It's an easy and common misconception that reading means visiting each and every node once and only once. Six or seven weeks into a course, students will realize they won't get to or have to read everything, and then they're okay. (This happened to me in college with my regular reading for my English major, but I'm not sure that the sort of physically overwhelming amount of reading is a common experience before grad school for everyone, so I'm not sure whether I can generalize that analogy.) Still, though ...

.... "Are we reading yet?" - how do students know they've done the assignment in a non-linear text that doesn't necessarily chart their completion like page-counts do? The structure in hypertexts are often too obscure. There isn't an easily accessible history. And students don't have any experience what it's supposed to be like. A counter-argument is: don't underrate existing "hypertext"... games, websites, etc. Students know when they've gotten through those.

Information retrieval as hypertext-- Is searching on google equivalent to reading Patchwork Girl? It's a different model of authorship, of literacy. Or is that difference simply a matter of mapping and of history? Wth CubicEye you clearly build a text out of a web search. The spatial model is hard to get for some people.

Is resistance to hypertext (on whose part?) intrinsic or habitual?

Why the vehement resentment to hypertext in general? A reluctance to give up existing competency and literacy for a new medium. Wanting a definitive experience, a closure even if it's misleading or false. Readers construct a schema as they read, their own mental map, and hypertext is disorienting 99% of the time. "[In most hypertexts] you don't have a schema for predicting what's next, for predicting where that text is going to go. We predict all the time." There's also an age aspect to this-- older writers are more invested in existing schema for reading.

Okay, but what about in social discourse, where you also can't predict the direction? Is hypertext better at representing social discourse and representing thought across discursive boundaries (i.e. multiple people's viewpoints)? Even in those cases, organization is still perceived as seriousness... and you can't fully get hypertext by maping it onto schema that you already know (ETA: it's like using tinderbox as just an outliner, or just a mapper). People resent you as an author taking their transparency away, and they don't know what's at the other end of the learning curve. There's also resentment from authors who feel that their artistry has been superseded or overshadowed by concerns or artistry of technology or of user interface. Who wants to worry about User Interface? It's ironic that "complex" pages end up getting perceived as "for kids".

David asks: why make hypertext optional? Answer: because the learning curve is too steep for time to teach the tools with the whole class. Because their habit is to add things on, not build - to create threads, not cross-linking or webs.

The scariness of the idea that the grade is dependent on reading everything affects the professor, too-- must the professor fight hypertext itself to read everything?

It's also scary for professors to make the transition to reading and grading hypertexts because it's not so clear where and how to 'comment in the margins'. (my own professor, who was comfortable with Storyspace and assigning hypertexts, still made me in one class hand out a linear version of a paper to my fellow students and that is what he graded and commented on.)

How do you explain to a class the utility of complex structure when other classes and the educational system privilege short stuff and argumentation along the five-paragraph model (intro, support, support, support, conclusion)

How do you grade a hypertext, especially if it involves programming code? It's probably better not to comment in the code, but to make a separate report on the code. You must be careful to communicate the criteria. You can have other students make comments with pointers (something that is especially easy in Tinderbox, since you can cluster comments in their own space, group them visually with adornments, but have them linking to anywhere in the main text. There's also no reason not to have comments accumulate, and it's easy to pull comments into a selection of "comments worth sharing" in class discussion or with later classes.

One aspect of student anxiety about hypertext is a matter of escaping the One True Reading model. This anxiety is fed by the game analogy-- repetition is often equal to failure in games unless there's evident proof of advancement. Another stone on a pedestal (thinking of Dark Castle, here), etc. In most hypertexts there's precious little evidence for a student's advancement outside of a reader's self-assessment... and we're teaching them how to be readers in the first place. Too few hypertexts make explicit that a circular structure in the reading is in fact a spiral.

An excercise for a class: "is this website hypertextual?"

"Who's Cribbing" by Jack Lewis - just tryto map it. ... trying to pushes students out of the comfort zone and gets them to see some of the difference between complex linear text and "native" hypertext.

Writers (among the students?) like "WOE", "Is me past" (?). No luck with "Afternoon". They also like "The Unknown".

You can create a hypertext physically with notecards, yarn, and colored paper. Using a physical example gets the metaphors, the paradigm, down and then you can bring it back to the writing: writing is a process of taking all those ideas, making each grammatical, and putting transitions (links) between them. This gets the visual learners. You can see them get it when they link words, rather than boxes (lexia).

We need forii (forums?) for pedagogical practice in hypertext that doesn't reduce to academic papers. That includes actual practice and discussion between practicioners as well as advice for those who might be willing to incorporate hypertext but unable to crest the learning curve.

Anne taught a course that used hypertext to examine the stories of a "silent generation" in Japan. Her class used hypertext to break through a cultural avoidance of WWII stories by having students interview their grandparents and relatives for their personal stories to create a hypertext from them. One student interviewed her grandmother, who was reluctant at first, then opened up and shared photos, talked... and talked and talked. She had never in her life had her story asked of her.

even with a web page, people think of each piece as linked linear bits, rather than a network made linear

Why was this project different as a hypertext than as a "conventional" text?

• The multimedia forced multiple voices. (Multimedia vs. hypertext was an issue in our discussion. "What is hypertext?" It's an irritating question but one for which we must have several, contextual, lucid answers.)

• Learning new tools and structure forced new kinds of questions.

• Creating the product as hypertext freed the student not only in structure but in content-- she could create contrasting voices.

The project led to empowerment and voicing so powerful that it could not be overstated. People break out of abusive relationships, discover family stories, discover multiplicity when faced with the inherent multivalence of hypertext. The students' responses to American responses to the WTC attacks.

Students struggled with the difference between a public act (for the web) and a personal one (for the professor) given the personal nature of the assignment. There was tension between commitment and participation in something larger. The public nature of the project, however, also gave the project an authenticity-- the students were involved in a real way in Japan's current discussions over its history in WWII, in building archives of memories. Their engagement in the assignment and the tasks behind it was immediate and (helped by the distance from known rhetorical structures) not defined by ideological authority structures. It was hypertext and it was on the web.

Would it be "selling out" a hypertext to give a "standard" version of a hypertext?

Mark introduced Tinderbox, since it was a fairly new product when the conference happened. Some quick notes from that presentation:

• Tinderbox split functionality off from Storyspace.

• Anders Fagerjord's site uses tinderbox, and he has tips sometimes. Markbernstein.org is a tinderbox blog.

• "The Victorian Web is like a blog of a community." NO - it's edited, 'posts' are approved by a secondary authority. YES - blogs get edited. NO - secondary editing party. YES - blog clusters get unlinked and ignored out of existence.

• In a war, hypertext would be lost. The community would be lost. They're all overexcited geeks.

• Agents can get distant blogs

Is a Tinderbox-built page a hypertext? No, there's no hiding and revealing. Yes, if you make links pop-ups? No - it's a rendered collage, hypertextual but not a hypertext. No, because you can't backtrack. Well, it's a hypertext with a non-hypertext projection, then.

Where Storyspace was about dynamic links and presentation, Tinderbox is more of an organizational tool. It's about notes, indexing, organization, agents. Eventually it would be nice if they were interoperable, so that you could drag things back and forth.

Storyspace has been around since 1987/1991 (what are those dates?) ... and has therefore spent more than a decade as a chunk of code. That's pretty rare. Victory Garden and Afternoon are still there after a decade. Things that were unusual about Storyspace then:

• links.

• Can find the spaces in another file despite being renamed, moved, a different version.

• Links were stored within the file.

• A text space = a window.

It was a tool for reading and writing large hypertexts, especially focused around narrative. There are features that are still unusual in 2002:

• Guard fields to qualify the destination of links

• external links are stored internally

• you can have overlapping link sources

• you can make anchors appear and disappear

• two-handed reading (for showing links, at least).

Sarah (?) then discussed how she's used Storyspace and seen Storyspace used with tree-killing writers. Writing is a non-linear process, and Storyspace tools let a writer work in a virtual multi-dimensional space. You can create assets and then draw upon them-- creating spaces then creating paths through them (multiple paths, even) which then become the narrative flow. You can create aliases which you can then organize in various ways. You could place the spaces' aliases in a two-dimensional chart with characters along the y-axis and the flow of the story (or chapters) along the x-axis to show when characters come and go. You could lay the spaces out along the narrative flow in the x-axis and place them on the y-axis according to the tension to plot that in your novel. Or you could have your proofreaders do the same to show you their perspective. This is a process of organizing and filtering your assets, of bringing the words from data (about characters, places) to story. Plus, you might note that it's easier to get (and see) nonlinearity from this approach.

It would be really useful to watch authors working, especially together. In this roundtable we're working with abstractions, but seeing people work, try to make meaning out of interfaces, see how people share ideas when working physically together... that would really help us see how to make the tools facilitate what they do. How to build Tinderbox into a personal knowledge manager, beyond being just an information manager. At the same time, you need to be aware of and careful with your niche-- misspent expectations killed Agenda. Be careful about what is under your control vs. automatic, etc.

What would a hypertext reading group look like? A house party? A workshop, like a writer's workshop (and what would that look like)? A correspondence course might be a sensible approach, given the relative paucity of hypertext scholars and geographic dispersion. This would be a great way for writers to perform usability testing on their hypertexts. Would you want to follow the "cabin in Maine" or the "weekend house party" model? One participant related having a workshop where each participant brought one work, sent it around a few weeks before, and then discussed them for two days.

Issues for any sort of group would be: ways to share commentary (including links or visual arrangements), ways to work independently yet together simultaneously. Could you build a tool that would help share hypertexts? (Some Tinderbox files are small enough to share over instant messaging, but not all, and there's no way to actively serve a tinderbox file so that multiple people can see and edit it simultaneously on different computers.)

DAlso, don't forget that there is fun in not knowing, in serendipity and re-vision. When is confusion, misdirection, not knowing fun and when is it frustrating? Is the author or reader at fault for the frustration with Finnegan's Wake? Why do some of these texts gain a cult following or critical respect, but not others? It's not just complexity or simplicity-- why do people go for 12-tone music or atonality for that matter, but not for hypertext?

Rob (?) points out that there are visible vs. invisible literacies. Sarah adds that it's also about text vs. language-- people have less trouble with graphic (multimedia) hypertext than with blocks of text. Ann wonders whether it's just the level to which the reader can cop out of having to figure it out. When is it a matter of "fixing" the interface, and when do you want not to? A diagram (say, architectural) still contains some intentional and unintentional ambiguity.

We then looked at argument. There's some use in understanding and displaying intermediate structures, the way we're thinking in the midst of an argument, before we've completed a design. A final product inevitably has had questions decided or choices made that we may want to reconsider or present as questions. There's the idea of a second choice, which might be better for another context. We can represent that ambiguity in hypertext by hiding alternatives, or revealing them. There's no reason not to put other viewpoints in.

Aren't there issues with presenting meta or side conversations in a publshed hypertext, though? Is there a way to present it that's not ironic?

UVA has a new imprint for electronic scholarship in the humanities.

Links can contain a lot of information. It's a challenge for any given interface or presentation to incorporate that liminal data into the narrative flow. You could use tooltips for mouseovers. In a linear presentation you could put link text between the contents of the lexia. What you want to do (often) is telegraph that you'll be sending someone (through a link) somewhere else... especially if it's outside your text. And in this context, it's worth thinking about what outside means. By linking resources you bring them into your own text to some extent... but don't have editorial control over them.

Then we thought about intermediate levels of "telegraphy". You don't always want to give perfect information about where a link goes because you risk the reader assuming they understand or not being interested: and they won't follow your link. Yet you don't want to leave a reader stumped as to where to go or wandering aimlessly (unless that's your goal) because then they're lost. One neat idea is to create indeterminate links or links that can be hidden or revealed according to some level of exploration that the reader is willing to undertake. They could set their willingness to stray from a topic, or stay within an area. I brought up the hypertext in which I colored links according to link type (definition or explanation).

Do you need guardfields or programmingto enforce argument, structure, or complexity?

Understanding narrative shape is even more critical to new readers.

How about random entry into a hypertext-- when someone drops into your website, how do they get context, a sense of where they are, and whether their interest in the page they're on translates to an interest in the rest of your site? Short of framing (sidebars, logos and explanatory text), you could present a pop-up window if the referrer doesn't match your own site, you can have textual breadcrumbs, and you could present a tiny icon of a map which is linked to a larger map.

How do you represent where the reader is in the larger argument (which may or may not map onto the site map?

Don't forget that when it comes to hypertext, most readers are "children"... they need to know strategies.

Mark Bernstein brings up the idea of thinking of links backwards-- everything is linked to everything else, and you as an author determine what *prevents* a visit. (Mark wrote a paper and presentation about a card game Thespis to model this.) Then set several players (agents?) who get points for going to certain lexia and provide links... and you've got structure. This goes back to the idea of a link server which revises the link structure for you based on criteria. What rhetorical structures (besides programming) will make the sculpturally, thespis-like approach accessible to authors? A timeline can help you set this up for a static site, but it's still dauntingly complex to make. And people will find the game analogy anaethema, not serious.

Also, the UI is difficult. You need an interface like the path builder, or like Windows Server's "resulting policy" feature which shows you: at this point on this path, what are my constraints and conditions?" Also, "what can't I do?" (Plus: do you show the reader that so that they have a sense of how far they've gotten through the potential of a text?)

There are some big challenges with the Thespis approach. It requires knowledge of the text; it is authorial only. Since it's drawing from a pool of lexia, the text itself must be compelling or the reader starts playing it like a game because the conditions become the focus of reading. If you can see where you're going and where you're not, a game develops around getting places.

In the metaphorical setting of thespis, what about false cards? Cards that a reader should play, but which reconfigure the goals like a game of Fluxx. At this point the people at the table wanted a way to see that possibility, a way to communicate loss ("cards you'll never see are..." or "this link has precluded you from...". Doing that discourages a linear text or linear reading. One interface for this might be a sort of possibility graph-- as you play cards your hand grows, shrinks, and eventually dries up. Parts of a map grow dim as they become inaccessible (or read)

There's a distinction between games and hypertext in the outcome. Having the outcome always the same might be bad game design, but good hypertext. It's a matter of agency and identification with the protagonist. The character in a good text, as in a game, must be limited.

Integrating transitions into a text where the order of the lexia is flexible is an enormous challenge. You need to either create nodes/lexia that are transitions or find ways to integrate transitions into the nodes. It would be wonderful to find a way to avoid separating the text into nodes in the first place, and to avoid making the transitions static within the nodes.

--------------Diane's highlights------------------

It's important to impress upon new readers the ida of relearning... to show them that it's a new and different way of reading and writing.

It would be neat to have a whole hypertext course, rather than just having a "unit" on hypertext in another course... you should use a hypertext as the course workspace, archive it, document it.

Emphasizing the game aspect of hypertext could be a way to get it to students, though it then faces the tension between ease and comfort of use vs. the seriousness

THere aren't enough models for what it means to read or write hypertext successfully. Some people seem to know it, but have a hard time articulating it.

We need a catalog of ways to show that life itself is hypertextual even outside of hypertext.

Links are a disappearing work-- how do we see links as a text themselves, how do we evaluate that text?

Mapping plot practices to hypertext and viceversa is wonderful, usefully interchangeable. Thinking of a book as a view onto a hypertext is a neat metaphor. (I feel that Proust's Rememberance of Things Past would be a particularly easy text to think of this way.) Tinderbox is good as a tool for creating multiple projections of a text.