Word Circuits


Forepaper by
for Messenger Morphs the Media 99

Greetings - I applaud the hopeful workshop title: the media *need* morphing and I am thrilled to see the progress brought about by many evangelists for hypertext writing and systems design. Thank you in particular to Deena Larsen and Rob Kendall for organizing and hosting (respectively) this particular e-collection of papers. - Julianne (julianne@trellix.com OR jchat@world.std.com)

Making Users Work via S l o w Reading

I. Definitions

Since the last workshop I have been s l o w l y reading Espen Aarseth's Cybertext. In fact, I'm still wrestling with the first page, where he uses the neologism "ergodic" to describe texts that make their users "work" to select or follow a "path" (the two Greek roots of the new word). Aarseth invites us to use his new words "in any way you find pleasurable." I'll gratefully use "ergodic" and "cybertext" as he defines them, and also offer a small quibble in regard to the former: the second root "path", hodos, might get overlooked. Anyone who wants to emphasize the second root could emphasize its presence with an h: erghodic.

I am also employing "users" as Aarseth does, a term "suggesting both active participation and dependency."[1] Not only does this usage avoid issues related to the term "reader," but as a 20-year veteran of (software & hardware) product development I welcomed this concept as an old friend - "both active participation and dependency" describes my "users" perfectly.

In last year's workshop it was useful to assume that there was such a thing as "user expectations" in confronting fiction (cybertext and otherwise). Each literate culture has a pool of "user expectations" in regard to text, which cybertexts may sometimes violate. Further, as more cybertexts are created and used, some users are developing "user expectations" in regard to cybertexts as a whole, which any particular cybertext may violate.

I use the term "attention" as defined by Simone Weil, Iris Murdoch, Hazel Henderson and others. (That's another paper!)

Finally I would like to register the term "playable", which I first heard used by Chris Crawford in discussing his game Balance of Power. His original design for this game had 60+ countries acting independently according to their own interests. However he found this was "unplayable." His choice of a structure with a duel between two superpowers was a fall-back choice based on the fact that his first (more realistic) system was too hard for users to work with and enter into. From this discussion I infer that calling a cybertext "playable" is an artistic judgement as to whether the text makes it possible to use it (participate in it and depend on it, as above) with some degree of pleasure.

II. Why This Topic Interests Me

The year I was 17, struggling to express myself within a narrow-minded cult, my then-boyfriend rebuked me [2] during one of our homework-session-dates. He said that my ability to read quickly was a "gift from Satan," because it allowed me "to be lazy." Reader, I dumped him (as Jane Eyre might say), but I have become very interested in the question he obliquely raised.

Actually, in addition to being able to skim like the devil, I enjoy several kinds of slow reading, all the way from reading-slowed-down-a-bit, which is analogous to paying more for organic vegetables, to almost-memorizing-on-the-spot, which involves, like moving to a farm and growing all one's own food, intense amounts of both effort and satisfaction. I began wondering whether there were ways (other than writing superbly well - not available to all) to invite or strongly compel other readers to slow down and in the process improve their attention skills.

On a banal level, this may be just one more attempt to grab the reader's eyeballs for longer. On an apocalyptic level, as our lives continue to speed up, the ability to be truly attentive grows ever more refreshing to its practitioners. Attention skills gained in reading can be used to be fully present in other parts of life as well.

III. Questions

  • - What are all the possible types of slow and fast reading, from (say) skimming to exegesis and/or memorization? [3]
  • - What are the implications (for the user) of each type?
  • - Are there ever benefits (again, for the user) in reading slowly?
  • - Are there ways to make a user read slowly, or at least suggest or tempt a user to do so?
  • - If there are such ways, what are they and how do they work?
  • - I assume that "making a user read slowly" is not identical to ergodic aka "making a reader work", but what is the relationship between these two? Is "slow reading" perhaps a subset of "working"?

IV. Assumptions & Experiments

- Slow Can Be Fun. As I enjoy slow reading in certain circumstances, so I assume there may be benefit to others, and would like to investigate this as my resources permit.

- We Know How to Speed Things Up. For years I have been paid to help users read nonfiction faster (or apprehend multimedia faster, or learn faster, or view diagrams faster...). At present I am helping to create hypertext writing software that has as one of its designated goals to help users skim large texts. Much is known [4] about factors that help users read more quickly (and in fact I am using the commonest such factors, titles and headings and summaries, in this piece).

- Can We Just Reverse Things? One of the easy, commercial ways to help users read faster is to find out what their expectations are and then make sure the text (plus helps if any) meets them. I am wonderfing whether systematically violating user expectations would produce a playable slow reading.

In Practice / What I'm Doing...

My attempt at violating all possible user expectations at once was not playable, producing work indistinguishable from inept, amateurish hypertext. (grin) Therefore I decided to choose just two expectations to violate in a systematic way [5]. I am now working on three versions of a short fictional hypertext. The same lexias exist in all three versions. My plan is to bring the three versions to some sort of "first draft completion" state (I can hear you all laughing - with me I hope) and solicit feedback from hypertext users. I will report on this, with URLs for the three versions, in my afterpaper.

- The sigma version is optimized for skimming (of the whole) as well as deep reading (of the individual sections).

- The omicron version is like sigma but with link titles that are clearly misleading [6].

- The omega version is like sigma but with different material linked, in an attempt "not to link the obvious," an idea borrowed with permission from Markku Eskelinen.

All are in Trellix so they can be used with or without a bird's-eye-view map of all the lexia (although not all the links).

Problems Being Non-Obvious

Once I decided "not to link the obvious," I realized that "obvious" could mean any of the following: (i) lexia that users would be most likely to want to read next, in the author's opinion; (ii) lexia with closely related content; (iii) lexia that are strongly related to the current lexia in terms of the structure of the hypertext; (iv) lexia provided to maximize user control and pleasure. It's proving hard to avoid the third type of obviousness without changing the structure of the hypertext, which (as was discussed last year) often is the plot. I am hoping some clarity will emerge as I continue working.

V. Acknowledgements & Extra Credit Reading

Last year, Deena Larsen introduced me to questions about closure. Perhaps in some texts with a structure less well defined than that of Deena's Stone Moons (forthcoming from Eastgate), the pleasurable feeling of achievement one gets after doing a lot of effortful reading might substitute for the thrill of closure. We'll see...

Markku Eskelinen tossed out the idea of not-linking-the-obvious during a speech at HT98. Tempting users into slow reading isn't the kind of temporal manipulation Markku has in mind, as evidenced by "A Node Is Still A Node Is Still A Node" (recently moved to http://www.kolumbus.fi/mareske/page14.html) , and his paper "Omission impossible: the ergodics of time" for Digital Arts and Culture - organized by this workshop's Jill Walker - sounds like it was great! - (http://cmc.hf.uib.no/dac/papers/eskelinen.h tml). However, if he does succeed in getting nodes to morph and/or expireaccording to temporal criteria, we'll need our attentive reading skills.

See you all in the workshop!  


[1] "...suggesting both..." is from page 174, Johns Hopkins, c. 1997.  

[2] For this relationship, I offer in extenuation my youth, hormones, and the fact that he was practically the only other cult member my age (locally).

For years now I have been examining the 'loony' pronouncements of my fellow cult members and trying to figure out the extent to which they might be meaningful. I use the method often recommended by Suzette Haden Elgin (paraphrased): assume the statement is true, and then try to figure out what it might be true of. The most interesting exercise was a study of a statement by my grandfather, an aerospace engineer, that "God blew up the space shuttle [Challenger]." Maybe I'll post that one.

Growing up in a cult, while trying to 'pass' in normal society, requires one to maintain and act according to mutually contradictory goals and beliefs. As described by Sara Ruddick in Maternal Thinking (Ballantine, c. 1989), the specific "practice" of "mothering" places even more dramatic demands on the parts of the brain that grapple with contradiction (e.g. must work to keep child safe AND must support child's autonomous exploration AND so on). Last year we discussed the way "hypertexts often 'feel realer' than linear narratives"; could this be in part because hypertexts are more effective than linears at letting their users wrestle among chunks of content that contradict one another? I assume that work has already been done in this area and I just don't know about it; would appreciate suggestions for further exploration...  

[3] To give just one example, the slowest reading I can think of is where one memorizes the text as one reads, or before moving "on" (whatever that means) to the "next" portion (ditto). When I have done this it produced breathtaking effects. I often practice something similar but short of full memorization. "What is the point of using a text, unless I let it enter into myself?"

User observations indicate that slow and fast reading are frequently combined. For example, a typical session on the world wide web might involve skimming (ultrafast reading) to get to what one is interested in, followed by slower and more focused reading/digestion of the interesting parts. For an applied or commercially oriented presentation of this type of finding, oriented towards nonfiction web content, see the publications of User Interface Engineering (www.uie.com).

 [4] For overviews of commercially-applicable tests of factors that affect reading, see the archive of articles by Jakob Nielsen (www.useit.com); his recent work on comprehension is also interesting. See also the "Usable Web" links maintained by Keith Instone (www.usableweb.com). A combination of academic and industry work on this can be found in the proceedings of our sister SIG, ACM-SIG for Computer-Human Interaction. The SIG name stays that way for sake of pronunciation, although many of us prefer to call the field of study HCI (putting the human first), "human factors," and so on.  

[5] These also involved assumptions, because I'm talking about "user expectations for cybertext" and as everyone knows these expectations are rapidly and constantly evolving. However, putting theory aside for a moment, I came up with several possible user expectations to violate:

  • - users assume that each element of the cybertext was intended by the author for some reason - "there must be a point to all this" - often an erroneous assumption
  • - some users expect that they will be able to use cues such as names, locations, and times (where these are mentioned) to build mental models across lexia involving chronology, identity of speakers, and so on
  • - users expect that the names (labels) of links bear some relation to the material in the lexia being linked to
  • - users expect that the links offered will be to the "obvious" lexia (see Problems Being Non-Obvious, above)  

[6] Clearly misleading (as opposed to "vague") means that they are clear in what they imply, but misleading as to the content they are linking to. Example: a link called "In the Casino, At Twilight" that leads to a dialogue that takes place neither in a casino nor at twilight.


Contact site director Robert Kendall at kendall@wordcircuits.com.
Visit Kendall's Home Page for more material.
Take an
online class in hypertext literature.