Subject: Re: middleware as an OS replacement
From: Ben_Tilly@trepp.com
Date: Wed, 11 Jul 2001 11:33:39 -0400


<gdwyer@dwyerecon.com> wrote:

> The Microsoft case record raises an important question that has
> substantial implications for the future development of open-source
> software. I am inclined to think that many of those analyses are quite
> wrong.
>
> Much of the Microsoft case hinges on the idea that Navigator posed a
> serious threat to Microsoft's _operating-system_ dominance. Essentially,
> the theory is that a programmer might at some future date use
> Navigator's APIs rather than Window's APIs to access the screen, memory
> and storage devices. The same argument is made about Java.

Actually it hinges on the idea that Microsoft thought that Navigator
was a credible threat to the application barrier of entry.  There is
little question that they thought this.  Their internal emails made it
clear that they did.

This does not mean that Navigator would have provided access to local
hardware.  It means that applications that people want could be
delivered through the web.  And it does not mean that all applications
would be delivered through the web.  It just means that enough of them
would be to make it take less work to make a competing OS competitive
with Windows.

Also there are good reasons to prefer having data *not* stored
locally.  For many applications it makes sense to store data on a
central server, and many of the current directions that Microsoft is
trying to take reflect this.

> This strikes me as improbable at best. This is an advanced version of
> (what I think is) a common idea in the history of programming -- write
> programs that can be executed on any computer with effortless porting.
> Interpreted programs are a common manifestation of this idea. The end
> result historically has been slow programs that do not reflect the
> hardware's unique capabilities and disabilities.

Indeed this is a common idea in the history of programming.  The
result on the one end can be pathetically slow programs.  The result
on another end can be the IBM mainframes and the AS/400.

But the truth be told, most of the "little programs" that businesses
need to run smoothly are *not* performance critical.  Today many of
those are being delivered over an internal website, or are delivered
in a portable language such as Java, Perl, or Python.  When I replace
what used to be a DOS batch script with a button on a web page on an
internal website written in Perl, I have just made it one step easier
for an internal desktop to be Windows or another operating system.  I
have made it one step easier for that server to be Linux or another
operating system.  I have made it one step easier for that company to
choose to switch operating systems rather than absorb whatever
licensing change Microsoft decides to institute.

This does threaten Microsoft's monopoly.  And part of Microsoft's
response is to make it as natural as possible for me to accidentally
make that button on that web page only work with IE on Windows.

> Rather than being broken into pieces with the underlying OS being a
> collection of drivers, OSes have been adding features at least since
> the advent of PCs. (I don't know much about mainframe OSes.)

It is safe to say that mainframe operating systems offer more features
than PC operating systems.

> Linux, FreeBSD, etc. are threats to MS precisely because they are
> full-featured OSes.

People don't buy operating systems.  They invest in applications.

Microsoft's power comes from trying to make all of the applications
work together easily, but only work on their operating system.  Then
people's investment in applications ties them to the monopoly, and
once they are so tied, Microsoft can proceed to charge monopoly rent
on the economy as a whole.

> Am I missing something? There is a lot of analysis out there assuming
> that Navigator was a serious threat to Windows.

I believe that you misunderstand the source of Microsoft's monopoly
power and the nature of the threat that Netscape posed.

> This list may not be the appropriate venue for raising this issue. If
> not, I apologize but I would appreciate being directed to an
> appropriate list.

I would think that slashdot and kuro5hin are more appropriate.  Or
LinuxToday.

Plus there are plenty of small sites out there.  For instance if you
went to http://z.iwethey.org/forums/render/forum/show?forumid=18 you
would be on topic.

Cheers,
Ben