Subject: Re: the walls have ears
From: "Stephen J. Turnbull" <>
Date: Wed, 2 Jun 1999 15:46:12 +0900 (JST)

>>>>> "rn" == Russell Nelson <> writes:

    >> If the free software upper bound is less than 100%, that
    >> implies that there is a proprietary product out there, and in
    >> the minds of customer it is significantly better to justify the
    >> price differential, if any.

    rn> Automatically?

Yes.  Think about it.  Assume the customers are smart and
well-informed, of course.  I should clarify my erroneous phrasing:
significantly better to justify the assumed significant price
differential.  Of course if there's no price differential it need not
be significantly better.

    >> Doesn't that cost free software credibility in the minds of the
    >> audience?

    rn> You're positing the existance of software which most people
    rn> here don't think exists.  There are boogeymen, too.

I set up a boogeyman to demonstrate the mechanism where the outcome is
obvious.  But we don't even want to get into fair fights!  We want to
go into markets where free software has a comparative advantage.  I
have no idea, but Martin Buchholz (former Sun, estimated
in a speech in Tokyo that 1 in 1000 developers is a primarily free
software developer.  If so, don't we want to concentrate those rather
scarce resources in areas where the impact is greatest?  And isn't one
factor in making the impact greatest identifying products where free
software's advantage is high and is likely to make a big win?  (Not to
mention that profit is probably fatter there.)

I'm not suggesting a central task allocating agency or anything like
that; I just think that a lot of developers would like to have some
tools to help evaluate the likely outcomes of proposed projects.

And some people will go for broke.  Thanks to Karsten Self, who
advised me of  I'm on record as saying it
won't work, by which I mean it won't ever get significant share (>=1%)
in the Windows market, nor will anything like 100% of Linux/Windows
dual-booting types trash their copies of Word for it (make that
">=50%" for the purposes of keeping score).  I lose if either one
happens.  People keep accusing me of saying "nobody buys insurance"
(sorry Russ, it's not just you, but that's the example freshest in my
mind).  Bah.  OK, now you see what I really am saying.  1% of the
Windows market ain't "nobody".

Don't like my rules?  Fine, make up your own.  But I don't see anybody
else thinking in quantitative terms like that; I think we should be.
I'll agree, my definition of a "free software win" above has some CYA
factor for me in it.  But Linux has won that big (defining the market
appropriately, obviously Linux can't have 1% of the Windows OS
market), Apache has done it (both counts, I believe), XFree86 (X
desktops only, obviously).  I'd be happy to discuss what a good
definition of "win" is, then go off and look for markets where it's
easier to "win."

BTW, Microsoft Word is not my boogeyman; as Russ says, most people
here, including me, don't think feature domination plus reliability
exists in the proprietary world, let alone in MS Word.  This is one of
the "fair fights we don't want to pick" IMO.  I won't take it back
just because somebody thinks they can do it, or because my real
opinion is "4:1 against," not "can't happen."  (I do reserve the right
to admit later that my current model is broken or my data was
insufficient, and try to do better in the future.)  And I'll try hard
to smile as AbiSource stockholders laugh on their way to the bank, if
they do ;-).  I'm rooting for them, really I am.

OK?  I know lots of you have put your money where your mouth is
already, in your businesses.  There's mine; I can't match yours---I'm
not a software developer.

(BTW:  I just got Brian Bartholomew's post regarding AbiSource; I'm
not going to change anything here in response.  I don't subscribe to
his broken software theory, anyway.  I have no opinion about the
quality of the software, I really should do more research before
sticking my neck out with predictions.  Talk is cheap.  ;-)

    rn> Not just orphanage, but also refusal to implement a desired
    rn> feature, fix a show-stopper bug, or just all-around provide

Stipulated.  No theoretical problem, just changes the input parameters.

    rn> bad support.  Again you're trying to tell us that people won't
    rn> buy insurance.  We *know* that's false, so try telling us
    rn> something else.

Who said people will _never_ buy insurance?  Not me.  What I _am_
saying is not false.  "Everybody buys insurance" is almost true, but
not a useful rule of thumb for getting into the insurance business.
"People who buy expensive new cars buy comprehensive coverage; people
who inherit junkers from older siblings don't" is a useful rule in
this context.

I'm saying certain people won't buy certain kinds of insurance at the
market price.  Aren't which people and which kinds of insurance
interesting questions here?

We know that some kinds of free software products work, to some
degree, in certain circumstances.  We've seen it happen.  Yes, I do
believe the evidence before my eyes: Linux, XEmacs, TCP/IP, sendmail,
XFree86, fvwm2, gnuplot, TeX, vm.el, procmail, and that's just what's
on my screen at the moment.  Still, wouldn't it be nice to know more
about _which_ ones?  And especially where the best growth potential
is?  And most especially in advance of committing development effort?

Free software is not a commodity.  When Tim O'Reilly talks about the
commoditization of the software market, he's making a valid point.
But what is becoming a commodity is not software itself, it's
individual software products.

These products are highly differentiated; an important question is
still _which_ products should we implement or improve?  Isn't it?

University of Tsukuba                Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
Institute of Policy and Planning Sciences       Tel/fax: +81 (298) 53-5091
What are those two straight lines for?  "Free software rules."