[Wired] Helping the Web Grow Up.

I Find Karma (adam@cs.caltech.edu)
Fri, 12 Jun 1998 03:11:43 -0700

Remind me never to talk to the press. For example, check out


included below for archival purposes. Oy. I consider myself neither
an object purist first nor a Web guru second, for the record...

----------------- 8< full article appears below --------------------

Helping the Web Grow Up
by Chris Oakes
12:55pm 5.Jun.98.PDT

The Web. That vast, global network of computers delivering almost
limitless amounts of information with unprecedented speed. All of it
connected by millions of miles of copper, glass, electrons, and light. A
miracle. Really. Yet, for all that, this phenomenal, gargantuan Net
just sits there passively. It isn't realizing its full potential.

All those pages, all that information -- and so little interaction
between them. Suppose you're editing a Spanish-language document. As
things are now, you'd better have a good grasp of spelling and syntax,
or have a good Spanish spell-checker bookmarked. Wouldn't it be nice if
your document and the spell-checker could find each other, without you
getting involved?

That's the idea behind Web computing, a software architecture that
enables information from widely disparate parts of the Web to speak to
each other. And WebBroker is the first official attempt at standardizing
Web computing.

Last month, the World Wide Web Consortium, the industry body that
recommends open standards for the Web, formally recognized WebBroker.

"WebBroker is a mechanism that enables you to talk to objects over the
Web," said Mike Dierken, senior software architect for DataChannel, the
company that submitted WebBroker to the consortium. Web computing is one
term DataChannel applies to the concept.

The components that technologies like WebBroker seek to connect are
small binary programs that perform a specific task and can open
themselves to other components and applications. Backers of Web
computing believe that much of what is networked via the Web and
intranets can be turned into usable components, if they already aren't.

When interacting via the Web, components offer some intriguing
possibilities. One example is the aforementioned Spanish-language
spell-checker. Encountering a document in Spanish, your word processor
would automatically locate, install, and employ a remote Spanish
spell-checking dictionary that it found on the Web.

"All you need to have is a way to look things up that identify what you
want to get done," said Dierken.

Rather than just having a simple news page on the Web with links to
related documents, the stories, graphics, and pictures could find
complementary information that would update or enhance them. Sites,
databases, and software applications could automatically integrate with
other text, data, and multimedia components that might complement them.

WebBroker's standout feature, DataChannel said, is that its mechanism
for connecting objects speaks "native Web," in the form of the hypertext
transfer protocol (HTTP) and the extensible markup language, or XML.
Thus, anything set up on a Web server, accessible through the hypertext
transfer protocol, can be accessed as a component.

The technology's Web-based design contrasts with the proprietary
protocols of other object "models," which include (breathe here)
Microsoft's Component Object Model (COM) and the Common Object Request
Broker Architecture (CORBA). All which ties into the equally dense
subject of middleware, the software that serves to provide a link
between disparate applications.

The folks who envision a distributed, self-aware Web would like to see
it go from just a static platform for retrieving files to a kind of live

But WebBroker isn't reaching quite as high as that scenario might
suggest. "We're solving a low-level technical issue," Dierken said.
"We're addressing how you can get your code on the Web and make it
distributed right away, without totally redesigning it."


But there is a project with somewhat loftier goals underway at the
California Institute of Technology, called Infospheres.

"Our system is a distributed system where all the objects are active
objects," said Joseph Kiniry, a Ph.D. graduate student who's a member of
the Infospheres research team. He and others are interested in not just
bringing objects to the Web, but making them as sophisticated as
possible -- automatic, self-aware, and intelligent.

As Kiniry puts it, "we are object purists first and Web gurus second."
The Infospheres active object, built using the Java programming language
because of its small size and portability, can "look at itself," Kiniry
said, and turn itself it into a threaded computer process, ready to act
on itself and other objects.

Thus, project members see a future for the Web as a worldwide pool of
objects interacting with one another. Everyone with a computer using the
Web keeps dozens of objects, each with individual functions suited to
that computer and that person or organization.

Infospheres' ambitious vision is one reason the project emphasizes a
system that can accommodate a massive size, i.e. scalability.

A handful of companies are already expressing interest, Kiniry said.
Norfolk Southern, an East Coast railroad company, is interested in using
Infospheres in crisis management over its internal network. When a
problem occurs on the tracks, a train engineer could alert a system
whose interacting, intelligent objects would kick into gear, supplying
people and computers with information necessary to handle the crisis.

Network vendor Novell is interested in Infospheres for creating a more
automatic operating system, Kiniry said.

To these ends, the Infospheres groups work on ways for objects to find
each other, reconfigure themselves, and adapt to changes. Among the
software building blocks are the software technologies of XML, Java,
CORBA, and COM. The project receives government funding from the
National Science Foundation and the Defense Advanced Research Projects
Agency, as well as corporate funding from Novell and Parasoft

The first implementation of the Infospheres architecture was released
earlier this year and the team is now at work on version 2.0.

A super-fluid Web future?

Ultimately, Web computing is a new application for the youthful medium
that, like browsing, uses the Web as its platform. But unlike browsing's
HTML, it's an application without a ubiquitous infrastructure. That
infrastructure is just what protocols like WebBroker and Infospheres are
trying to provide.

Many obstacles remain. Developers will have to resolve bandwidth
limitations, older, object-resistant software, and sheer complexity of
Web computing when compared to Web publishing.

Kiniry expects Web computing's success to take hold when a company the
size of a Sprint or MCI sees the cost and efficiency benefits, and
adopts it.

But if and when that day comes, DataChannel's Dierken notes that the
implications and effects of a Web that becomes a kind of giant computer
are difficult to fully grasp. If every piece of information becomes
dynamic, the Web could become too fluid, too changing. Any sense of
permanency might become all too rare when documents, systems, and other
"objects" constantly update themselves.

"On the one hand, you could have complete chaos," he said. "On the other
hand, you could have complete nirvana."

Nonetheless, Kiniry said he, for one, is eager. "We know that this is
eventually going to happen. It might be 10 years down the line, but it
will happen."


And all that dwell upon the earth shall worship Him, whose names are not
written in the book of life of the Lamb slain from the foundation of the
-- Revelation 13:8