words of wisdom from Phil Agre

From: Strata Rose Chalup (strata@virtual.net)
Date: Mon Jan 15 2001 - 12:26:05 PST

Snipped this from a recent RRE, very topical bits for this crowd. Most
of the P2P crowd is essentially concentrating on new channels, but some
of the folks here have companies which are making enabling technology.
What's the key factor? The technology must respond to economy of scale
and enable its users to leverage their economy of scope. Think about
various firms technology ploys in that context, and some of the things
that are wowing people on this list don't look so good (Groove, for
instance), whereas some sleepers have excellent prospects.

The stovepipe effect and platform cycle that Phil mentions is another
thing to factor in. Dave's recent collaboration and hatchet burying is
a great example of how, possibly without thinking of it explicitly,
visionary technologists route around the stovepipe effect by extending
their features out into the rest of the protocol world. You go, girl!


----begin RRE News excerpt, Phil Agre author, fwd with attribution----

We're laughing at the "new economy" companies that are trading at 15
cents a share right now, and justly so. But mixed in among the downer
cows, I suspect, are some legitimate ideas that are simply premature.
Here are a few reasons why that might happen:

Scale. Most information technologies huge massive economies of scale.
They require substantial one-time investments, like writing software,
that must be paid off across many customers. Economists often pretend
that prices are driven by the marginal cost of production, which in
the computer world are usually low, but in scale-intensive industries
prices tend to be driven by fixed costs. And fixed costs tend to
be counterintuitive. If a company with fixed costs of X can find
N customers who are willing to pay X/N, then they can at least break
even. These could be 10 customers who are willing to pay X/10, or
1,000,000 who are willing to pay X/1,000,000 -- any N will do. If
the product can be released in several versions that justify charging
different amounts to different customers then you can combine N1,
N2, and N3 to get the cash flow you need. But in many cases those
N customers don't exist yet, for any N. Either enough people aren't
online yet, or the product requires complements (such as personal
computers) that are themselves still too expensive. As a result, lots
of good ideas have to wait until the market builds sufficient scale.
This idea takes some getting used to: one day your worthless business
plan will become valuable, but that day will not be announced by any
thunderclaps -- the market will simply and silently have grown big
enough. Since most of these markets are natural monopolies, again
because of their high fixed costs, the game is to predict when scale
will be achieved. Get started early enough to be positioned when it
comes, and late enough so you don't run out of cash while waiting for
it. The bigger the prospective monopoly, the greater the incentive
to burn cash for the sake of the first-mover advantage. That was the
theory behind a lot of the .com start-ups. They got it wrong for the
most part, but their investors believed it long enough to put a bunch
of Ferraris on I-280. Maybe next time they'll get it right.

Scope. Once upon a time, giant media mergers were always justified
economically in terms of "synergy". If you paid $50M for a media
company that only seemed to be worth $30M, you would claim that the
deal isn't just ego, and that the combined company would be valuable
enough to make up the difference. These claims got discredited in
sufficient numbers that nobody utters the S-word anymore, but that
obscures the fact that a real economic principle was at stake. This
principle is called economies of scope. Whereas economies of scale
are efficiencies that can be gained by producing large numbers of the
same thing, economies of scope are efficiencies that can be gained by
using the same facilities to produce several different things at once.
Economies of scope were first discovered in the early 20th century
by the chemical industry and other businesses that could use the same
equipment to combine the same inputs in different ways. If a company
produced only one chemical, then it would lose in competition with a
company that could produce several chemicals, distributing its fixed
costs among all of them. The same thing is true in media industries.
The effort that goes into creating a cartoon character or reporting
a news story can be turned into products in several different media.
This is why the Internet is not going to revolutionize the economic
structure of the media industry: the Internet is one more channel for
the distribution of synergistic products. Someone who produces media
content only for the Internet must compete with people who use the
same facilities and effort to produce media content for several media.
A company's business plan can be premature because the technology or
distribution channels do not yet exist to capture economics of scope.
A concept might only be viable if it is turned into both a Web site
and a cable show, but if not enough people have digital cable yet then
the concept will have to wait. Some concepts may require distribution
channels that are so numerous that we'll just have to wait for them
all to be invented.

Stovepipes. Networked applications are organized into layers. The
Internet Protocol is a layer. So is the HTTP protocol for moving Web
pages around. Layers manage complexity by defining abstract bundles
of functionality that are likely to have many uses. Layers are also
economically efficient because they enable economies of scope: the
more services you can build on top of a given layer, the more ways
you have to distribute its cost. In fact, if you have a layer like IP
that supports a wide range of valuable services, then you can probably
make it free, just from the labor contributed by the people who make
money from the stuff built on top of it. It's a good system, and much
of the long-term quiet politics of the computer world is about the
ways in which future layers are going to be defined. If you try to
make them all things to all people then they will crash, but if you
try to focus them then someone will be upset. Current examples of
long-term layer-development are digital libraries, cooperative work,
large-scale simulation, and distributed objects. Digital computing
(as opposed to analog) is itself a layer. A product can be premature
if it combines a lot of functionality that should really be provided
by layers. Such a product is called a "stovepipe", and earlier I have
described a technical and economic dynamic called the "platform cycle"
by which stovepipe products get made obsolete by the coming of new
layers that abstract away the common elements of their functionality.
To think up examples of stovepipe technologies, start with any major
layer, define its functionality, and then think about products that
included that functionality before the layer existed -- any special-
purpose network that preceded the Internet, or any distributed display
system that preceded the Web. Right now, the world badly needs layers
upon layers of middleware to support emerging distributed applications.
An example is Akamai-style distributed database technology, which sits
between your Web browser and the Internet and invisibly redirects your
information request to a nearby uncongested server. The same sort of
thing will be needed for wireless and ubiquitous-computing services,
and it would probably be a mistake to build those services before the
general-purpose distributed database layers are available.

It's entirely likely that the smart players know all these things.
They have the intellectual property, capital, and skills that they
will need to join the game, and they are calmly waiting for the time
to come -- for the market to acquire scale, for media synergies to
develop, and for new service layers to be standardized and deployed
on a large enough scale. They don't have to engage in hype. What's
unfortunate is when unsuspecting investors plough their cash into
new ventures that really are good ideas but aren't good ideas now.
--------end RRE News excerpt----------------

Strata Rose Chalup [KF6NBZ]                      strata "@" virtual.net
VirtualNet Consulting                            http://www.virtual.net/
 ** Project Management & Architecture for ISP/ASP Systems Integration **

This archive was generated by hypermail 2b29 : Fri Apr 27 2001 - 23:18:41 PDT