[
Lists Home |
Date Index |
Thread Index
]
The problem of arguing the network effect as an incentive
is it assumes everyone wants to be popular when some merely
want to be close to those they choose. Business services
aren't group gropes; they are secret societies with
secret incantations. They are communities that self
select through sharing vocabularies and by this, they
accumulate and conserve wealth among their members.
Classic keiretsu.
What if the network effect is deliberately limited and this
is precisely the sensitive requirement? They don't want
to grow it; they want to control its growth, and by that,
I don't mean the SOAP system vendors; I mean their customers.
Not surprisingly, web services, that is, roll your own API,
is a return to the "play your own game" approach: wealth made
through sharing some aspects of the system, but differentiation
based on aspects that are private, that do lock in the buyer
for the duration of a contract, but don't lock them out
of another system as long as it can process the same SOAP
messages. The lock-in is the message set.
Like the StorySpace example in the article, the motivations
to build services are self-sustaining and self-modifying, so
it can win because it has a different definition of winning
than a game based on achieving a network effect of a user
interface; the effect is based on the message set, the
transaction types. The client is irrelevant except insofar
as it is convenient (you don't need HTML, XForms, SVG,
but they are convenient).
I do know about the difficulty of getting a common schema
out of an industry. One man's ontology is another man's heresy.
Getting vocabulary agreements in place is the two-headed
alligator of the markup industry:
two heads, both have teeth, it can't sh*t, is always cranky and wants to go in
different directions so mainly just sits there and bites anyone who comes
too close.
The SOAP part is easy. The message set is hard.
len
From: Mark Baker [mailto:distobj@acm.org]
FWIW, here's a great paper on network effects in hypertext systems;
http://www1.ics.uci.edu/~ejw/papers/whitehead_ht99.html
|
