Subject: Re: Non-proprietary software?
From: ian@airs.com
Date: 26 Sep 1998 10:20:31 -0400

   Date: Sat, 26 Sep 1998 04:34:41 -0400
   From: Brian Bartholomew <bb@wv.com>

   > I have a nearly infinite supply of zeros and ones that you are
   > welcome to use in any way you see fit.  These ones and zeros are not
   > branded, QC'ed, marketed, documented, and have no future.

   I think that's silly and doesn't address my point.  When I buy a
   railcar of corn, I get corn -- but I don't care very much who I buy it
   from, and it has a very competitive price.

From the perspective of somebody who is not in the corn business,
there is no significant difference between one piece of corn and
another.  The only reasonable comparison I can see between buying corn
and buying software is to consider buying copies of the same piece of
software.  In that case, your non-proprietary software already exists,
in the form of public domain software.  You can choose the ISP you
want to get it from, and the site you want to download it from.

Presumably, that isn't what you mean.  You want to be able to say ``I
want a word processor'' and get a choice between different word
processors, all commodities with as little reason to prefer one over
the other as there is to prefer one ear of corn over another.

I don't know if I have stated your wish correctly, but I feel
confident in saying that the situation I described will never arise.

As long as software requires human intelligence to create, there will
be significant differences between different pieces of software.
Software packages will be driven to be different, not the same.  There
will never be very many different versions of a particular type of
software, and you will have to choose among the available versions
intelligently.

Software will not be a commodity until it becomes so cheap and easy to
create that you don't even need to think about it.  At that point
you'll just create a new piece of software when you need it; you won't
buy it from anybody.

(For example, using Unix filter programs is like this.  Nobody would
purchase a particular arrangement of piped programs at any price, even
free.  People just recreate them as needed.)

On the other hand it's easy to imagine a commodity market in software
services, in which you purchase the right to use a particular service
for a limited amount of time.  This would be a natural path for search
engines, such as the existing (free) Internet search engines and the
existing (expensive) services for particular domains such as legal
documents.  But I don't think that's what you're talking about.

   There is the technical question of whether software can be specified
   or tested cheaply enough to ever be a commodity.  If it can't be
   tested, then probably I'm stuck with a brand.

Specification and testing are not intrinsically the problem.  One can
imagine a formal specification language which leads to automatically
generated tests.  The problem is that creating the software requires
human attention--quite unlike the problem of creating an ear of corn.

Of course, any formal specification is itself a piece of software, but
I think that shows that the problem still reduces itself to one of
creating the software.

   However, this brand
   could be managed in the least proprietary manner that still preserves
   the brand technically.  For instance, the FSF brand seems to survive
   copying much better than the commercial Linux distribution brands.

I don't think I follow this.  There is a great deal of overlap between
the FSF brand and the commercial Linux distribution brands; in some
sense, that's why RMS pushes the term GNU/Linux.  The fact that you
can distinguish them (and I can too) is that they are branding
different kinds of things.  Commercial Linux distributions are
branding binary ready to run software.  The FSF is branding source
code, which you have to turn into a useful binary yourself.  So your
comparison is like comparing seed corn with the final ear of corn.

But probably I don't understand your point.

   > Well, it seems clear that what he wants is the instantiation of an
   > idea, without paying someone for that instantiation.

   It's not paying that irks me, it's paying proprietary-derived prices.

I think it's a good rule of thumb that over the long run nothing will
cost less than it costs to produce it.  Software development is
expensive because it takes intelligent human time for each program.
``Proprietary-derived prices'' are just one mechanism to recover the
costs of software development.  Free software businesses pursue other
mechanisms.  You can do your own software development if you choose,
and thus pay the true price directly.

The gratis software market tends to defy this rule of thumb to some
extent because some hackers take their payment in the form of
recognition or a feeling of contribution rather than money.  In other
words, the price is worth more to the payee than to the payer, so it
looks like a good deal on both sides.  However, if software does
become a commodity, then I predict that that will no longer be the
case.

   > For each software construct that you create and place in the pool of
   > shared software, you get a dynamic number of softbucks, which can be
   > used over and over again to procure more software.

   As a software creator I wouldn't be excited by such a plan, because it
   seems unlikely to gain me credits in US Dollars, which I could use
   towards the rent.

Now I just don't get it.  If software becomes are a commodity, where
are you going to get those US dollars?  Farmers only survive--to the
extent that they do survive--by applying economies of scale, and I
can't imagine how to apply that to the production of new programs.

Ian