Why did we build cyberspace like that?
Stefik's collection of essays on the Internet are good for your brain
Where did the ideas for e-mail, searchable digital libraries, electronic commerce, and AOL chat rooms come from? Why are they successful? In his book Internet Dreams: Archetypes, Myths, and Metaphors, Mark Stefik suggests that it's not just because some programmers implemented these technologies. It's because humans have attached importance to these ways of communicating for a very long time. This fascinating book will make many people think afresh about what we are doing online. (3,500 words)
This book is not an easy, breezy, commuter-train read, as last month's The Friction-Free Economy was. It is a collection of articles, speeches, and other writings, assembled by Stefik, a scientist at Xerox PARC. Stefik's commentaries before and after each article make Internet Dreams ideal as a textbook for a challenging college-level course called, say, sociology of the Internet.
Deep thoughts on technology and society
Better, then, to admire the ideas themselves. Internet Dreams's main theme is one that has occasionally been addressed by shallower thinkers, with trivial results. The theme is: How do network communication technologies reflect or influence our humanity? Where other writers have answered this question with inanities like "Duh, it's a return to tribalism," or by talking about various forms of fantasy-filled cyber-utopia, Stefik tries to understand how we use technologies to enhance the way we live and work, not simply for the sake of mental onanism or entertainment. His particular emphasis, as a scientist, is on how scientists can use information to help them get their work done.
Stefik chooses four metaphors of information use: digital libraries, e-mail, electronic commerce, and virtual worlds. There is a section of the book devoted to each. He introduces each section by talking about the "deep structure" of each metaphor -- e.g., what makes a library a library? Is it merely a place for books and magazines? No, it's also a community center and a place to get advice, and it uses a very particular way of classifying information.
The success of each section is inversely proportional to the maturity of the technology. The section on e-mail is the weakest: all of the observations about how and why we use e-mail seem obvious. Some of the selections in this section strike me as the work of academics who come to conclusions and then write them up in journal articles without bothering to check them out in the real world. (Rather like some work in computer science research, which describes so-called innovations that some startup company implemented two years ago and then tanked, or that Microsoft included as "dumb little features" in last year's version of Word or Excel.)
The sections on digital libraries and electronic commerce are the strongest. Perhaps this is because these technologies are still evolving. But I should admit that these topics are also nearest and dearest to my heart: I have spent the past few years of my professional career dealing with digital libraries and with the type of electronic commerce on which Stefik concentrates: commerce in bits (digital works, as Stefik calls them), not atoms.
Digital libraries then and now
Digital libraries exist today, to be sure, but not on the scale or in the way that the authors of most of the articles in the section called "The Digital Library Metaphor: The I-Way as Publishing and Community Memory" would want. The idea of the computerized library is a utopian one: a vast repository of knowledge available instantaneously on one's desktop. According to Stefik, the idea of the library itself dates back to ancient times, when various cultures mythologized the idea of the keeper of knowledge.
The first articulation of something resembling a digital library was in 1945, in an article that is, in retrospect, one of the most important pieces of technology vision ever written. "As We May Think" was written by Vannevar Bush, who headed a precursor of the National Science Foundation during World War II. The article, excerpted here, introduces the memex, Bush's concept for a mechanical device built into a desk that implements a personal information library. Although it uses microfilm and does not have a connection to the outside world (remember, this was 1945), the memex not only bears uncanny resemblance to a Web browser but also seems, the way Bush describes it, eminently implementable.
This book follows Bush's article with another fundamental writing on electronic libraries. Stay in Washington and jump-cut from 1945 to 1988 for the coinage of the term "digital library" by Bob Kahn, co-inventor of the TCP/IP protocol and by then (as now) head of the nonprofit Corporation for National Research Initiatives (CNRI). He and Vint Cerf, co-founder of CNRI and now a senior executive at MCI, wrote an entire book on digital libraries. From this, Stefik excerpts a chapter on the concept of the knowbot. A knowbot is an autonomous information agent, which Kahn and Cerf describe as the way to make vast digital libraries truly useful. In many senses digital libraries already exist. Although the phrase "digital library" is catchy, the concept is quite complex -- analogous, say, to "public transportation" -- and, therefore, means different things to different people. This lack of uniformity has resulted in a deficiency of uniform technical infrastructure, such as search protocols, that make more advanced technologies like knowbots possible.
While their technical details may be the same, digital libraries often serve different purposes for different builders. To a commercial media company, a digital library is a "content management system:" an organized collection of all its publishable assets, from which the company can draw to create a wide variety of revenue-producing products and services. To a traditional public library, a digital library is a way to open up the world's knowledge to the public.
|Perhaps libraries will eventually reemerge as some kind of institutionally sponsored information source for the economically disadvantaged -- a sort of "information food stamp" service on the Internet|
The conflict between these two points of view manifests itself quite simply. If your neighborhood public library buys a published work and puts it up on a Web site for the world to access, then -- brushing aside questions of how easy or how much fun it is to read things on Web browsers -- why should anyone else ever pay for a copy? Furthermore, does that library really have the right or responsibility to use a world-accessible medium to distribute information?
At the same time, the Internet is causing the publishers' and libraries' points of view to converge, because it almost forces them to impinge on each other's territories. Where should the boundaries be drawn? Another of the articles in this book, an excerpt from "What is the Role of Libraries in the Information Economy?" by the British author John Browning, offers answers to that question based on pragmatic reality rather than on abstract notions.
Commercial Web sites provide a reference point for these answers. We're currently seeing a model for content access on commercial Web sites emerge in which basic content is free to users, while premium content and services are available only in exchange for user registration and/or payment. Libraries' electronic services are converging on the same model: basic services are free, but libraries are beginning to charge for premium services. Libraries are also, through their own Web sites, beginning to resemble content formatters and repackagers, just like commercial media companies. Meanwhile, more and more commercial media companies are letting users search through large collections of assets on their Web sites.
The implications of convergence between the "business models" of commercial media companies and libraries are interesting. It's likely that the two camps will remain at loggerheads for some time. Perhaps libraries will eventually reemerge as some kind of institutionally sponsored information source for the economically disadvantaged -- a sort of "information food stamp" service on the Internet. That scenario is possible if it's based on technology that Stefik introduces in the section on electronic commerce, in an article he wrote himself called "Letting Loose the Light: Igniting Commerce in Electronic Publication." This is the article that introduced me to Stefik's work when I read it three years ago. Actually, it reintroduced me to Stefik: I had known about his work with Daniel Bobrow on object-oriented programming languages when I was an objects geek back in graduate school. But I had no idea that Stefik had gone in the research direction he currently pursues.
Selling bits with bits
"Letting Loose the Light" is, like Khan's piece on knowbots, a visionary technology prescription whose implementation will take time to realize. It seeks to provide a technological basis for commerce in digital content in a way that walks the tightrope between ease of access and protection of the author's interests.
Currently, content creators who want to publish electronically and get paid for their work have few alternatives when it comes to copyright protection. Let's say I wanted to sell one of my books in digital form on the Web. At the simplest, I could create some Web pages that collect payment from a user and then let her download a regular file in a format such as Adobe PDF. The problem with this, of course, is that once someone has that file, she can make many copies of it.
A slightly better alternative would be for me to encrypt the file, using something like the Pretty Good Privacy (PGP) public-key encryption scheme. This requires that the user know a private key -- essentially a password -- to get at the file, but otherwise it does little good, because once the file is decrypted, it is once again a plain file that can be copied at will. In other words, simple encryption protects the file from unauthorized access on my Web site, but after any user has decrypted it, it's fair game for anyone.
Stefik's scheme for trusted systems offers a better way. A trusted system is a sort of "black box" that allows access to its contents only through certain strictly defined operations. For those of you familiar with object-oriented programming, you can see how Stefik got from object-oriented LISP to trusted systems: a trusted system is nothing more than an object. A regular file is not a trusted system, because users and programmers have full access to its raw contents and can do whatever they want with it. A trusted system that contains digital content for use on a personal computer might have only three access methods: view, print, and copy. View opens an on-screen viewing program on the content; print sends it to the printer, and copy, rather than creating a perfect copy of the file, creates a clone of the trusted system.
There are no other ways to access the content. Furthermore, each such access may have a price, which the user could pay automatically through some digital payment scheme. There may also be an access method called lend, which is like copy but with a difference: the clone it creates has a limited life. During that life, the original trusted system does not function. At the end of the clone's life, the clone becomes inaccessible and the original comes back on. (The borrower may then also get an offer to buy her own copy of the trusted system.)
The greatest advantage of the trusted system scheme is that it ensures that each transaction results in payment to the author, without breach of copyright. It is also relatively easy to use -- although, admittedly, not as easy as an unencumbered file.
Trusted systems: the drawbacks
From a practical standpoint, however, trusted systems have a few drawbacks. First, there are some technical difficulties in implementing a trusted system that works in a standard computer. There would be too many opportunities to break the trusted system's security: for example, one could easily write a print driver that, instead of printing, captures content and puts it in a plain file. Stefik acknowledges this and says that the best way to implement trusted systems is with separate hardware devices. Such schemes (remember "dongles?" Didn't think so) have found very little acceptance with average consumers, yet there are several vendors working on software-only versions of trusted systems. They do not yet possess all of the attributes that Stefik specifies, but the technology is getting there.
The second drawback relates to the fact that trusted systems are a transaction-oriented commerce paradigm. Currently there are two problems with transaction-oriented electronic commerce: First, your device must be connected to the Internet in order for payment collection to work. Second, the cost of processing each transaction can be prohibitively high, leading to the need for "microtransaction" support.
Trusted systems and the "hunter-trickster"
These problems will surely be overcome in time. The world eventually will move to a transaction-oriented paradigm for selling content online. Then, content creators and other rightsholders will be compensated per transaction, (i.e., each time someone accesses a piece of content). Stefik's paper is a blueprint for this type of scheme. It appears in the third section of Internet Dreams, the one called, "The Electronic Marketplace Metaphor: Selling Goods and Services on the I-Way."
Another highlight of this section is "Slaves of a New Machine: Exploring the For-Free/For-Pay Conundrum," excerpts from an unnervingly prophetic 1994 speech on Internet publishing by Laura Fillmore, the owner of the Online BookStore. The Online BookStore was a much-watched but notably unsuccessful attempt, before the rise of the Web, to sell downloads of books in plain ASCII text format from an FTP server. In her speech, Fillmore bemoans the lack of applicability of traditional publishing models (such as making entire books available as a unit of publication) to the Internet. She admits that a large problem is the dismal readability of ASCII text. But at the same time, she predicts models that would ultimately be successful online, such as content with user participation (e.g., chat groups), sponsored Web sites, and using the Web as a tool for promoting physical products (like books).
Stefik is, in effect, disagreeing with Fillmore. His trusted systems idea implies that traditional publishing models, including digital distribution of complete book-length works, will be viable as long as connectivity is ubiquitous, microtransaction support is available, and electronic formats can be as readable as physical ones. Whether he turns out to be right or not is yet another fascinating question raised in this book.
Stefik's primary purpose in selecting the electronic marketplace metaphor is to show how it lines up with the human "hunter-trickster" archetype, which was prevalent during times when man was nomadic and his main problem was finding food for the night's meal rather defending his land against invaders or doing some invading himself. The hunter-trickster archetype, Stefik says, is more appropriate to today's network-connected world, in which cleverness, adaptability, and speed are necessary ingredients to success -- they translate to flat organizations, entrepreneurial attitudes, and "virtual corporations" in today's business world. Contrast this with the older style of large organizations run by command-and-control hierarchies, a paradigm that is more in line with the "warrior" archetype that held sway during times of farmland and territorial conquest. Another article in this book, "Electronic Markets and Electronic Hierarchies," by a trio of MIT Business School professors, makes these differences in paradigms explicit in their application to modern business practices.
The final section of Internet Dreams is on so-called virtual worlds: metaphorical places that people can inhabit in cyberspace. This chapter is about two things, neither of which are the emerging idea of virtual worlds with "avatars" as graphical representations of the worlds' inhabitants. When this book was put together, that technology didn't really exist. Instead, the virtual worlds section spends a lot of time on MUDs, or Multi-User Domains, which are the more interesting text-only precursors of avatar worlds. In fact, one of the articles actually says that graphics would do nothing to enhance the overall effect of digital worlds. The other topic addressed in this section is that of tools for scientific collaboration, such as networked digital whiteboards. Two long articles describe the evolution of a civilization, of sorts, in a MUD called LambdaMOO (the MOO stands for "MUD, Object-Oriented", while Lambda is a reference to the LISP programming language in which the program was implemented).
One of them talks about a "rape," in which a woman was "sexually assaulted" by another denizen of the MUD who happened to be a sociopath. The sexual assault came in the form of actions that the sociopath made MUD denizens perform on the woman's online persona, by means of devious MUD hackery. I found the arguments in this article self-absorbed and utterly unconvincing. The discussion, of course, was about to what extent the woman was actually raped -- i.e., to what extent she (or her online persona) could be said to have inhabited the virtual world. She felt greatly offended -- rightly so -- and called for the "rapist" to be punished. But the discussion of whether she inhabited some world or not is completely beside the point. Someone broadcast disgusting sexually explicit messages involving her. Why is this significantly different from publishing the same messages in a magazine, or for that matter, e-mailing them to hundreds of people? I don't see how one's connection to a virtual community differs, in this context, from one's connection to a community implied by a magazine or group of e-mail recipients. The question of offense or punishment in the MUD scenario is simply an issue of pornography versus free speech. This is a debate that has taken place for decades outside of the cyber-world.
Moreover, I don't see the point of MUDs, except as escapist entertainment for the socially inept -- despite Stefik's noble disavowal of this. He sees virtual world technology as being potentially helpful to people in practical ways, such as his experiments with collaborative electronic whiteboards, which are also included in the book. Networked digital whiteboards are now available as commercial products. They're very cool.
So, as a whole, is Internet Dreams. This is one of the most intellectually stimulating books I have read in a while. If it ever gets accepted as a college-level textbook -- as well it should -- and it comes time to create a second edition, then I urge Mark Stefik to make two important changes: First, mention explicitly when each article was written in the prefatory material to the article. Don't make the reader look it up in the Sources list at the end of the book. Dates would help contextualize the ideas in time, so that even if they describe technology that is trivial by today's standards, the reader can be encouraged to appreciate the ideas for their intrinsic worth. Second, please cut down on the length of your own intros and outros. Often they are much longer than the articles themselves, and at times they belabor the relatively obvious. Let the elaboration come from the professor teaching the course and from the lively discussion students will have in the classroom. And in the library and coffee shops on campus. And in the term papers they will write. And in the startup companies they will go on to build. And all the other high-quality thinking that this remarkable book will surely engender.
Title: Internet Dreams : Archetypes, Myths, and Metaphors
Author: Mark Stefik
Publisher: MIT Press
List price: $15.00
About the author
Bill Rosenblatt is market development manager for media and publishing industries at Sun Microsystems. Reach Bill at firstname.lastname@example.org.
If you have technical problems with this magazine, contact email@example.com