Subject: Re: economic efficiency of free software
From: "Benjamin J. Tilly " <>
Date: Tue, 10 Feb 2004 06:06:02 +0800

Jamie Lokier <> wrote:
> Stephen J. Turnbull wrote:
> >     Jamie> Also, one of the problems with any proprietary product (not
> >     Jamie> just software) is that some "demanders" are refused access
> >     Jamie> to the product simply because the economic process requires
> >     Jamie> that refusal, not because there is any real cost to the
> >     Jamie> business in those demanders having the product.
> > 
> > You've apparently missed Thomas Jefferson's point, too.  Refusing
> > access to those who don't pay is not a _problem_ resulting in
> > undersupply, it is a _proposed compromise solution_ to undersupply.
> Yeah but you missed my point: it doesn't work, and the quality-of-life
> cost is too high :)

You have claimed the point, but not convinced me of it.  At least not
in general.

In some contexts, eg public health, I agree absolutely.  My personal
choice to monitor which restaurants have a cook with cholera doesn't
keep others from being careless, and I can't easily protect myself
from the resulting epidemic.  Therefore preventing epidemics becomes
a public good, and it is well understood in economics why those are
poorly provisioned from private efforts.

In other contexts, eg Levis, I strongly disagree.  Only giving Levis
to people who are willing to pay works pretty well in practice,
particularly when compared to alternatives that have been tried.

Drawing a solid line between the two kinds of endeavours can get
complex.  But it is important to recognize that both are good.

> > Another way to grasp the magnitude of the computational issue is to
> > recall that 50% of the non-material costs in the US economy go to
> > "knowledge workers" who are not directly related to physical
> > production of goods or consumer services, but rather to instructing
> > the rest of the economy about what goods to produce and where to
> > deliver them.
> > 
> > Ie, the US economy is a distributed computer whose running costs are
> > about $5 trillion/year.
> You don't know the true cost of it.  Perhaps it's running cost is $5
> million/year, plus an extra $5 trillion/year is wasted on peoples'
> need to justify their needs, which is a consequence of the economic
> process.

Wrong.  Stephen does know the true cost OF WHAT WE SPEND.  It is about
5 trillion per year.  

Stephen doesn't know the OPTIMAL cost.  That may well be much less.  It
could be more.  (Perhaps a better focus on decision making would give
better outcomes.)  My personal belief is that the optimal level varies
widely depending on the value system of the person judging.

However I'll point out that anyone claiming that their alternate system
does significantly better should do their homework carefully.  Many
people over the years have proposed alternate systems, and some have
even had the freedom to try to implement them.  As chaotic as it is, our
current system has done astonishingly well.

> You can start with the premise that there is a computational economic
> problem, and proceed to estimate it's cost, in which case you have no
> idea what its cost is, because you have no idea how many people are
> toiling away at work created for the sake of work.

The premise is true, though.  There is a production and allocation
problem to be solved.  Ignore it or not, it is there.  You can try to
solve it by top-down planning.  The Soviets did, with a notable lack
of success.  You can try to solve it by finding a system that solves
it implicitly.  Capitalism does that and overall has done fairly well.

> What proportion of those knowledge workers, or the work they do, do
> you think are necessary to handle the computational issue which you're
> referring to?

In a perfect world, or the real one?

In the decision making question at hand there is a real complexity
issue.  When you try to centralize decision making, knowledge gets
dropped.  If you try to increase the number of key decision people
naively, the number of lines of communication increase quadratically
while work only increases linearly with the number of people
involved.  Before long diminishing returns leaves everyone in
meetings, with no work being done.  Finding an appropriate
compromise is hard, and any answer chosen will dissatisfying to most.

In a perfect world, of course, the right answers would magically
appear with much less effort.  In a perfect world you could just
trust employees.  In the real world, right answers are not so
obvious and you have to worry about how people's self-interest does
not match their putative job interest.

A random example is that traders in financial firms have to get
astronomical salaries.  If they don't, then it becomes cheaper for
their opposing numbers to say, "Let me buy that for X and I'll slip
you Y under the table."  Of course that happens anyways, but it
happens less when your trader has more to lose from being bribed...

> How many of them are doing work for the sole purpose of "doing work",
> because the economic process requires people to work in order to get
> their basic needs met, even when the work is not required for those
> needs to be met?

Fewer than you would think because employers are notoriously loathe
to pay for people that they don't need.

> A clear example in our field is all the effort spent creating patents
> by companies for the sole reason of defending against the other
> companies' patents.

The patent system is broken in more fundamental ways than that...

> > [2]  See Richard Stallman's essay "Misinterpreting Copyright---A
> > Series of Errors" in _Free Software, Free Society_.  (I suppose it's
> > available on the web somewhere, but I prefer to buy the book and pay
> > royalties to rms. ;-)  I believe rms is wrong when he argues that in
> > software there are other reasons why "enough" of the "right kinds" of
> > software will be produced without proprietary modes of distribution,
> > but his Constitutional scholarship is accurate.
> Can you describe any kind of software, that is in some way useful,
> that would not be produced without proprietary modes of distribution,
> assuming _some_ kind of sponsorship whether from individuals or
> organisations, and the appropriate sponsorship criteria?

I can think of kinds that I think would not, but I doubt that I could
convince you of that.  I know that I can't convince you if you
postulate sponsorship of kinds that people like to dream about which I
believe are impossible.  Conversely it is likely that you believe that
there are modes that are possible which I find implausible.