Why Multimedia Publishing is a Crock

(Appeared in Whole Earth Review, Winter, 1991.)

Many computer hardware and software companies are attempting to develop a new, computer-based medium, which has been tagged “multimedia.” Most of these systems use mass produced optical discs to store information, which you might someday buy at the corner audio or video store like tapes or audio CDs. This scenario is called multimedia publishing.

The purpose of any medium is to send messages. Since consumers of today already live in an information-saturated world, these new messages compete for a scarce resource: time. Messages compete for time in two ways: salience and production value. Salience is the relevance of the message to the consumer's personal interests. Production value is the amount of attention-getting media "heat" embodied in an message.

To give two examples: A car commercial on network television is likely to have extremely high production value - fast cuts, lots of action, and fancy graphics - but its salience to most consumers at the point of viewing is low. The production value of a three-line newspaper classified ad for a used motorboat is very low, but its salience to those who purposely turn to that column in the newspaper is high.

While both factors might be maximized, economic constraints usually encourage a trade-off between salience and production value. For example, targeted direct mail is a way in which advertisers pay to optimize salience. Very few direct marketers go the next step and also increase production value by, for example, mailing video tapes instead of print.

The fixed nature of current media emphasizes the tradeoff of salience and production value. For instance, network television has been inherently broadcast. With no means to optimize salience, the uniform trend in television is toward attention-getting high production value. This is seen even in news programs, which, though timely, are limited in their salience to any particular viewer.

Reader Relevance

Information which is optically published - compiled, edited and then pressed onto a compact disc - has limited currency and novelty due to the time taken to prepare and ship the material. Also, while the presentation order may be varied (by hypertext, search, guides and so forth), the underlying content always remains the same, and the portion relevant to a particular consumer will be exhausted at some point. These are two ways of saying that in spite of claims of “interactivity,” a medium based on optical publishing is inherently limited in salience.

Books, magazines, and newspapers have had similar limitations for 500 years and have done quite well. But multimedia has some further obstacles to overcome: the lack of portability of most players and the time to swap discs limit the "ready-to-handness" of the medium in comparison to books. A multimedia “reader” device must be purchased before any discs can be used. Since a new medium must be better than existing media in some respect to justify the initial purchase, multimedia systems have to find a way to compete in spite of these limits.

But optical publishing is locked into the wrong basis for competition. Since it cannot further improve salience to the end user, it must hike production values to compete. The medium must become capable of accepting these higher values and rendering them for the consumer, hence the push for full-screen motion video, rich stereo sounds, faster animation rates. But this leads to a series of problems:

• Today's consumer priced hardware is stretched to its limits to deliver acceptable video, graphics, and sound. This keeps system costs high, and will lead to frequent model changes as manufacturers try to push performance. This makes it harder to create multimedia content with any shelf life, and may leave consumers stuck and frustrated with expensive, incompatible equipment that keeps going obsolete.
• Greater production values are only purchased with greater production costs. This is a barrier to entry in the multimedia content marketplace. This cost structure favors existing “infoglomerates” who have the capital to make the initial investment, rather than smaller companies which might produce novel content. These large information owners are encouraged to hedge by publishing titles of “broad” interest, which in turn have reduced salience to any particular consumer.
• The characteristics of the resulting medium make it hard to make use existing content. Many previous new media, such as CD audio, were able to reuse existing material with minimal conversion. Some, such as videotape, allowed both conversion and the fresh creation of personal content by consumers. Nonwritable optical media prevent the latter option. Straight conversion of linear video or traditional book editorial structures to optical media gains little competitive advantage, resulting in a more expensive version of content that was already accessible. We are still working toward interactive media forms that will be compelling (see the 'Storytelling' section, WER#71), but it seems clear that simple copying or automatic reassembly of existing content has limited appeal.

To make multimedia publishing take off, there has to be a positive feedback cycle of hardware availability, content availability,and consumer purchases. Each of these problems puts a damper on that cycle. Several billion dollars have been spent to date trying to get the feedback going, with meager success and little examination of the underlying structural problems.

Competing with Salience

If we cast aside the current limits of optical media, we can think about optimizing salience, rather than production value, in a new medium. A medium which is designed to support salience should be:

• Timely: This means online, real-time data, instead of information that is batched up for publication.
• Convenient: The delivery mechanism can’t be chained to a desktop or living room - it needs to be easily portable. Startup and user interfaces must be fast and simple, to compete with books and the Walkman. This requirement plus timeliness yields a need for continuous, remote connection, probably by spread-spectrum radio modems.
• Personalized: Personal relevance is the greatest component of salience. This requirement can be satisfied through content based searching in text, combined with the use of “agent” software which continuously scans available data for relevance and routes useful bits to the consumer.

Note that these goals could be accomplished for most applications with text only. Motion video and high quality sound can be added, but where they satisfy the need for use values in the content they convey, rather than production values. An example is voice mail and annotation, which is casual in quality but usually relevant to the recipient.

As a corroboration, consider the one overwhelming success in wide area networks: Usenet. Here's a network that's text only and often comes with a lamentable user interface. Yet it has achieved world-wide penetration and sustains numerous invisible work and social communities, because it allows anyone with access to contribute, and is structured into discussions which allow readers to shop for salient topics. It competes effectively for the time of its audience.

The apparent cost of these information transactions for most Usenet clients is effectively zero. The Usenet community exists only at the sufferance of the industrial, academic, and governmental bodies that subsidize the net. How might one create a widespread network with Usenet’s benefits, but without the necessity for subsidy? To figure that out, we should know what people value about information.

Value in Information Transactions

People are willing to pay for two aspects of information transactions, which I'll call "push" and "pull," borrowing from marketing jargon. The most obvious form of a push transaction is broadcast network advertising. The advertiser pays to push out his information at the audience, whose viewing is totally subsidized by this activity. Trade publications and free "advertiser" newspapers are similar examples.

In a pull transaction, the consumer pays to acquire information of perceived value. Book buying is the oldest example, but purchases of video- and audio- tapes fit the same pattern.

Mixed systems are common. For instance, consumer magazine revenues commonly consist of two parts advertising to one part subscription. The readers pay a third of the freight for getting a perceived useful slice of information, the advertisers pick up the rest in consideration of the audience delivered by the publisher. Newspapers' revenues are likewise split between advertisers and subscribers.

The characteristics of existing media make it difficult to change a given system's position in this value spectrum. It would be impossible for a magazine to be subsidized trade one month, fully subscriber paid the next, and mixed the third. A television channel might be advertiser-subsidized, “public,” premium cable, or pay-per-view, but it won’t change overnight. There is even less possibility of a consumer altering the nature of his or her relationship with the system on an individual basis.

Networks as a Medium

Today’s commercial computer networks can also be positioned on this spectrum. Dialog is a pure pull network - it delivers very salient information at a high cost. Compuserve is a pull model at the consumer level. Prodigy is the first mixed-model network - the users' access is partially subsidized by the push value of the advertising scrolling across a portion of the screen.

In existing consumer systems, such as Compuserve and Prodigy, the network owner is the only authorized vendor of information. That is, users may post messages, often through a moderator, but they do not receive value when others read them. All revenue goes to the system owner. In some cases, technical experts may receive subsidies or even royalties for their participation, but these arrangements must be specially negotiated with the system owner. Messages which might be construed as advertising are forbidden on most systems.

These practices add up to centralized control of the means of information production, and a complete lack of market interaction among the users. These information networks effectively embody state socialism, and like the states of Eastern Europe, they have failed to create viable economies.

Like traditional media, these commercial networks have fixed themselves, their users, and all their transactions ito one point in the value spectrum of push and pull. There is no structural reason why this must be so in a distributed electronic medium. Existing database and transaction-management systems are fully capable of mediating value exchanges among myriad parties - Visa and Mastercard do it every day.

The Network is the Market

A consumer information system needs to be a fully functional marketplace. Each user, individual or corporate, should have the freedom to buy and sell push and pull information values. For instance, as a user I could choose to wholly subsidize my pull habit by letting others' push advertising take over three-quarers of my screen. I could publish a useful database online, and use part of the revenues from user's pull of that information to subsidize some push to others who might then become subscribers. I might remain effectively anonymous, or I might choose to make public some aspects of my information and material consumption habits so that relevant "pushers" would bid higher for my screen space.

The essence is the reduction in transaction costs between information value buyers and sellers. This requires arranging rendezvous between pushers and pullers in a manner more efficient than competing media. The technology which makes this efficiency feasible is indexing (a way of structuring informaton so the right information is more easily found by those who seek it), search (a way of sifting through the information to find themost immediately salient bits), and agency models (software that has information about the requirements of an individual user) that match the information needs of the pullers against the information being provided by the pushers. The competitive basis of such a system is optimized salience.

The smart network medium should be able to "single-cast," that is, arrange one-to-one transmissions, compared to conventional media which are broadcast or narrowcast only to a degree. Advertisers (pushers) should be willing to pay more for transmission to consumers who are self-identified targets. For instance, a local auto tire store would presumably pay more for exposure to users who had posted interests in new tires than it would for an ad in the local newspaper which reaches many with absolutely no interest.

The potential spread between value of information to sender and receiver creates opportunities for arbitrage by the system owner. If a consumer is willing to pay to pull information on, say, tires, the system might know that there was a client willing to pay to push that same information. It might split the difference and transmit the information free, but it could also act as intermediary broker and keep part of the spread to support the network and earn a profit. Arbitrage plus transaction fees should suffice to keep a system going without external subsidy.

The ground for competition among networks becomes the success of a system owner at creating indexing and profiling software which matches the needs of its clients, at maintaining a appropriate balance of privacy rights against guarantees of information quality and delivery, and increasing the efficiency of the information marketplace over time.

How Do We Get There?

Where would such a network come from? While the current Internet is a harbinger, it’s too hard to use, is tied to desktop machinery, and has the wrong economic structure. One of the requirements for progress is more bandwidth, which enables users to send higher volumes of information in shorter periods of time. The proposed NREN (National Research and Education Network) high-bandwidth backbone (powerful computers connected by high-speed communications lines) has the problem of being overtly subsidized and controlled by government, which is unlikely to lead to a market structure — and it seems unwise to entrust the bureaucrats that gave us NASA and DOE with the potential core of an information society.

Many parts of the physical infrastructures that could be employed for a market network do exist; they are owned by the regional Bell companies, the long-distance carriers, and (to some extent) cable TV operators. The Bell companies may allowed to enter the information services marketplace within the next year. At the same time, they should be prevented from using money from voice (monopoly-market) consumers to gain unfair advantage in the information market. The market network may be a way of setting up the appropriate incentives, providing the Bell companies were required to act as common carriers for all comers, in addition to being buyers and sellers of information on their own account.

An intriguing possibility is that such a network could be created without laying wires and optical fibers. Cellular telephone technology is already so cheap that it has been proposed as an alternative to a wired network in the former East Germany. Petitions have been filed with the FCC for bandwidth for two-way radio connection of portable computers, and hardware firms are scrambling madly to bring such devices to market. The success of regional public bulletin-board systems, like the WELL, indicate that it may not be necessary to build an national network to begin bootstrapping an information marketplace. If this vision works out, the new medium could be coming to your neighborhood, in the air, sometime soon. The year 2000 seems a reasonable guess, but favorable breaks on public-policy decisions, technology advance and venture funding could move the date even closer.