computers, they are
for text layout only so you should wait until there is a more
appropriate SGML application."
(In fact we did have to alter the "image" tag to allow for image
type, size and placement.)
c. "SGML can not be used for hypertext links because they were
(I argued that with our B-tree structure and the "Back" button, that
navigation was not a problem and demonstrated it.)
2. ERWISE, 1992
"The first graphical Web browser was named Erwise, an application
written for Unix computers running the Windows X System. It was
developed by Kim Nyberg, Kari Syd„nmaanlakka, Teemu Rantanen, and
Kati Borgers, four college students who attended the Helsinki
University of Technology in Finland. The project began in 1991 and
was released in April of 1992. Erwise is largely credited as a
pioneering program and is the precursor to the modern Internet
"Tim Berners-Lee, the creator of the World Wide Web, was
instrumental in the development of the first graphical web browser.
Berners-Lee realized that a browser with a _graphical_ user
interface would make the Web significantly easier to use for
Internet surfing. Believing that the concept would be an ideal
project for students, he made numerous requests to prospective
developers around the world."
"Berners-Lee's request was picked up Ari Lemmke, the instructor who
supervised the four students that eventually created the browser.
The four Finnish students developed a graphical, point-and-click
browser with numerous features that would not be seen until much
later. Their work was inspired by Robert Cailliau, the Belgian
computer scientist best known for working with Berners-Lee to
develop the World Wide Web."
"As the first graphical Web browser, Erwise had many unique
qualities. Among them was its ability to simultaneously load
multiple pages. For example, users could click on a hyperlink that
automatically opened another window featuring the linked Web page.
In fact, Erwise had a number of capabilities Mosaic lacked,
including integrated text searching that could be performed from any
given Web page. Mosaic, the first graphical Web browser that helped
popularize the Web, still is widely recognized for laying the
groundwork modern browsers would follow."
"The development of the first graphical Web browser came to a halt
after its creators graduated and moved on to other endeavors.
Berners-Lee took a trip to Finland in an attempt to encourage the
group of students to continue their work on the project.
Unfortunately for Erwise, the group could not proceed with
development due to inadequate funding. It has been said that
Berners-Lee would have liked to continue the project himself but
could not do so because the code was written entirely in Finnish."
3. Unknown CALS browser, 1988
Lost somewhere in my archives is a document outlining the RFP
process for a Navy CALS project. We were able to demonstrate
Silversmith on several CALS technical documents, yet lost the
proposal to a beltway vendor. At this point it became apparent that
even with a conforming CALS working browser, we were not able to win
Federal RFP competitions against vendors with no extant tools.
4. VIOLA Browser, 1993
...Jennifer Doan, a Texarkana lawyer representing Yahoo and Amazon,
led the questioning.
“Mr. Berners-Lee, why are you here?” asked Doan.
“I am here because I want to help get some clarity over what was
obvious, and what was the feeling of computing [in the early
1990s]…. The tools I had in my knapsack, so to speak,” he said.
After describing how Berners-Lee worked at CERN in Switzerland
back in the 1980s, Doan moved on to the web. When Berners-Lee
invented the web, did he apply for a patent on it, Doan asked.
“No,” said Berners-Lee.
“Why not?” asked Doan.
“The internet was already around. I was taking hypertext, and it
was around a long time too. I was taking stuff we knew how to do….
All I was doing was putting together bits that had been around for
years in a particular combination to meet the needs that I have.”
Doan: “And who owns the web?”
Berners-Lee: “We do.”
Doan: “The web we all own, is it ‘interactive’?”
“It is pretty interactive, yeah,” said Berners-Lee, smiling.
Then Doan moved on to the heart of Berners-Lee’s testimony: to
establish the importance of the Viola browser, created by Pei Wei,
who was at that time a computer science student at UC Berkeley.
The Viola browser is a key piece of “prior art” that the defendant
companies hope will invalidate the UC/Eolas patent.
Berners-Lee described Viola as “an important part of the
development of the web.”
The jury was shown an e-mail from Pei Wei to Berners-Lee dated
December 1991 — almost two years before Doyle’s invention — which
read in part: “One thing I’d like to do soon, if I have time, is
to teach the parser about Viola object descriptions and basically
embed Viola objects (GUIs and programmability) into HTML files.”
Later Tuesday, Wei would testify that he had demonstrated
interactive elements working in the Viola browser to Sun
Microsystems in May 1993 — several months before Doyle claims to
have come up with his invention.
Finally, Doan turned to Berners-Lee’s book, in which he described
Wei as “a very inventive student at UC Berkeley.” Berners-Lee
described how the web community at that time wasn’t focused on
patents or even money — Wei simply put his invention online for
“It was ahead of its time,” said Berners-Lee. “The things Pei was
doing would later be done in Java.” Java was heavily used in the
late 1990s to add interactivity to web pages, later to be
His own act, creating the World Wide Web, was more a matter of
personalities and persuasion than it was a matter of hammering out
code, he explained.
It is my understanding that TBL also used the AAP tag set in the
creation of XML. I was dismayed to find that the implementation
was incomplete, and had stripped much of the power from the
"After further study of these documents and others in the same
subdirectory, it became apparent that most of the early HTML tags
were actually taken from the CERN SGMLGuid language, which itself
was a variant of AAP (an early SGML language). For example,
hn, p, ol and so on are all apparently taken from this
language. The only radical change was the addition of the all
important anchor (<a>) link, without which the WWW wouldn't
have taken off."
So the W3C graphical (XML) browser followed the same development
path as did Silversmith in 1986.
Tim later said he decided that he would not patent his browser. I'm
not sure to which part he was referring.
- - - - - - - -
About the book, "The Next Computer Revolution"
This is a popular
book for managers, executives and
developers interested in semantic systems. This title is a working
title, and will be changed to sometime a little more palatable to
the target audience. It outlines the nuts and bolts of a system
along with a smattering of jargon needed to bootstrap the reader
into more technical works. The premise of the book is that we need
semantic systems to move forward and this is one approach to
implementing such a system.
The basic goal is to give a background in implementing semantic
systems along with a suggested architecture for a general purpose,
semantic system. There is also a healthy dose of discussion of the
myths that hinder and restrain new technology. I mention at least a
dozen that mislead discussions into non-productive approaches to
Concord, MA USA
On 4/8/2012 5:40 PM, Rich Cooper wrote:
die poor are more the rule
than the exception. I am told that the likelihood
of an arbitrary patent paying
its owner more than the owner paid to prosecute
the patent is about one
percent. Today, if you use an attorney or agent,
it costs about ten grand to
get an average patent through the USPTO. About
one grand of USPTO fees, and
the rest for the attorney/agent.
I did the
prosecution myself (you can do
that too) and it cost me about one grand for
prosecution, and another two grand
to pay maintenance fees late. So you can get a
better deal, but you have to
understand patent litigation to do it right,