Subject: Re: the walls have ears
From: craig@jcb-sc.com
Date: 31 May 1999 06:35:35 -0000

>>>>>> "craig" == craig  <craig@jcb-sc.com> writes:
>
>    craig> But, yes, in the end, as I've said for many years now, if
>    craig> the customers don't *value* having the source code
>    craig> available to them, BSD license will beat GPL "hands down"
>    craig> for business.  Not because it *is* more valuable, but
>    craig> because, due to what I feel is a failure to properly value
>
>Them's fighting words.  You're welcome to your feeling, and it may be
>accurate.
>
>But without measurements, nobody, least of all you, will ever know.  I
>wish I had them, but in their absence, I think theory speaks in
>opposition.

Yes, just my feeling.  But I've seen plenty of people write stuff like
this, paraphrased, such as in the foreword to the book "Linux at Work"
(not an ORA book, but I forget offhand whether it was a Wiley book
or something else):

  I thought having the source code to programs we were using didn't
  matter much, but, after [various experiences] resulting in [big
  problems with proprietary code] compared to [great experiences]
  due to [having the source code and the choices that gives], I
  decided otherwise.

If anyone could point us to contradictory experiences, as in:

  I thought having the source code to programs was important, but it
  turned out to be a bad thing because [bad experience] so now we
  have a rule in our organization to not use any product for which
  we actually have the source code, although we are permitted to use
  products that are derived, by others, from BSD'd and PD'd code,
  as long as we get only the proprietary binaries.

Since I have *never* heard an experience like that related in any
serious way (if at all), it's hard for me to understand why you
have such a problem with my statement regarding the importance of
*customers* valuing source-code availability vis-a-vis the success,
freedom, or whatever in the BSD/GPL argument.

>One problem with programmers' discussion of management decisions is
>that programmers rarely consider all the costs that management does;
>they don't have the experience to do so.  (Of course, the big problem
>with programmers not discussing management decisions at all is the
>converse ;-)  Keep talking guys!

Interesting assumption there, that, because I'm a well-known programmer,
I'm therefore not a manager.

In case it matters, I was a Lead Technical Writer and then a Manager
of Documentation (basically starting a writing "group", such as it
was, from scratch, on nearly zero budget) for a string of several years.

I have a clue about what is involved at the management level.

And, in my discussions of the free-software concept among management
and entrepeneur types from Board members of churches all the way to
John Templeton, whenever I've pointed out, e.g., the fact that having
the source code to programs being used in-house means having *more*
information and *more* control, I've yet to have a *single* one of
these people respond by saying something like:

  "But a good businessman does everything he can to reduce the amount
   of information he has on, and control he has over, his business,
   so your free-software notion will never catch on."

>    craig> the assumption that source code is of little value will
>    craig> have nearly vanished.
>
>Who assumes this?  I would say almost nobody.

The answer is "almost everybody" when it comes to people arguing that
the BSD is "more free" *or* "more successful" vs. the GPL, in the
context I wrote, specifically, the importance of source code to
*end users*.

>The issue is whether
>clients will have the resources to exploit it.  If not, the value is
>less than the cost.

Without the source code, you never get to find out.  Period.

>Note that one of the necessary resources is the
>knowledge, in management, that you have the other resources; as
>high-level languages have come to dominate, managers will start to
>realize that in-house programmers or non-author consultants without
>specialized knowledge of the application may be sufficient, but when
>managers had the impression that programming was a black art and all
>the geeks around them were mere sorceror's apprentices, they didn't
>have that essential knowledge of possibility.  And from a manager's
>point of view, geek hubris is just as dangerous as any other kind.

That's certainly one dynamic in operation.  As more of them realize
the freedom to have source code in-house gives them the same kind
of flexibilities they like to employ when choosing to outsource
activities not central to their mission, they'll tend to value that
freedom (and the information that comes with it) more.

Again, once enough *end users* decide that the risks of being "caught
with their source down" (no source code for their products) are just
too great, the profit-making potential of proprietary or BSD vs. the
GPL will stop appearing so great.

>    craig> Always remember: the only reason people make money off of
>    craig> proprietary software is that people are willing to buy
>    craig> software without having free access to the source code.
>    craig> The moment that's no longer the case,
>
>I assure you, that will forever be the case.  People will always be
>willing to buy closed-source products, if (1) they monopolize an
>important feature and the price difference is not too great, or (2)
>the closed-source software is cheaper and the price difference is
>great enough.  There is no exception to those rules for business
>demand (really; that's both theoretically and empirically verified)
>and most people are suprised at how small a price difference is
>necessary to generate some substitution in demanded quantities, even
>for consumer goods.

How many businesses that buy fleets of cars (looking at only those
that don't do it as part of their central mission, or at all of them,
e.g. whether you include rent-a-car companies) deliberately choose
to buy them with the hood welded shut?

>The fact that (2) is not true for any known (to me) products today
>doesn't mean it will always be so.  At some point, some customers may
>be willing to risk lock-in through closed-source licensing, while
>others will refuse to do so and demand source-available licensing
>(thus submitting to lock-in by NDA), and sufficiently public-spirited
>ones may contribute fees sufficient to acquire an open-source license
>(it's conceivable though I think unlikely that avoiding lock-in would
>justify this).  Broader source availability will result in higher
>license fees.

Yes, but please think through the wholesale implications of what you
and I appear to agree about, that *more* customers (a higher percentage)
*will* demand end-user source availability, and more of those that do
will demand GPL-style freedoms to circulate improvements among other
customers who have similar interests in the code continuing to work.

When I think it through, I see the following pressures this puts on
the proprietary model, which is also the only model that provides
better profitability/penetration of code based on non-GPL free/OSS
licenses:

  -  Customers value having the source more, so more of those who do
     choose to *not* have it in specific instances, do so because they
     are expecting higher value from proprietary products than they
     would otherwise.

  -  Putting that higher value into proprietary products requires bigger
     investments up-front in those products, and thus better prediction of
     the market size and tolerance for the pricing.

  -  But those same customers also know that, in addition to turning
     down a potential source-available option (however mythical),
     they're turning down an option that they could obtain for *free*
     from some channel.  So the pressure on pricing of the proprietary
     code is greater, just as is the pressure on the market size, and
     (especially) market penetration.  E.g. the days of 90% (or whatever,
     I don't keep market numbers handy) penetration of proprietary compilers
     is long-gone, that of proprietary OSes is probably long-gone (soon if
     not already), and there's reason to believe this trend will continue
     through to the desktop.

  -  Further, the pace at which free software is developed, especially
     in response to (creating clones of) highly successful proprietary
     products, is generally increasing, not slowing down.

  -  That increased pace not only pressures the proprietary vendors
     to increase *their* pace of development, but reduces the window
     during which they can command high prices, market share, etc.

  -  And, it reduces the window during which *customers* believe their
     best short-term choice to be the proprietary code.

  -  That means VCs are going to be expected to provide more funding
     to create higher-valued proprietary products for a smaller
     penetration of a market that tolerates lower and lower prices
     and which *plans* to be loyal to those products for shorter and
     shorter periods of time.

  -  Which, in turn, means VCs with actual cash in hand are probably
     going to be fewer in number, with many going off to help fund
     other businesses, e.g. those in other lines of work that happen
     to be exploiting GPL'ed software well.

  -  More on the "empathetic", or whatever, side of things, once the
     "bloom is off the rose", in public perception, for the proprietary-
     software model -- once the phrase "proprietary software" no longer
     causes the average joe to swoon, with visions of megabuck-laden
     Microsoft in his head -- I *think* we'll see a general decrease
     in the assumption that proprietary software offers inherent value
     over free software.  (As I believe I noted previously, I often
     get the impression, when discussing this concept with "average
     joe" types, that they really believe "Since Microsoft made it
     big with the proprietary model, it Must Be Good for me to be on
     the buying end of it".  In their comments, they campaign quite
     vigorously for the continued success of that model, and hardly
     take the time to consider what *freedom* would mean to *them*!
     Once the old model *feels* old to the general public, I predict
     there'll be a lot more awareness of the value of freedom to
     have the source code.)

Since I'm not hypnotized by the huge chunks of change involved in the
software industry by the past 15-20 years -- in fact, I don't even
find that amount of *time* to be particularly noticable on the scale
I tend to think about -- I don't believe some Magic Pill is going to
change any of the above, unless it comes in the form of extreme
government interference in the free-software movement (e.g. software
patents), making the frictions of its development and deployment cycle
artificially high.  (Actually, there's already *plenty* of such
interference already, but I think it shows the robustness of the GPL
model that it's succeeding anyway.  E.g. there are no laws that get
people convicted of felonies and thrown into prison for ten years
just for shipping ten copies of a proprietary program with GPL'ed code
inside, but let those ten copies be of MS Office, and it's another
story, last I heard.)

>My point, with respect to Craig's argument, is not that he's wrong,
>but simply that this is not a bang-bang issue; market penetration for
>free licenses and in particular the GPL is going to be an interesting
>statistic to watch for some years, at least, and there's no guarantee
>it will ever have a majority, let alone go to 100% (although we can
>hope).

It won't need to get near there to make the viability of creating
proprietary software vastly lower than it is today, and I don't think
I've *ever* represented this as "a bang-bang issue", except in the sense
that to the *extent* every customer requires source code, the present
market dynamics in favor of proprietary software go out the window --
so, *if* they all magically chose GPL'ed-type freedoms, *then* it'd
be "bang-bang", but I don't claim that's going to happen.

I don't think GPL vs. proprietary software will necessarily get near
to being a calculator vs. slide rule thing anytime in the next 20
years or so (barring government interference on *that* side, e.g.,
some kind of unlikely post-Y2K hysteria where everyone becomes convinced
that having the source code would have prevented all sorts of
problems).

But there are more convincing arguments in favor of choosing a slide
rule than there are in favor of choosing to not have the source code
to programs you're using.  Someday customers will realize this en
masse, and start asking for a heck of a lot more out of the sourceless
products they *will*, I expect, still choose -- which will have to be
as much better than the free products as today's calculators are than
yesterday's slide rules.

        tq vm, (burley)