[svlug] Re: Open Source Criticism Questions

Sanatan Rai sanat at stanford.edu
Wed Aug 20 21:46:59 PDT 2003

: > All open source programmers are relieved from all
: > responsibilities, and owe absolutely nothing to
: > anyone, unless you specifically hire one for
: > development or support.
: That seems to be the consensus.
: > If a customer wants support, there are certainly
: > companies that will sell you a support contract.  If
: > you want a feature, there are people you can hire to
: > add the feature.  But it's completely inappropriate
: > to ask a developer to provide this as a
: > "responsibility".
: That seems to be the consensus as well.
: > Is that really a free ride?  On a value-for-price
: > basis, it is impossible for open source software to
: > be a bad deal.  Not so for most Microsoft products,
: > which can't even be returned for a refund if you use
: > it for a while and decide you don't like it.
: A reasonable argument, I think.  I disagree but don't
: think that it is worth arguing over.

	I should point out at this juncture that Open Source as a
movement has its roots in Universities. It has been seen that academics
tend to be happy to share their ideas with others. As a case, it is
interesting to consider TeX.

	For those who might not be that familar with the history of TeX:
TeX is a document _typesetting_ system developed by Don Knuth in the
late '70s because he was dissatisfied with the quality of proofs
supplied by his publisher. He was in the process of publishing the first
volume of The Art of Computer Programming, and Wiley's poor typesetting
drove him to develop TeX.

	As a computer scientist, it was of immediate intellectual
interest to him to abstract the fundamental algorithms and to quantify
seemingly subjective judgements on aesthetics. He learnt the principles
of typesetting and font design and produced TeX and MetaFont.

	As a software programme, TeX is hard to beat. At this point
there are no known bugs. If you find one, send Knuth a report, and he
will send you a cheque. The code is available freely: you can download
the .web file from any of the CTAN archives.

	He discusses the code in detail in the book he wrote about it.
It is likely there better algorithms are now known for many of the
things a typesetting system has to do, but at this moment, there is no
comparable piece of software that implements them all, lout included.

	The cost of downloading and installing TeX is 0.0. It is the
user's responsibility to maintain and learn TeX and its variants. There
are a few companies who make their living by selling products based on

	TeX is therefore the supreme example of Open Source software:
free, and a canonical example of the high quality that Open Source
developers are proud of.

	Naturally, Open Source is not a company, and therefore there is
no control over what is available. Programmes written by academics of
the calibre of Knuth or by experienced and dedicated people are of a
high quality, from TeX to Samba to Apache to the Linux kernel.

	There is of course a growing number of enthusiasts, who write
code and share it with others. Not all of it is of a high quality. So
before using a programme one has to excercise one's judgement.

	The same holds for commercial software: the final determiner of
quality is the people who write it. The fact that they make a living out
of it only means that to the extent their licences require them to, they
provide support and bug fixes. Therefore, given that the standard MS
license provides no warranty, it is not surprising that their software
is so buggy and bloated. In fact it is surprising that it works at all.

	Therefore, the correct style of criticism for Open Source
software is the academic style. Be objective and to the point. For an
example, search Google for a criticism of the GNU Numerical Library by
one of the developers. This is reasonable, because Open Source follows
the model of academia: one is judges on originality and quality. These
are not the criteria that are used in a trade magazine.

	As far as commercial software is concerned, there is this big
component in its production and promotion that comes from `marketing'.
No one `markets' Open Source software. Some companies base their
livelihood on Open Source software, but that is not the same thing. So
you are welcome to say `Red Hat 9.0 sucks.' But if you have complaints
with the Linux kernel, then provide a carefully argued paper that
describes in detail but with precision why you think that the Linux
kernel is not a modern kernel, what you think is the flaw in the
approach and indicate better algorithms. Ask youself the question: would
Andrew S Tanenbaum consider my paper worth his time?

	Eventually the pupose of criticism to foster improvement.
Commercial software can perhaps be improved by abuse and litigation. For
a model based on the academics spirit, it is the academic style that
should be preffered.


Sanatan Rai,                      | EMail: sanat at stanford.edu
Dept. of Management Sc. & Engg.   | Home: 143D Escondido Village,
Stanford University,              |       Stanford, Ca 94305.
Stanford, Ca 94305.               | 'phone: (650) 498 1655 (R)
                                  |         (650) 736 2109 (O)

More information about the svlug mailing list