Subject: Re: NYT Times on Mainstream
From: Frank Hecker <frank@collab.net>
Date: Tue, 29 Aug 2000 18:50:57 -0400

Brian Bartholomew wrote:
> > CODE NAME: MAINSTREAM
> > The President's Information Technology Advisory Committee, a
> > group of university researchers and corporate executives, will
> > issue a report advising President Clinton to support open-source
> > software as a way to meet the increasing demand for better
> > software.  The support of the federal government would be a
> > significant victory for proponents of open-source, who recently
> > have seen their ideology move from the fringe to the mainstream
> > of the computer-programming world.
> 
> The US federal government has collectivized and centrally planned
> every other sector of the economy, why shouldn't they believe they own
> software?  Does anyone else see this as a screaming horror?

With all due respect, I think that you're overreacting here; IMO there
is no such "screaming horror" on the horizon.

You can find some general information about this advisory committee at
this URL:

http://www.ccic.gov/ac/

including a list of committee members and a number of the documents the
committee has generated. As you will note, most of its work is concerned
with Federal funding of information technology research, particularly
with regard to "next generation Internet" research and scientific
computing. The committee also had some hearings and issued some general
statements around the "digital divide" question.

I don't have detailed information on all the committee's activities, but
as it happened I did attend and speak at a meeting of a panel associated
with the committee which was specifically considering the issue of
software used for high performance scientific computing, that is, the
sort of software used in research initiatives hosted or funded by
agencies like NASA, Department of Energy, National Science Foundation,
National Weather Service, etc. (This technology also has applicability
to agencies like NSA, of course.) This particular panel at least was not
concerned with the general-purpose desktop or server software used by
the vast majority of users inside or outside the US government.

As it turns out high performance computing is a special case, because
the overall market is relatively small, and because the US government
directly or indirectly drives a great deal of the customer demand,
especially at the high end (e.g., "teraflop computing"). Because the
market is so small relative to the investment to create products for the
market, traditional vendors of proprietary hardware and software for
high-end computing are finding it unprofitable to continue in the
market, and some are leaving it for good.

So what the committee panel has been doing is exploring the
possibilities of basing the next generation of high-performance systems
on open source software and commodity hardware -- basically extending
the Beowulf cluster idea, taking advantage of the Linux kernel and other
work already ongoing to support high-end systems, and adding more
software as necessary to support high-end scientific applications
(basically middleware for high-end computational computing). If this
particular panel does recommend some level of government funding then I
suspect it will basically be directed at funding development of software
specific to this application domain. Again, I suspect that the overall
committee recommendations will be similar, and will be in large part
directed at the role of open source software in relation to
government-funded research projects.

In terms of the broader question you raised about Federal control of and
influence over software development, I worked with IT departments and
others in US government agencies for several years, and I can't ever
remember anyone proposing anything along the lines of "central planning
for software". In fact, throughout most of the 1990's the watchword in
government was "buy COTS", i.e., purchase "commercial off the shelf"
software as opposed to contracting for software to be developed
specifically for government use (so-called "GOTS" software). That's why
at the same time the Justice Department was prosecuting the Microsoft
anti-trust case, US government agencies were standardizing on Windows on
desktops and servers, and purchasing site licenses for Microsoft Office,
database software from Oracle/Informix/Sybase, Internet software from
Netscape, and so on.

Unfortunately agencies also found that they still had
government-specific requirements in some areas (e.g., security) and that
commercial software vendors were not always motivated to address those
requirements in their proprietary software products. So now I think some
government people are speculating that open source offers a potential
middle ground between having all your software be custom written (and
thus incurring huge development costs) and having all your software be
shrinkwrapped (and having to depend on the vendors to meet all your
requirements).

If the US government does get involved with open source software, then
my hope is that they will do the smart thing and leverage the general
base of software already created and being created, and limit government
funding specifically to software and software features that are unique
to the government IT market. We'll see what the upcoming advisory
committee report recommends, but whatever the detailed recommendations
are, I very much doubt they will be the harbinger of a future "Ministry
of Software".

Frank
-- 
Frank Hecker            work: http://www.collab.net/
frank@collab.net        home: http://www.hecker.org/