Museums and the Web 1999

Best of the Web

Archives & Museum Informatics
2008 Murray Ave.,
Suite D
Pittsburgh, PA
15217 USA

Join our Mailing List.

Published: March 1999.


Telling Stories: Procedural Authorship and Extracting Meaning from Museum Databases

Steve Dietz, Walker Art Center, USA
The Analytical Engine weaves algebraic patterns just as the Jacquard loom weaves flowers and trees.
--Ada Lovelace, 1834

You want to write a song but first you have to invent a piano.
--Toni Dove, 1998

From the Library of Alexandria to the World Wide Web

It has been suggested that the pigeon holes for the vast collection of papyrus scrolls at the library of Alexandria were a technical innovation that greatly facilitated access to the library's holdings. (Canfora, 1990) Ever since, cultural heritage institutions and their collections management software vendors have been attempting to wrestle under control the information they know about their collections. Generally, however, museum automation systems, computer-based and otherwise, have been as much for the management of resources as access to them. It is only with the rise of the World Wide Web that, arguably, a new emphasis on public access to museum informational resources has arisen. Inevitably, as a field, we have begun to succeed in providing the public access to much of the information we as managers have spent the last 20 plus years automating--that is, making "machine readable." The question is, in part, is this the information the public is looking for and, in part, how can we make it compelling.

In 1997, just as projects such as The Thinker ( and the National Gallery of Art's (Washington, DC) powerful Leonardo system ( were coming online, at the first Museums & the Web conference, the panel "Accessing Museum Collections Over the Web" ( was raising important questions about the value to the user of all this effort.

Not surprisingly, the Web, despite its inadequacies and the need for museums to rethink their information practices, was seen as critical to the success of providing access to museum information integrated across knowledge domains, telling educational stories, and creating dynamic and compelling interactive exhibitions.

Jim Blackaby and Beth Sandore focused on "practical approaches to data organization and access," using work at the U.S. Memorial Holocaust Museum and the Oregon Historical Society to demonstrate how "you could put your fingers on all of the information about a specific topic in a museum, regardless of whether it was drawn from the objects collection, exhibit catalogues, the library's holdings, or the prints and slides collection." (

Kevin Donovan gave an impassioned plea that "Simply providing the public with access to data is insufficient to satisfy the goal of public education." Museums, at least in their databased, public outreach efforts, need to kick the object-centered habit.

To achieve on-line Public Learning as I have described it museums need to offer enriched, value-added content that supplements label copy and object records with well-told stories that captivate and enlighten. . . . museum information systems must evolve from object-centric collection management systems to context capable content management systems. (
Howard Besser treated us to a mini-history of museum information systems and how they differed from interactive exhibition systems:
Historically, most museum automation efforts have been driven by the need for record-keeping and inventory control, and have resulted in collection management systems. A smaller set of automation efforts has centered in museum education departments, and has focused on interactive exhibits. The vendors, software, tools, and platforms for computers used in collection management have been very different from those in interactive exhibition. (
Today, as even a cursory scan of the 1999 Museums & the Web conference program ( shows, there are countless remarkable efforts to yoke databased museum information to web-based outreach programs. Are we there, then? Can users put their fingers on all the information they want? Are we telling stories that matter? Are we creating compelling, interactive experiences?

The answer is yes, there are some provocative models, and no, it will be impossible to implement them on a universal scale as they are currently constructed.

From the Collection to the Library to the Archive:
Integrated Information Access

Arguably, we are furthest along in the arena of integrated information access. A joint project between the Walker Art Center and The Minneapolis Institute of Arts, ArtsConnectEd ( is indicative of the direction and state of such efforts.


Figure 1: ArtsConnectEd, an integrated information access joint project between the Walker Art Center and The Minneapolis Institute of Arts, with a particular emphasis on use in the K-12 classroom by teachers and students. (

Contrary to Donovan's assertion that collection management systems needed to become more comprehensive MIA and Walker separated information management from access. We take periodic dumps of the information in our various managements systems, "scrub" the data, and put it into an Inquery database. This allows users to find information about a specific topic in either or both museums, regardless of whether it was drawn from the objects collection, exhibit catalogues, the library's holdings, or the archives.

The system is quite powerful and includes over 100,000 collection, library, and archive records from both institutions, more than 1,000 dynamically zoomable object images, hundreds of hours of audio and video, numerous curriculum units, and interactive hypermedia modules. Perhaps most usefully, any search provides related items that the searcher may not have thought to ask for.


Figure 2: Search Deluxe allows users to specify the scope and type of search results. The results identify related items that are not necessarily part of the initial search result set.

While the search engine is only one part of ArtsConnectEd--there are also extensive interfaces to synthetic classroom materials--after three full days in the Dayton-Hudson usability lab at two different stages of the project, we are forced to recognize that there is strong corroboration that the information we are providing--i.e. the information we have created over the past 20 years--is not always what the user is looking for. As one participant who had successfully completed one of the usability scenarios put it, "where's the information." He didn't recognize what the museum collection database had to say about an object as something of interest to him.

Donovan succinctly states what many art museums in particular do not like to admit:

as any paleontologist who has walked into a gallery of Trecento Italian paintings can tell you, or any art historian who has been confronted with a display case filled with geological samples knows, objects do not give up the richness of their history, context, and meaning easily. Museums need to wrap layers of interpretation around the bare fact of an object before the public can begin to grasp its significance. The same holds true for the information we offer on-line to our publics.
Simply providing users an object's creation date, creator name, medium, and so on is not enough, even if the information has been culled from different domains at different institutions. For many users, we need to provide a richer environment, a more compelling online experience.

From Categories to Stories

While object-centeredness may be good from a museum collection point of view, Donovan argues that it is not the best way to engage the public.

Instead of leading with the object, lead with the story of the culture, historical context, important people and places, and their importance. Tell engaging stories with objects woven through them. Do so via entertaining, prescribed paths that both lead the user lightly by the hand and encourage curiosity, exploration and serendipity.

This approach of animating the museum experience and museum objects with a narrative structured is echoed by the eminent information architect and exhibition designer Ralph Appelbaum (1996), who designed the magnificent installations for the American Museum of Natural History and the U.S. Memorial Holocaust Museum:

In our recent work for the American Museum of Natural History and the United States Holocaust Memorial Museum, we have experimented with ways to activate the total experience of communication goals. We look for architectural and environmental metaphors for the key pedagogical concepts behind an exhibition, so that space traditionally left neutral is given voice. This approach casts a broader informational net to engage the receptiveness of different visitors. We feel the results are seen in people's sense of immersion, attention span, and enhanced memory of their experience, provoking them to discuss the exhibit with others and engage in activities such as reading more on the subject, visiting related sites, or becoming more involved with the museum.
As one example of an online story-context, Donovan cites the Monticello Web site ( html), which allows the visitor to follow a day in the life of Thomas Jefferson. This narrative may be particularly powerful because it replicates our internal diurnal clock. The day-in-the-life narrative is, in fact, fascinating and providing a quotidian context for the use of Jefferson's writing desk probably does mean a lot more than most people than the straight collection database record that probably exists for it.


Figure 3: A day in the life of Thomas Jefferson's Monticello. (

Nevertheless, I do question how far one can actually explore with just a pre-determined narrative, which remains largely textual and which the visitor reads more than she participates in. Perhaps more significantly, it presents, basically, a single point of view. Presumably, Jefferson's slaves had a different idea of a day at Monticello. As James J. O'Donnell (1998) writes:

The author is already an endangered species and rightly so. The notion that authoritative discourse comes with a single monologic voice thrives on the written artifact. Both oral discourse . . . and the networked conversations that already surround us suggest that in the dialogue of conflicting voices, a fuller representation of the world may be found. The notion that reality itself can be reduced to a single model universally shared is at best a useful fiction, at worst a hallucination that will turn out to have been dependent on the written word for its ubiquity and power
World Ceramics
Through Your Eyes

Figure 4: World Ceramics (, The Minneapolis Institute of Arts; Through Your Eyes (, Walker Art Center.

Museums rightly strive to present authoritative information, but this need not be limited it to a single, authorless (institutional) point of view. And while multi-vocality is not specific to digital media--think of Rashoman---the ability to present parallel and even conflicting narratives is an easy task. In the ArtsConnectEd project, for example, both Walker and MIA have created multiple-narrative presentations. In "World Ceramics," you can choose the point of view of an artist, an archaeologist, or a curator ( In "Through Your Eyes" (, you can "tour" the Walker's varied programs through the particular experiences of different members of the local community.

From Multi-Vocal to Hyper-Linear

Arguably, it is more natural for humans to think in associative webs rather than in strictly linear fashion. Until the advent of the hyperlinking capability of the World Wide Web, however, non-linear narrative was largely the province of experimental literature and cinema.

And whether one sees it as a vice or a virtue, museums increasingly have to recognize that their future audience will have been brought up on the jump cuts of MTV, the branching game worlds of Nintendo, and the hyperlinking of the Web. In such an environment, hypertext is a survival strategy.

Artists remain the primary expanders of hypertext narratives. A brief sampling might include Mark Amerika's Grammatron (; Olia Lialina's My Boyfriend Came Back from the War (; Melinda Burgess's Line (; Darcey Steinke's Blindspot (; Judy Malloy and Cathy Marshall's, Forward Anywhere< (; and Claude Closky's Do You Want Love or Lust? (

In "Hypertextual Consciousness," Mark Amerika, one of the best-known practitioners and theoreticians of hypertext and hypermedia, posits hypertext as a kind of subversive experiential space that also hints at auto-generation ("Writing Machine").

Hypertext, as a concept, suggests an alternative to the more rigid, authoritarian linearity of conventional book-contained text. In the middle of reading or viewing a hypertext . . . the reader/participant (co-conspirator) is given a number of options to select from so as to break away from the text-block being presently read, thus enabling the reader/participant to immediately enter a new writing or textual space.. . . A hypertextual viewing style would be one where the reader/participant (co-conspirator) actively clicks their way into new writing or textual spaces . . . . Hypertext, as a more narratologically-minded (fictionally-generated) clickual reading/viewing style, could be construed as kind of Writing Machine.
Active essay

Figure 5: Mitchel Resnick, "Exploring Emergence," an active essay. (

In a not dissimilar vein, Mitchel Resnick, a professor in the Epistemology and Learning Group at the MIT Media Lab is researching the idea of "active essays" (
), by which computers make possible a new form of narrative expression, in which manipulable computational objects are integrated with text, graphics, and video.


Figure 6: Julie Luckenbach and Louis Mazza, "Beuys/Logos," Walker Art Center (

The Walker used the format of the hyperessay as its online "catalog essay," Beuys/Logos ( for an exhibition Joseph Beuys Multiples. Written by Julie Luckenbach and designed by Louis Mazza, this approach allows the reader to chose what interests her the most and follow that thread, including beyond the essay, so to speak, to other resources on the Internet that may have no formal relation to Beuys--certainly not to this specific exhibition--at all.

The "traditional" complaints about hypertext, of course, are that it destabilizes our notion of narrative as being a specific trajectory, it diminishes the authorial role, and it simply isn't rewarding. The first two concerns, as the philosopher Wiilliam J. Mitchell (992) suggests, are well founded:

So we must abandon the traditional conception of an art world populated by stable, enduring, finished works and replace it with one that recognizes continual mutation and proliferation of variants - much as with oral epic poetry. Notions of individual authorial responsibility for image content, authorial determination of meaning, and authorial prestige are correspondingly diminished.

From Storytelling to Conversation

Socrates said that writing forces you to follow an argument rather than participate in it. He hated the idea that someone could write something down, and then go off and die, and you would never be able to argue them back out of it. What we think is so important about writing he thought was horrible.
--Alan Kay (1998)
Kay's Socrates story points to an important difference between storytelling and conversation. Storytelling, while certainly influenced by audience responses, is one way, declamatory. A good conversation, on the other hand, is two-way, equally influenced by both parties. As Roger Schank (1994), director of Instructional Learning Systems at Northwestern University, and developer of "ASK" puts it:
ASK is a form of hypermedia based on the metaphor of having a conversation with an expert (or a group of experts). In this conversation, the user provides questions and the ASK system provides the answers. In a real conversation, both participants influence the flow of discussion. In an ASK system, the same holds true. The user influences the flow by selecting which questions to pursue and the ASK system influences the flow through the answers it. (

Amerika's "Writing Machine," Resnick's "active essays" (a term coined by Alan Kay), Mitchell's decentered author, Schank's conversations all point to an important and distinctive characteristic of digital media. This characteristic is commonly referred to as interactivity, although in this context a better term might be computability, the basis for procedural authorship.

From Interactivity to Responsiveness

Besser, in his 1997 panel presentation, was making a distinction between the software for museum information management and "interactive exhibitions." Exemplary examples of such an interactive exhibition might be "Academic Interactive Exercise: Robert Cumming" ( or "Artwork of the Month: Lorna Simpson, Wigs" (

Both of these projects take advantage of the powerful capabilities of digital media to create activity-based interactions based on an original artwork. Neither, however, is interactive in the sense that artist and media theoritician Simon Biggs argues for:

Both the term "navigation", and the sense in which it used, represents a narrowing of the possibilities for interactive media. The idea of navigation is primarily founded on a very traditional notion of what an artwork might be. Fundamentally, the use of this word implies work which is more or less fixed in its content, and through which the reader can "navigate" in a non-linear fashion. This allows the emergent illusion that the reader is experiencing a dynamic and interactive work. Such work however is not interactive. . . . An interactive work is significantly different. . . . . The term interactivity can be used to refer to those works which feature some form of responsiveness to the reader, where that responsiveness causes the content of the work to be altered.
Great Wall of China

Figure 7: Simon Biggs, The Great Wall of China, screenshot (

Biggs's own project, The Great Wall of China (, is a "writing machine" that is more like a musical than written composition, and at first blush it may not seem very appropriate for the serious world of vetted museum information. It is important to keep in mind, however, that the apparently random texts are responding to user input--in this case the movements of the mouse.

One of the perennial problems with very large databases, like The Thinker or as we hope AMICO ( will become, is that of browsing. How do you give a sense of the scope of the collection in a way that allows for serendipity? The traditional answer is a random slide show that has some kind of filtering mechanisms. What if, however, there was some other kind of responsiveness (two-way) that did not rely on either fixed categorization that the user may or may not be familiar with or completely random display. It is easy to imagine such navigation as being both valuable and entertaining.

Many other artists are working with responsive interactivity, including: Paul Vanouse, Consensual Fantasy Engine ( and Persistent Data Confidante (; Toni Dove, Artificial Changelings (
people/murtaugh/tdove.html); Rebecca Allen, Bush Soul (; Jane Prophet, Technosphere (; David Blair, Waxweb (; Natalie Bookchin, Databank of the Everyday (; Jeffrey Shaw, Legible City (
guggenheim/mediascape/shaw.html); Maurice Benayoun, Is God Flat? (; Ken Goldberg, Jester (; Joseph Squier, io (; Richard Rinehart, An Experience Base-A Boolean Typhoon (

From the Author to Procedural Authorship

It is arguable that all artworks are interactive. Lev Manovich (1998), for instance, argues that:

New media takes "interaction" literally, equating it with a strictly physical interaction between a user and a screen (by pressing a button), at the sake of psychological interaction. The psychological process of filling in, hypothesis forming, recall and identification--which are required for us to comprehend any text or image at all--are erroneously equated with an objectively existing structure of interactive links.
Procedural authorship, however, moves the author's traditional role over, so to speak, and without the reader/interactor, there is nothing authored except possibility. Simply put, procedural authorship makes the rules, which the reader en(inter)acts.

One simple way to explain this idea is with Ken Perlin's Wide World Web. ( This java applet produces an earth-like rendering. Every time it is launched, the world is a bit different, but from Perlin's point of view, every time, it's exactly what he intended. The authorship is under the hood, so to speak, in the algorithm. The "story" he has authored--a dynamic world--varies within appropriate parameters, which he has determined.

Another example of procedural authorship is baseball. Baseball is defined by both a set space and a set of rules. Every game, hundreds of thousands of them each year, is the same--set of rules. Yet every game is different. Each game has its own personality, based on the interactions of the players. And while some might call the game dull, for others it is a thrilling drama, precisely because the outcome is never known and different every time.

From Responsive Interactivity to Interactive Narrative

To return to Manovich, not every responsive program is a narrative.

To qualify as a narrative, a cultural object has to satisfy a number of criteria, which literary scholar Mieke Bal defines as follows: it should contain both an actor and a narrator; it should also consist of three distinct levels consisting of the text, the story, and the fabula; and its "contents" should be a "series of connected events caused or experienced by actors. Obviously not all cultural objects are narratives. However, in the world of new media, the word "narrative" is often used as an all-inclusive term, to cover up the fact that we have not yet developed a language to describe these new strange objects. . . . Thus, a number of database records linked together so that more than one trajectory is possible, is assumed to constitute "interactive narrative." But to just create these trajectories is of course not sufficient; the author also has to control the semantics of the elements and the logic of their connection so that the resulting object will meet the criteria of narrative as outlined above.

Hence, Paul Vanouse's Persistent Data Confidante, while an excellent example of procedural authorship--based on viewer ratings, secrets in the database can either die off or mate with other secrets and create new hybrid ones--is not a narrative, while his Consensual Fantasy Engine is an exemplary narrative whose trajectory is entirely determined by audience responses to a series of questions and the corresponding responses of a number of algorithmic "critics," which stitch together a narrative on the fly.

Consensual Fantasy Engine

Figure 8: Paul Vanouse and Peter Weyhrauch, Consensual Fantasy Engine, screenshot. (

Just because a procedurally authored projects meets the criteria for being a narrative doesn't mean that it is good, of course. Walter Ong (1982), however, has suggested that In the 1930s, scholars concluded that Homer, one of the greatest storytellers of the Western canon, probably used stock phrases according to context, using a simple replacement strategy. And it is such a strategy that Janet Murray, in her excellent book Hamlet On the Holodeck (1997), concludes Vladamir Propp validated.

When he finished all the extant titles, Propp was able to summarize all the variants of the Russian folktale in one inclusive representation. His work suggests that satisfying stories can be generated by substituting and rearranging formulaic units according to rules as precise as a mathematical formula.

From Interactive Narrative to
Machine-to-Machine Understanding

Beyond an interest in experimenting with innovative presentation methods, the other reason new ways of creating stories becomes so important potentially, is hinted at in a talk by Cliff Lynch back at the 1997 Museums & the Web conference. In a session on "Dublin Core and Warwick Framework Metadata for the Description and Location of Networked Information Objects" (, Lynch, the current executive director of the Coalition for Networked Information, suggested that one of the dominant issues facing institutions such as museums and libraries is the geometrically cascading explosion of information bytes, the majority of which will never be able to be processed by humans directly. We must figure out systems for machine-to-machine processing and comprehension of information.

Lynch was referring primarily to enabling search and retrieve functions, but the need can apply equally to creating rich information contexts, narrative stories, and responsive interactivity about and around the objects in museum collections. We are beginning to understand how to do this well in one-off situations, but it is truly a Sissyphean task and for every 6-month or year-long project completed, there are hundreds more waiting in the archives and dozens more possibilities being acquired on a daily basis. We will never catch up with the flow of facts to hand-knead them into coherent information environments that open up the possibility of increased knowledge by the user without additional tools.

Joseph Squier's new project, io (, uses an interactive engine to allow user input of a limited vocabulary to have a primitive kind of conversation and related dynamic "illustrations"--or vice-versa. Again, it is not hard to imagine using a similar function to understand better what people are looking for and translating it to the way that museums manage their information--a gap that can be very large and at times insurmountable.


Figure 9: Joseph Squier, io, screenshot. (

Richard Rinehart's new project, An Experience Base-A Boolean Typhoon (, is interesting for the way that it flexibly structures user input to create a self-generating database using data-multipurposing-friendly meta-language for describing life, an emerging proto-standard, which Rinehart has registered with the ANSI and ISO standards organizations, and this experience base is the first testbed project exploring the feasibility of such a language.

experience base

Figure 10: Richard Rinehart, An Experience Base-A Boolean Typhoon, screenshot. (

Ken Goldberg et al's Jester: The On-Line Joke Recommender ( is a collaborative filtering project that asks you to rate a series of jokes and then, based on your ratings, it recommends jokes "personalized" for your sense of humor. This kind of process, collaborative filtering, has proved successful in databases of movies, books, music, and other matters of "taste." It is a project/process waiting to happen at museums.


Figure 11: Ken Goldberg, Dhruv Gupta, DiGiovanni, Hiro Narita, Jester: The On-line Joke Recommender, screenshot. (


Amerika, M. "HYPERTEXTUAL CONSCIOUSNESS 1.0: an exploration into cyborg-narrators, virtual reality and the teleportation of narrative consciousness into the electrosphere..."

Appelbaum, R. (1996) "Activating Public Space to echo a museum's message and provoke multiple levels of response," In R.S. Wurman (Ed.) Information Architects (p. 150). New York: Graphis Press.

Besser, H. (1997). "Integrating Collections Management Information into

Online Exhibits: The Worldwide Web as a Facilitator for Linking 2 Separate Processes." Museums and the Web: An International Conference .

Biggs, S. (1998) "On Navigation and Interactivity," The Great Wall of China,

Blackaby, J. and Sandore, B. (1997). "Building Integrated Museum Information Systems: Practical Approaches to Data Organization and Access." Museums and the Web: An International Conference .

Canfora, L., Ryle, M. (Translator). (1990). The Vanished Library: A Wonder of the Ancient World.

Donovan, K. (1997). "The Best of Intentions: Public Access, the Web & the Evolution of Museum Automation." Museums and the Web: An International Conference .

Kay, A. (1998). In J. Abramson (Ed.) If/Then .

Manovich, L. (1998) Database as a Symbolic Form, Nettime.

Mitchell, W.J. (1992). The Reconfigured Eye. London: MIT Press, p. 52.

Murray, J. (1997). Hamlet On the Holodeck: The Future of Narrative in Cyberspace. Boston: MIT Press.

O'Donnell, J.J. (1998). Avatars of the Word: From Papyrus to Cyberspace. Cambridge, MA; London: Harvard.

Ong, W. (1982). Orality & Literacy: The Technologizing of the Word. London and New York: Routledge.

Schank, R. (1994) Engines for Education,.