"What we need most, right now, is a better understanding of the *craft* of hypertext. We need to master links that control pacing, evoke place and character, that invite us here or lead us far afield."

-- Mark Bernstein

Eastgate Systems, Inc.

The Connection System -- another side to the hypertext programming coin -- is seeking a solution to cross-platform issues for Web-based hypertext.

Coauthored by Robert Kendall and Jean-Hugues Réty

Tools of the Trade

Mark Bernstein, for more than a decade, has been president and chief scientist of Eastgate Systems, Inc., a pioneer company in the development of hypermedia and hypertext writing tools, and world-renowned publishers of original hypertexts--poetry, fiction, and nonfiction. Eastgate Systems, founded in 1982, was referred to in the New York Times as "the New Directions of electronic publishing," a reference to the famous vanguard print publishing house of the 1930s to the present, in recognition of the high innovative and intellectual quality of the company's products. At Eastgate, Bernstein, a Harvard-trained scientist, has created and managed hypertext tools and technologies, including the Storyspace hypertext writing environment which has been widely regarded as the tool of choice for serious and literary hypertext.

With html authoring of hypertexts gaining in popularity, but besieged by problems inherent in the way the two major browsers interpret advanced programming like DHTML and javascript, it seemed a good time to sit down and talk to Bernstein about the origins of Eastgate, Storyspace and the plans this dynamic software developer and hypertext publisher has for the future.

How did Eastgate get started?

When we started Eastgate in 1982, we wanted to improve how computers and people worked together on hard problems, using three new technologies:

    - direct manipulation interfaces
    - literate computing
    - hypertext

These three technologies have always been central to the Eastgate vision. Direct manipulation interfaces are now widely used and fairly well understood, while literate computing -- finding ways to make program code as easily understood as prose -- remains much more problematic.

Hypertext, of course, has been the big change. When we began, hypertext systems were laboratory curiosities. In the past decade, they've transformed writing, journalism, education, and business.

We've always envisioned Eastgate as a small, technology-driven company, a dedicated group of scientists and artists working together to do new things.

Who was the first author you published? What was the public reaction (press, etc.) to the emergence of Eastgate?

Our very first hypertext publication was a collection of position papers from the first ACM Hypertext Conference, Hypertext '87. We saw this as a service to the research community, not as the beginning of a publishing house.

Our first hypertext fiction was Michael Joyce's afternoon, a story. Joyce's work was already circulating in the research community -- you could get a copy if you knew the right people -- and was widely discussed and widely admired. Again, the most pressing need was to create a way for hypertext people to read and study work that others had seen, to create a common ground for talking about hypertext.

But the appearance of afternoon, followed by Sarah Smith's King of Space and Stuart Moulthrop's Victory Garden, also demonstrated that this new medium was going to matter to the wider literary community, that hypertexts were going to be a very important and central part of the future of fiction.

Was it the need for software to write hypertext that led you to developing Storyspace?

Storyspace was begun by three professors -- Jay David Bolter (a classicist), Michael Joyce (a novelist), and John B. Smith (a student of writing). All three, I think, saw a need for hypertext tools that people could use outside the laboratory, tools that writers could use right away.

At the same time, Eastgate was developing a separate system, Hypergate. By the late 80's, it was clear that working together made good sense. Some ideas from Hypergate found their way into Storyspace, others migrated to other systems. For example, the idea of showing whether a link takes you back to a place you've already seen was a Hypergate innovation called a "breadcrumb." We've never done breadcrumbs in Storyspace, but Mosaic picked up the idea in the early 90's and now they're ubiquitous.

Tell us a bit about Storyspace. Specifically what an author can do with it that they can't do with plain html, javascript or dhtml?

Storyspace is a hypertext writing environment. It's focused on the process of writing, not on presentation. It's a tool for building complex link structures.

When you write HTML directly, you're confined to a simple kind of link -- a GOTO from here to there. Worse, making a link interrupts your work; instead of thinking about words and ideas, you need to think about files and markup:

    [A href="....."] .... [/A]
In Storyspace, you make a link by drawing a line. That means it's easy to make lots of links -- to build really complex and fluid structures. It's easy to delete them, to move them around, to try different arrangements. Storyspace tries to make it less tempting to build a rigid framework like the static "information architectures" now popular on the Web; you don't need to pour every hypertext structure into a rigid precast shell.

Also, Storyspace links are dynamic. They can change as you read, by taking into account what you've already read. This turns out to be terribly important for large hypertexts, especially narratives, because dynamic links help create and sustain narrative thrust. Although you can do this with javascript -- Robert Kendall and Jean Rety are building some scripts to make it easier -- the process is tricky and ad hoc.

So it sounds like the 'breadcrumbs' translated into HTML as 'visited links' but dynamic links in Storyspace are different? How so?

HTML links always do the same thing. They don't know what you've already done or seen. They always go from here to there.

Dynamic links change. They take into account what you have already read, or what you've read recently. That means writers can adjust tone and pacing in ways that you simply can't manage without them.

For example, suppose you want to explain a tricky installation procedure and need to be sure that readers have been warned about safety guidelines. A dynamic link can take you to the safety guidelines IF YOU HAVEN'T ALREADY SEEN THEM.

How do you perceive Storyspace functioning for authors, as contrasted with programs like Dreamweaver, Shockwave or Flash? Does it include many of the same features, alternate features?

This is a real grab bag of nice software!

Dreamweaver is an HTML editor; I use it quite a bit. Dreamweaver is great at making pages -- it's best when you have a complex page with lots of design questions that need to be just right. Storyspace doesn't do much with pages; Storyspace is about links, about the structure of connection between pages. It helps you break away from trivial navigation architectures -- simple lists and outlines -- without getting lost in a hopeless tangle of links.

A very nice technique is to use something like Dreamweaver to design the look of a site -- the way each individual page appears -- and save the Dreamweaver files as a Storyspace *template*. Then you write the hypertext in Storyspace, so you can get the linking and structure exactly the way you want them without getting bogged down in file names and tags. Storyspace then pours the pages into the template, building a custom site exactly to your specifications.

So what kind of learning curve could an author expect to encounter in order to figure out how to use Storyspace effectively? How would this compare to learning DHTML or javascript?

Storyspace is very easy to use, but this is really the wrong way of looking at things. The underlying skill for Storyspace users isn't a programming language -- it's language itself, the craft of writing. Specifically, the craft of writing hypertext.

It takes real work to learn to write well, to craft structures that are clear and yet flexible, that guide diverse audiences without imposing a single structure on everyone. But this kind of challenge is not a lot like learning a programming language. Apples and oranges.

Given the problems inherent in the way the current browsers interpret DHTML, how does Storyspace approach the cross-platform problem? Will the created hypertext function only on the platform for which it was written?

Storyspace itself uses no DHTML or javascript or css.

Platform issues really only arise when you're working with intricate page designs -- crafting elaborate templates. Then they're exactly the sort of cross-platform issues you're accustomed to.

Lots of people use Storyspace to create stand-alone hypertexts -- self-contained applications that they distribute on CD or disk. In that case, there's no browser involved and so the usual cross-platform issues go away. Of course, there remain the differences between Storyspace for Macintosh and Storyspace for Windows, but these too are usually easy to overcome.

Can the new E-Book readers access Storyspace authored hypertext?

Yes, if you export to HTML. But the real winner will be E-Book readers that read Storyspace documents directly, so they can handle dynamic links. That will come, probably in about a year.

Right now, the e-book reader business is mostly about distribution deals. Sooner or later, they'll want to offer something really new, something that conventional books clearly can't do. Right now, it's all about branding and strategic partnerships. That will change, and Storyspace will be right there when it does.

What about Flash?

Flash is for animation. Interestingly, it has strong hypertext research roots -- one of its architects, Norm Meyrowitz, was deeply involved in the classic Intermedia system at Brown. We'll always use tools like Flash to create animations for our hypertexts, but these are primarily spot technologies, concerned with an object rather than with the big picture.

What's next?

We need a new kind of hypertext tool -- a tool for planning and for taking notes, for making sense of the worlds we inhabit and imagine.

We have all this speed and power now -- Storyspace runs a thousand times faster today than it did when it was new. We can use that power to help us work with ideas, not just to add movies to our pages.

We're working on some really nice new tools. Still lab curiosities -- returning to our roots, really. They'll build on Storyspace, of course, but they'll go in very different directions.

Of course, there will always be a Storyspace, too, and designing a fresh Storyspace is going to be lots of fun.

Now this sounds intriguing. Without divulging any information that might be proprietary, can you tell us a bit more about what a 'fresh' Storyspace might look like, and when it might be available?

We've given a lot of thought to the question, "What *is* Storyspace?" On the one hand, it's an obvious question: it's a program that's been used by hypertext writers over more than a decade, so it's very familiar. But it's not just a cardboard box or 100,000 lines of code; at its core, Storyspace is a way of writing, a way of looking at structure.

So we want to keep a lot of the elements that make Storyspace. Terrific hypertext maps. Links that are easy to make. Writing spaces you can pick up and hold. But at the same time, there are many new things we can do now that weren't possible, or weren't practical, even five years ago. Plus, there's a lot of wonderful research to assimilate and make generally available.

I'm sure many authors would appreciate any tool that would allow them to concentrate more on their creative work, and less on programming. Early HTML was relatively easy to learn, if limited in implementation. Does it concern you that the hypertext field is requiring increased technical knowledge from its authors?

Not at all! The appearance of change is an illusion: serious work and serious writing always requires knowledge.

Some of that is learning jargon. That's all HTML (or DHTML) is: a way of expressing yourself concisely within a very limited domain. It's just like learning to read proof; it's a nuisance to learn the the lines and squiggles, but once you take the trouble you can do things that need to be done.

Hypertext systems people have exactly the same perception, incidentally; in the 'good old days' they only needed to know about programming, but now they also seem expected to know about interactive fiction and postmodern philosophy.

The hard part is learning theory and learning craft: learning about hypertext structure, and mastering its use. Learning how to use a few additional tools is easy, but the craft (like everything in art that matters) is always hard.

What impact do you, as a publisher, wish to have on the future of hypertext, its content, possible markets, software innovations?

It's now absolutely clear that the future of literature lies on the screen, and that interlinked literature -- hypertext -- is the kind of writing best suited to screens.

What we need most, right now, is a better understanding of the *craft* of hypertext. We need to master links that control pacing, evoke place and character, that invite us here or lead us far afield.

Studying hypertexts -- understanding what makes them effective -- is terribly important. We've had more than a decade, now, to read afternoon and Victory Garden, but we're still learning how they work. Look at Jill Walker's Hypertext '99 paper, for example: a wonderful parable that shows how different ways of approaching a hypertext lead to very different responses.

We need to look at hypertexts as they are. For a long time, people saw links and hierarchies, and they wrote as if hypertexts *ought* to be links and hierarchies. But if you look at actual work, you see much more. Cycles are the core of real hypertexts. The tension between simplicity and elaboration, between the garden and the forest, is crucial; that's why my paper on Hypertext Gardens has gotten such a wide response

We need to master mirrorworlds and montage, feints and tangles, and all the rest. We're still finding *names* for these things. That's the beauty of this moment in history: we have lots of serious hypertext to read, and we're still free to wander, to observe, to be the first person to see something and say, "let's give this a name."

We also need more hypertext, of course. Serious, finely-crafted, innovative hypertext. Nothing beats that.

Back to the Means