The passage below by Phil Agre on disintermediation was buried in a Red
Rock Eater post from yesterday. The definition of "disintermediation"
is perhaps as elusive as the definition of "virtual community."
It's the point I was trying to make earlier today about the PITAC report
http://xent.ics.uci.edu/FoRK-archive/august98/0315.html
because in their vision of commerce, they are missing the concept of
intermediaries.
> What, then, do intermediaries do? Once we compare and contrast them,
> we discover that they do all sorts of things. They deal with the
> innumerable details that resist standardization. They hear about all
> of the pitfalls that their clients run into so they can warn you about
> them. They process, analyze, and synthesize information. They watch
> over the social environment for relevant issues. It varies a lot...
The rest of his rant is pretty interesting, bouncing from XML to EDI to
libraries in a quest to unlock the essence of disintermediation.
Includedn below if you're interested.
> [I forward the following from Phil <pagre@weber.ucsd.edu> hoping
> reassure people he has not been the victim of a coup or palace
> revolution, but is alive and well and living in Los Angeles.]
>
> =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
> This message was forwarded through the Red Rock Eater News Service
> (RRE). Send any replies to the original author, pagre@weber.ucsd.edu
> You are welcome to send the message along to others. For information
> on RRE, including instructions for (un)subscribing, send an empty
> message to rre-help@weber.ucsd.edu
> =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
>
> ...
>
> I poured hot oil on the whole concept of disintermediation the other
> day, and got several interesting messages in response. Rather than
> try to respond to these messages individually, let me expand on the
> point a little bit. The problem with the concept of disintermediation
> is not that nothing resembling it ever happens. Rather, like most of
> the concepts that we use to talk about the networked society, it tends
> to reify one end of a complex spectrum of phenomena.
>
> The problem is analogous to the problem with the concept of virtual
> communities. Amidst my railing against cyberpundits, I do make an
> exception for Howard Rheingold, who is a real intellectual. For
> one thing, he wrote his "Virtual Community" book way back in 1993 --
> and it still holds up well today. But the phrase "virtual community"
> is nonetheless disastrously misleading. Even though Howard knows
> better himself, the phrase tends to set up an opposition between two
> extremes: the mythical face-to-face community of American civic ideals
> where nobody even picks up the telephone, and the totally Internet-
> mediated community where nobody even knows where anybody else lives.
> In fact, in reality as we find it in 1998, most communities lie at
> points between these two extremes, and we only confuse ourselves when
> we try to use the exteme concepts to analyze any real case.
>
> And that's the problem with the concept of disintermediation. One
> could (and some do) try to define a concept like "pure intermediary",
> which is basically the kind of intermediary for which the theory
> of disintermediation applies. In the market context anyway, this
> is the kind of intermediary who does absolutely nothing except pull
> standardized information from sellers (or buyers) and transfer it to
> buyers (or sellers) with minimal processing. Even if such a concept
> could apply to some real case in the world, it is still misleading
> because it invites us to ignore all of the other stuff that the vast
> majority of real intermediaries do. Disintermediation does have the
> great virtue, in addition to the other sarcastic virtues I mentioned,
> that it identifies a class of social enemies that information
> technology can be employed to purge. But if our goal is economic
> efficiency, social justice, rational design, or honest social theory,
> then we need to throw away that drama and get ourselves a serious
> anatomy and physiology of intermediation.
>
> What, then, do intermediaries do? Once we compare and contrast them,
> we discover that they do all sorts of things. They deal with the
> innumerable details that resist standardization. They hear about all
> of the pitfalls that their clients run into so they can warn you about
> them. They process, analyze, and synthesize information. They watch
> over the social environment for relevant issues. It varies a lot, and
> it not terribly helpful to try to analyze the matter in the abstract.
>
> So to make matters concrete, let us spell out a particular scenario
> that proponents of the disintermediation theory are likely to approve
> of. That scenario involves XML. XML is cool. Perhaps the purists
> will cringe at me for putting it this way, but it's a cross between
> a document markup language and a database notation. XML documents
> come in definite, predefined types, and each type has its own set
> of specialized tags. Someone could write (and probably already has
> written) a "DTD" (document type definition) for resumes. People who
> are looking for work could mark up their resumes using the relevant
> tags, put them on the Web, and make them visible to some Web crawler
> that indexes XML pages. Then an employer could search for suitable
> job candidates by typing some keywords into a resume-specific search
> engine. The search engine could probably be constructed automatically
> from the DTD. Presto, no more want ads, and not even any more "job
> boards" on the Internet. Likewise, industries that produce heavily
> standardized goods like electrical or mechanical equipment, where
> any product can be almost completely characterized using a fixed list
> of parameters, could get together to create DTD's for their products.
> Then instead of going through market-makers, they could simply publish
> their catalog online as a set of XML pages, and anyone wishing to shop
> for such goods could simply type in a search request at a generic XML
> search engine.
>
> That's the scenario. I didn't invent it; quite the contrary, it's
> the conventional wisdom in many quarters. So what's wrong with it?
> Not that much, so long as you stick to the cases for which it works.
> The key is standardization. If the task at hand is simply matching
> "wanters" with "havers", and the match can be defined in standardized
> ways, then XML provides a generic platform for building the necessary
> tools. The deep principle here is the symbiotic relationship between
> information technology standards and standards out in the big world.
> The more information and communication technology you have, the more
> incentive you have to standardize things like agricultural goods,
> mechanical parts, computer programmers, etc etc. And, reciprocally,
> the more standardized things are in a given social world, the more
> incentive that world will experience to adopt another round of
> information and communication technology. That doesn't mean that
> things necessarily *will* get standardized -- standardization can be
> costly. But it does shift the balance.
>
> To see this, think about two more examples, EDI and libraries. Read a
> brilliant article by Eric Brousseau, "EDI and interfirm relationships:
> Toward a standardization of coordination processes?", Information
> Economics and Policy 6(4), 1994, pages 319-347. It's about the reason
> why EDI (electronic data interchange) has stalled, and secondarily
> why electronic commerce faces unexpectedly steep challenges. EDI
> was supposed to integrate companies more closely with their suppliers
> by creating an open channel for orders and other associated data.
> The problem, as Brousseau points out, is that relationships between
> companies and their suppliers vary a great deal in their stability.
> In some cases, the relationships are very stable and predictable.
> Orders don't vary that much, and it doesn't take much data to specify
> them. In those cases nobody needs a complicated, expensive computer
> system to send and receive orders; the telephone is plenty, or just a
> standing order with fixed quantities. At the other extreme, as in the
> construction industry, things are so unpredictable that orders cannot
> be specified over a data line using a moderate number of parameters.
> Once again the telephone is required, or face-to-face interaction with
> complicated paperwork.
>
> EDI, like many technical systems, requires fairly large economies of
> scale before it becomes affordable -- that is, a lot of companies have
> to adopt it so that the costs of developing it can be distributed.
> And that never happened because it was simply not useful enough to
> enough companies. Electronic commerce on the Internet might be more
> successful because it unbundles several layers of functionality that
> EDI systems enclosed in one complicated package. The bottom layers
> -- the functionality of the raw Internet before any applications are
> installed -- can be applied to many, many purposes. As a result, many
> firms have multiple incentives to connect to the Internet, and the
> fixed costs of developing the Internet can be spread so widely that
> nobody even thinks of them as costs. The pattern can then repeat
> with all of the other elements of electronic commerce functionality,
> layer by layer. In each case, however, the story is the same: the
> technology is useful when it fits in with standardized elements of
> the institutional world around it. The more such elements one can
> identify, the more value the technology delivers. And as the payoff
> from standardization increases, the incentive to standardize things
> increases as well.
>
> Libraries epitomize another pattern: stuff in the world that is sort
> of standardized, but not standardized enough that all of the effort
> can be pushed to the margins of the system. Library collections
> encompass an amazing variety of stuff, in hundreds of languages, in
> all kinds of formats, with all kinds of exceptions and quirks and
> details, and librarians have spent centuries refining methods for
> cataloguing the stuff. To make the administration of their catalogs
> as efficient as possible, the librarians keep using information
> technology to push the effort, so to speak, upstream. The OCLC
> database in Ohio, for example, pools much of the world's cataloguing
> effort. Publishers are asked to prepare rough drafts of catalog
> entries. Individual libraries customize catalog entries to their
> local purposes. Each record is the result of a whole elaborate
> division of labor organized by a sort of monastic cult that lives and
> breathes the very detailed standards that make the results consistent
> and thus useful.
>
> These people, the catalogers, are intermediaries in the vague sense
> that they stand between authors and readers. (Another problem with
> trying to apply the concept of disintermediation to everything is that
> you're tempted to interpret everything as an intermediary, whether
> that's a useful way to talk about it or not.) You're not going to
> disintermediate them. Instead, what needs to happen is happening:
> the people themselves keep reshuffling and reorganizing the work.
> We can learn a lot from them. See Chris Borgman's forthcoming book
> for details.
>
> That's not to say that the whole publication system should stay the
> way it is. Just about everybody in the research world is longing
> for the day when we can get rid of commerical journal publishers who
> acquire intellectual property from research institutions for free and
> sell it back to them for thousands of dollars at a time. This won't
> happen, however, until the value that those intermediaries add to
> research publications gets added someplace else in the value chain.
> That is why most of the best innovations in journal publishing
> are happening in technical and scientific fields. ARPA and NSF,
> visionaries that they are, have thought hard for many years about the
> infrastructure of scientific and technical research, and they have
> encouraged the use of standard tools. That includes spectacular tools
> such as the Internet, but it also includes mundane tools like the
> LaTeX document formatting system. If everyone who publishes technical
> information uses LaTeX then it becomes much easier to run a journal,
> and much of the value that publishers add (typesetting, copyediting,
> and so on) becomes unnecessary.
>
> Is this disintermediation? Again, you can stretch the concept to fit.
> It's much more useful, however, to map out the reshuffling division
> of labor that becomes possible as the rising tide of standards makes
> information technology more useful. We'll still have intermediaries
> in research publishing, but their functions will be different. Their
> functions should be moved away from commercial firms to university
> consortia and professional organizations so that intellectual property
> and other contractual constraints don't motivate the intermediary to
> slow down the transition. (If a journal editor decides to move his
> or her whole journal and editorial board from a commercial publisher
> to a nonprofit organization overnight, a lawsuit is likely to result.)
>
> It's a trade-off: as we move those functions around in the system, we
> lock ourselves into a whole new set of standards. Those standards may
> facilitate compatibility and cooperation and efficiency, but precisely
> for that reason the standards themselves may become hard to change,
> and the fixed costs associated with adopting the standards may tend
> even more conclusively to marginalize those institutions, for example
> in less developed countries, that cannot afford them.
----
adam@cs.caltech.edu
My Gorgonite brothers are doing what the Gorgonites do best. They are
hiding.
-- Small Soldiers