Radical computing thoughts: Data impedance & automatous computing
Sun, 28 Oct 2001 18:55:07 -0800
> My guess is that the Web Services camp is implicitly buying the
> worse-is-better* argument
This is an interesting point.
"worse is better" argues that it is of primary importance for a piece of
software to have a relatively simple implementation, presumably for
maintainability, correctness, and portability. If people/programs using the
software have to do a little more work as a result, so be it. That is,
simplicity-of-implementation trumps ease-of-use.
However, it never occurred to me that the Web Services camp would be pushing
this, since they promote very actively that the service protocols have to be
very easy to understand and use, so that lots of people can write and use
Web Services. And I believed this, since otherwise I can't imagine a single
reason for making every thing (a) text-based and (b) self-describing (unless
this is secretly backed by vendors of disks, memory, bandwidth, or CPU, all
of which will be consumed in great quantities.)
But, on the other hand, another important element of Web Services is
portability. This stuff is supposed to run on everything from handhelds to
72-way servers, from CE to Solaris to Lynx. And presumably in order for that
to happen (quickly), the protocols have got to be *simple to implement*.
Which makes me wonder -- which wins out? Easy to use? Or easy to implement?
At this point, it seems quite apparent that "easy to implement" is winning
out, and that "ease of use" is only superficial.
Any thoughts? I wonder, if this is the case, why it is. Why are
Microsoft/IBM pushing this so hard? To sell more stuff? To provide an
alternative to Java? To promote a charade of populism while in fact only
they have developers skilled enough to specify/infer the context needed to
build a successful Web Service? Other?