Subject: Re: A few here may have an opinion on this
From: "Stephen J. Turnbull" <>
Date: Thu, 24 Oct 2002 20:47:22 +0900

>>>>> "Brian" == Brian Behlendorf <> writes:

    Brian> It's an interesting thesis; sounds like you're essentially
    Brian> arguing that the more open source software replaces
    Brian> proprietary code, the lower the "whole chunk" would be,
    Brian> meaning - GDP would be less?

Yes.  I have a student who has figures that say that in Japan, where
overall investment has an internal rate of return (IRR) of about 20%,
IT as a whole has an IRR of 75%, and software has an IRR of 200%.  We
know that both the IT and software numbers are huge overstatements,
but we expect the qualitative result (IT is "much more" productive
than general investment, and software "much more" productive than
hardware) to hold up.  So ...

    Brian> Whatever money the end-user (corporate or individual) would
    Brian> have spent but won't doesn't just disappear.  It goes into
    Brian> funding other needs, or reducing the cost of the
    Brian> product/service you sell.

Yes.  And since it's much less effective in producing value-added
(according to the hypothesis above) than investment in software, net
total end-user benefits decrease.  I don't see any way around this
conclusion---if Chris is right, that OSS is necessarily less
productive than proprietary modes.

But I haven't yet accepted Chris's thesis that encouraging GPL
licensing necessarily means less software being produced.[1]

    o A lot of improvements that will be made in the OSS ecology,
      which (as a non-cult phenomenon) is less than 10 years old.
      Maybe Eric's cauldron really _is_ magic.  We can hope.

    o Many CS students will happily become "lifestyle" businessmen, or
      join such businesses.  There will be another wave of foundations
      of FSBs in a couple of years when this recession ends.

    o The infrastructure argument (made by several people already) is
      correct.  This won't support market-oriented FSBs, I'm afraid,
      but OSS for infrastructure projects is clearly (IMHO ;-) socially
      beneficial, ie, generates more value than proprietary modes
      can.  So government-sector FSB contractors should get business.

Still, Chris's point is plausible, and it's wrong for Kragen to ignore it.

    Brian> Since IT "drove" (for better and worse) the world economy
    Brian> in the 1990's, does that mean a threat of recession?

No, I'm Schumpeterian.  I don't think that lack of productivity
creates recessions (it results in decreased standards of living).

    Brian> What value do you think could be placed on the Internet?

On the order of Word.  Not as high as the whole Windows platform by at
least an order of magnitude.  In the future the Internet will
contribute a higher order of magnitude than any other platform, I
imagine.  But much of that Internet-generated value should be
attributed to proprietary software (Solaris servers, Cisco routers,
Microsoft clients).  Not because it's better than OSS, but just
because in fact it happens to be in place and proprietary.

Except that I don't think the correlation between "in place" and
"proprietary" is accidental.  (Nor is it due to the Illuminati.)

    Brian> Would it be crazy if I said the Internet would not have
    Brian> been built (that TCP-IP would not have won) without free
    Brian> software?

Stipulated.  I don't know how to prove it, but I certainly believe it.

    Brian> Not that free software alone built it, but the value
    Brian> derived from it is sometimes 2 or 3 orders of interaction
    Brian> removed.

Yes.  I don't mean to ignore that (in fact, the remark about the BSD
stack was supposed to hint at it), but unless the discount factor per
"order of interaction" is pretty close to 1, after a couple of orders
of interaction the faction of value derived from the original free
software is pretty attenuated.  I think it's enough to conclude that
any of several blockbuster proprietary apps (Visicalc, anyone?  how
about NintendoOS? :-) have created more value than all of free
software combined.

Also, Minix, Linux, and *BSD must all cede much of their claims to
generate value back to the innovator---AT&T's proprietary UNIX---in
exactly the same way.  That goes in spades for the purely taillight-
chasing apps, of which there are not a few.  Not to mention GCC, flex,
and bison, all of which have taken substantial inspiration from their
proprietary forbears.  And---pace, Mr. Edison!---inspiration is far
more than 1% of a software product.  How much of the value of OpenSSH
should be considered to accrue to the RSA patent?

[1]  "University" licensing can't possibly have this effect, it has to
be net positive.  Why?  Because if you accept the contract to produce
the free software, it must be because you can't make more money doing
it yourself with a proprietary license.  But the software is free, and
therefore available to others to use, generating external value at no
sacrifice of profit.  This external value includes any derivatives
which must be proprietary to support themselves, which is what the GPL
foregoes.  QED

Institute of Policy and Planning Sciences
University of Tsukuba                    Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
***   I grew up in the 70s.   Tom Hayden married Jane Fonda, Jerry Rubin   ***
*** wears a necktie.  Dammit, let's make _this_ Revolution work!!  Please? ***