RFC soon on essay "Does Free Software Production in a Bazaar obey the Law of Diminishing Returns?"

[Jacques Chester]

>X-Eric-Conspiracy: There is no conspiracy

*Coughs politely*

>Jacques Chester <thunda at manor.downunder.net.au>:
>> Fifthly, the possible conclusions so far are:
>> * That ESR is completely correct, that Free Software *does* break 
the 
>>   LODR and that it represents a new economic phenomenon in 
production
>> * That ESR is completely wrong, that Free Software *does* obey the
>>   LODR, it just happens to be far more 'scaleable' than the 
traditional
>>   "Cathedral"
>> * That ESR, unaware of the economic framework in which this can be
>>   viewed, does not realise that Brook's Law is the same as the LODR;
>>   and has therefore failed to distinguish between production in the
>>   Short Run (where the LODR applies) and the Long Run (where it does
>>   not).
>
>My take on the situation differs from any of these three :-).

I was wondering how you'd view an outsider's slant of your
specialty. Those possible conclusions were, of course,
*possible*. The caveat still stands that they may be wholly
incorrect (for several reasons).

>I'm not familiar enough with the technical literature on LODR to form
>a firm judgement on whether Brooks's Law is equivalent to LODR.  My
>gut reaction is one of skepticism.

I'm afraid that, having delivered a sermon on quantitatively
testing your qualitative assertions, my own attempt to draw
approximation between Brook's Law and the LODR is itself
qualititative. Hypocrisy of the highest level, and great fodder
for the Caveats appendix.

>I also want to point out that I do not assert anywhere in my writings
>that open-source development is immune to LODR -- for the very good 
reason
>that I would coinsider any such claim patently ridiculous!

So would I, and so would any economist. That's why, if a
relationship between Brook's Law and the LODR is established,
the subject ought to be investigated. Even if the answer
is deadly dull, it's worth checking anyway :)

>I'll put the case more positively.  I strongly suspect the following 
things:
>
>1. Bazaar-mode development *is* susceptible to LODR effects, but (in 
your
>   own words) "happens to be far more 'scaleable' than the traditional
>   cathedral".  This, in fact, is almost exactly how I would have 
phrased
>   my own answer if the question had come up before.

The preliminary Total Programmers vs Average Total Output
seems to concur with the "highly scalable" conclusion. To
the naked eye, it seems to be a slight curve upwards, and
this curve itself seems to match the beginnings of a
stock-standard LODR curve.

But the gotcha is simple to pick. We cannot say at all that
it *is* the beginning of an LODR graph, because for all we
know it could be any of an infinite set of graphs with that
beginning. It is just that we *expect* to see what we are
seeing. And that's all I can say.

Despite the selection of GNOME as a data source, I am afraid
that my conclusions will be tentative, at best. While some
elementary guesses might be hazarded, the data set is just
too small to make any concrete statements.

Alas, GNOME is one of the largest community-based OSS projects
that uses CVS. I would have liked to have used GNU or Linux data,
but such data is not available in the high resolution of a
CVS logfile.

>2. Brooks's Law is not precisely *equivalent* to LODR, but is rather a 
special
>   case of it involving *particular* nonlinear scaling phenomena.  
Accordingly,
>   one may assert that the bazaar mode repeals Brooks's Law without 
making
>   any commitment about the applicability of the LODR in general.

And, conscious of the theory of knowledge, I too would suspect
that Brook's Law is a special case. In an earlier draft, I said
so. For simplicity, however, such a reference is removed from the
main body of the essay and into the Caveats section.

This is for efficiency of reason. Logic, of course, is the act
of creating perfect structures from imperfect ones, which
implies a certain degree of fitting fuzzy cubes into velcro
circles. It doesn't always work cleanly, and stuff sticks
where it shouldn't. For convenience, and in a conscious
decision, we choose to carve off the edges so that the basics
fit.

With the essay, I seek to create an equivalency between the
LODR and Brook's Law. While I do think Brook's Law is one
special instance of the LODR, I also think it inherits
enough of the core function to be "approaching equivalence".

No, it is not the same. But, for the purposes of the main
body of the essay, it will be. Why? Because this allows us
to bring to bear some of the economic analysis that we
use when dealing with the LODR. I am only a student of the
most elementary and fundamental economic theories, but
even I have a toolkit that seems to fit the problem at
hand.

This issue will most certainly be addressed in the
Caveats. Indeed, the Caveats might be more numerous than
any other part of the essay.

Thanks again;

JC.