Subject: Re: a model of competition between free and proprietary software
From: Ian Lance Taylor <>
Date: 20 May 2001 17:56:34 -0700

Seth Gordon <> writes:

> The number of developer-hours that F can attract is roughly
> proportional to F's quality, and if F's maintainer is competent, F's
> quality is roughly proportional to the number of developer-hours
> spent on it.  Therefore, if the first version of F is good enough to
> attract a few good volunteers, F's quality will grow exponentially.
> However, the number of developer-hours spent on improving P is *not*
> proportional to P's quality, but to its revenue.  Furthermore, not all
> of that revenue can be spent on developers' salaries.  Some of it must
> be spent on salespeople, lawyers, and the other
> accoutrements of a functioning corporation.  Most importantly, the
> investors who supplied the original capital must be rewarded, and they
> must be rewarded at an attractive rate of interest, so P's sponsor has
> one *cost* that grows exponentially.

One potential problem with this analysis is that it appears to assume
that developer-hours are more or less equivalent.  But in the present
state of the art in computer programming, that is not true.  Some
programmers are vastly more productive than others, by as much as two
orders of magnitude.  If the vendor of P is able to pay well and hire
highly productive programmers, P may have an advantage over F.
Conversely, F may be able to attract the attention of highly
productive programmers.  F has the advantage of being able to attract
from a wider range of developers; P has the advantage of being able to
aim precisely at developers with relevant skills (e.g., specialists
from academia, where that is relevant).  I can't decide who has the

You seem to disregard QA.  F may have the advantage of permitting
source code review (or may not, if P's source code is available for
inspection).  P has the advantage of being able to pay people to do
intelligent QA; it is likely that few people will be motivated to do
so for F.

You seem to disregard the auxiliary stuff which P can provide, like
professional documentation and packaging, things which F is likely to
lack especially in the early going.

In other words: developer-hours is not the only criterion for quality.

> At a certain point, F's level of quality is so close to P's that if a
> customer needs one feature from F that P does not have, it will be
> cheaper for the customer to hire someone to extend F than to buy P,
> and there are enough volunteers extending F that they can quickly
> match any improvements to P.  Once F passes this point, the return on
> investment for P's sponsors is no more than the ROI for a company
> specializing in F-related services; that is, they may be able to stay
> in business by providing support, maintenance, and consulting
> services, but the copyright on P itself has no market value.

It seems like your argument misses a step here: why should F's level
of quality get close to P's?  It might, but is that necessarily going
to happen?

These days I incline to an infrastructural view: if a product is
infrastructure--very widely used--then a high-quality free version is
very likely to appear, one way or another.  In other cases, whether a
high-quality free version appears or not is a matter of chance.

> Unfortunately, my primary economics education comes from reading Paul
> Krugman's columns, so I'm not sure how to develop my theory into
> something that would convince the pros -- and I'm afraid that if I
> tried, I would waste a lot of time reinventing wheels.  If one of the
> business or economics gurus on this list could give me a list of
> background reading, or even a background vocabulary ("the differential
> equation that captures all of the variables involved in a business
> reinvesting its profits is known as Mumblefrotz's Formula"), I would
> be very grateful.

Most of what you are talking about is micro-economics, which I tend to
feel is basically common sense.  Economic texts are about as likely to
confuse you as to enlighten you.  Economics is not a rigorous
discipline: if you can tell a convincing story, then you can find math
to support it.  Look at the Laffer Curve, for example.  I suspect that
there aren't enough examples of free software projects competing with
proprietary projects for you to produce any relevant statistics.  (I
may be slightly biased since my father is a professor of economics,
but I think I'm right.)