🚀 go-pugleaf

In article <ki8aqcINN7so@early-bird.think.com> barmar@think.com (Barry Margolin) writes:
>Software is different.  Yes, there's some initial capital outlay, but it's
>relatively small compared to the cost of building a factory.  It presumably
>costs hundreds of thousands of dollars to build the manufacturing facility
>for paper clips that cost a penny apiece; on the other hand, it costs
>$5-10K to buy a computer that can be used to write programs that go for
>hundreds of dollars apiece.  The paper clip manufacturer has to sell
>thousands of boxes of paper clips to recoup his investment, while the
>software developer only needs to sell a few dozen.

Economies of scale.

(How's *that* for an expletive? :-))

I can start a one-man paperclip factory with a pair of
pliers and a castoff spool of steel wire from the Acme
Chicken-Coop Company.  Total cost:  two bucks and a mild
salmonella infection.  I can't charge fifty bucks a
paperclip, however.  Nor would I expect to pay "several
hundred dollars" a copy for something you twiddled out of a
Quadra in your underwear.

Real-world software-development, of the sort that large
companies protect, is undertaken by large teams of individuals
who occupy expensively zoned office space and require copious
amounts of Mountain Dew, Dental Care, Cash Money, and capital
equipment (an average employee costs ~100k/yr to pay, heat,
and manage (including the managers), and is allocated ~150k
of per-capita revenue by the accountants (who often know how
many atomic beans there are, but rarely know how many of them
are fissionable matter...)).  And then there's the many, many
people needed just to copy the software.  I remember the days
when there was a designated Chief Floppy Copier, who would
spend as much as ten hours in the middle of the night after
a beta deadline, making all of 10 copies of the release for
the beta sites (about 50MB each).

It may appear from the outside that the product is just so
much plastic and iron oxide, but from the inside there's a
couple of dozen man-years of blood, sweat, and debentures
to be recouped.

I can guarantee you that the GNU software now in
circulation, if valued at its development costs, which I
bet are considerable since I don't know of many UNIX
machines ca. 1982-90 that would have cost anywhere near as
low as $10k, either to buy or to run, would probably be in
the several tens of thousands of dollars per major program
(call EMACS $1e6, given the economic cost of the talent of
its author(s)), and easily over a thousand dollars per
executable file.

We're merely fortunate that the variable costs of software
are dinky (a few hundred to pay for a tape-set or network
transfer and the hard-disk space that will, eternally it
seems, be taken to store it; plus the several thousand for
the system guys to unpack and rebuild it, and to announce
its availability, and maintain it (none of this is required
when you buy a package from DEC, for instance...); plus the
few hundred in time taken by users to report and work
around the bugs, plus the <mumble> in time, waste, and rework
when users don't realize there's a bug and the results are
gunk but get used, anyway...) and that the developers and
their sponsors have donated their time and capital
depreciation to the users of the software.

Besides, economics isn't any sort of factor in patent
differentiation, so the whole issue's a red herring...



There's another point:  what does a patent really prevent and
how long does it hang around?  It certainly doesn't prevent
progress, since it's designed to promote progress.  What it
prevents is profit from copying, but nothing done by FSF
ever promoted profit, so profit isn't a motive.  Is profit
a motivation to progress?  Not necessarily; a billion[*]
economic-cost-dollars worth of software exists in GNU, and
nobody's profited per se from it, yet.

A patent exists for 17 years, give or take.  In 200 years
we've come from being able to make N parts the same to
being able to do a trillion mathematical calculations in
a single second (a story I read recently gave that a palpable
metaphor:  if every human on earth were given a pocket
calculator and told to wail away for something over a
minute, you'd get a trillion operations, and there's the
physical capability right now to plug together a few
boxes of circuit boards and do that every second; of course,
this doesn't do justice to the net = ~1.5e6 machines
times ~1.5 Mflops/machine = ~2Tflops/Earth).  Comparatively
it took 7000 years from poking seeds into the ground to
being able to mechanically pluck them out of cotton bolls.

If a patent does have a discernible chilling effect on
progress, just what sort of time-frame is it in which you
feel chilled?  Are you worried about progress itself, or
progress in your lifetime?  Is it a crucial element of
human existence that progress be achieved manically, or are
we killing ourselves with it?  (Hey, _I_ could do with a
vacation right now, but a certain twiddle I thought up
eleven months ago is taking just a little longer than I
scheduled to get itself fleshed-out, so I have to write
thousand-word diatribes on software patentability in lieu
of leaving my computer for a little R&R...)


A final point on the prevention of profit on technology:
How many people don't get the economic benefits owed them
(though they may not perhap demand it) for their
participation in this progress, because the product is
free?  How many don't get the benefits of participating in
the business of producing and distributing this thing,
because there's no unit product margin to support them?
How many don't have jobs providing the goods and services
used by the people in this business, because none of them
have incomes?  How many don't get the benefits of the
business of building a supporting infrastructure to
facilitate all of this direct and collateral economic
activity, because now nobody can pay taxes?

Progress is like minerals; you don't make it, you mine it;
it's the basis of an entire sector of the economy; but you
can't consume it, so it's no good economically unless it's
tradeable, not even to the miners.  But will it run out?


It's paradoxical, isn't it?  We get what we want, and it
don't cost us nothin', but we may starve because of it.


				--Blair
				  "Thanks for the editor;
				   now how about consing
				   me up a cheeseburger?"


[*] That's a conservative estimate.  Take Microsoft's
eternal revenue and multiply it by ten and you'll get what
I think GNU software will be worth once the OS is out, if
it works on micros.  That's not what it will earn,
certainly, nor is there really going to be that sort of
demand for it, maybe, but in a quality-based valuation,
that's the general ballpark.