[ox-en] open source and national interests (the australia case)
- From: "geert lovink" <geert desk.nl>
- Date: Fri, 8 Aug 2003 15:20:41 +1000
Australia & Open Source Software - Responses
Tony Healy, Aus Innovate, 27 July 2003
Thanks to the people, including parliamentarians, who provided some views on
open source software. The contributions were of quite a high quality. Here
is my response.
One parliamentarian pointed me to an interesting paper regarding the plight
of Argentina and other South America countries with respect to participation
in the software industries and access to modern computing platforms.[1] I
sympathise with Argentina's plight as described in that paper. If that's
what inspired Congressman Nunez of Peru, then I salute him. However my main
points remain.
1. Is Open Source Profitable or Not?
------------------------------------
Many people picked up on the seeming contradiction where I bemoan the impact
of open source on software developers, yet talk about foreign outsourcers
gaining hefty consulting and support fees from open source. The point is
that they are different types of business.
Installing and configuring software products is low-skill work, completely
different from software development. It will not help Australia at all.
Although services cover more than this, including some custom software
development, they are generally not as involved as developing software for
the marketplace. Services are dominated by large foreign outsourcers
including IBM. Increasingly those companies outsource even that work to
cheap foreign labour. By comparison, Australian companies who build useful
software can maintain their place in the economy, provide valuable
intellectual capital for the future of Australia and gain export dollars.
This point is made well in a German report predicting healthy growth in open
source software sales, which a parliamentarian pointed me to. That study
includes and even emphasises services: "When we talk about sales of open
source software, what we mean, essentially, is a service that a vendor
provides in copying the software on a CD and distributing the disc to a
customer," said Soreon company spokesman Christian Lipski.[2] Services also
figure strongly in Congressman Nunez's belief that open source would grow
Peru's software industry.
Robert H challenged my point that Linus Torvalds, after creating a product
worth billions of dollars, must still scrabble for a job. Robert pointed to
Linus having numerous job offers. The point is that Torvalds needs to take a
job with someone else at all. By comparison, Bill G hires people. In
economic terms, the creation of Linux has been a process where wealth
created by programmers was transferred to the corporations that use Linux
without paying any fees to those whose work created it.
Robert also points out that one motivation for open source contributors is
that it lets them work on projects they find interesting, as compared to
their day jobs. But if those open source contributors protected their work,
it would become a saleable product and that would become their source of
income. They would have created a business.
2. Australia Must Protect Our Software Competitiveness
-------------------------------------------------------
I mentioned that open source actually benefits a number of big foreign
corporations, including IBM and other foreign outsourcers, and including
hardware vendors Sun, IBM and HP. Those companies are actually the ones
funding the organisation that now employs Linus Torvalds. (Open Source
Development Labs www.osdl.org)
Those companies do not do this out of the goodness of their hearts. They do
it because, firstly, open source software increases the availability of
software and thus the demand for computers. Deleterious effects on software
developers are not important to hardware companies, but they are to
Australia.
Secondly, open source software generally runs on a wider range of computer
types than Windows, thus increasing the market for expensive workstations of
the type sold by Sun, HP and IBM. Contrary to all the hype, particularly
associated with Java, this type of cross-platform capability in software has
costs for end users. Cross platform programs are slower than programs
written specifically for Windows, and often have sub-optimal user interfaces
and bugs associated with trying to straddle diverse platforms. So-called web
service applications often involve an enormous amount of bloat, especially
if they involve J2EE and especially if developed by poor (read cheap)
programmers. Users and government lose big time with some cross platform
targeting, particularly in emerging handheld markets.
The strong involvement by IBM and others in funding open source development
nowadays also produces another question not addressed by open source
advocates. Why is it acceptable for IBM to make a profit but not for
software developers?
If we applied this attitude to other industries, we would say that movie
makers should provide their movies for free, and accept what they can get
from advertisers. We would say that journalists should provide their stories
for free, and accept what they can get from public relations firms. Clearly,
those groups would not see that as beneficial, and neither should software
developers.
3. Licences Don't Protect Small Developers
------------------------------------------
Several people cited licences as protection against copying. Licences might
work for massive pieces of software, especially well-known ones, but they're
useless for protecting smaller products such as those by Australian
developers, where it's easy to change the source code and feasible to claim
the product has indeed been developed by the firm that's selling it. Robert
H's comments about detecting copying probably relate to large, well-known
code bases. They also seem to relate to cases where attempts to mask the
copying were amateurish.
The hard work in developing software is in resolving the complexity of a
requirement into a working design, and then building the complete product so
everything works. Once all that work is done, it is very easy to take the
resulting source code and modify it so the origins are not provable. In any
case, the freeloaders wouldn't have to even make their own source code
available for inspection.
Robert H asked why someone would bother copying source code if it was
already available for free. His point was that surely there would be no
market for a commercial product. The most relevant answer is that large
companies can capitalise on such work and build markets much faster than
small ones, thus smothering innovative Australian companies. This occurs in
consulting too. Large consulting companies that obtain rivals' source code
can suck up all the relevant work and leave the original developer broke.
Tom W asked this question from another angle: why would someone copy code
from an open source product if, as I contend, a lot of it is junk? The
answer is that I don't believe it's all junk; I said some is fine but most
is junk. More importantly, I am talking about the danger of government
mandating open source for supply to government, which would require
commercial developers to reveal their valuable source code.
In another context, Richard C actually pointed out that commercial firms are
known to have copied code from open source products. Richard C was
misinterpreting my point. I don't argue commercial firms are more noble than
open source providers; far from it; I argue that disclosing source would
harm the best software developers.
Tom also compared the disclosure in open source with that required in
patents. However patents are generally for inspectable processes and
machines, usually of high or distinctive value, and infringements are
relatively easy to prove. With software, infringements are much harder to
recognise let alone prove.
4. Yes, Switch To Public Data Formats
-------------------------------------
I agree and am a strong supporter of using public formats for data. Open
source is not necessary for this. Government just needs to define the
formats and provide validation suites so developers can work with them. A
corollary is that government would benefit from having a much higher level
of software development expertise in a lot of business projects, to ensure
that formats were useful and that there was a good environment for external
suppliers to work with.
In particular, government should not rely on outsourcers to do this sort of
work because, in my experience, their work is poor, they show poor
understanding of the principles of good governance, and they generally skew
environments against Australian firms. Would that we have a Senate inquiry
into these issues one day.
5. Helping Third World Countries
--------------------------------
I entirely support the idea of third world countries using the cheapest
platforms that work and would see merit in wealthy nations funding projects
to see that they obtained what they need.
6. Is Developing an OS Hard or Easy?
------------------------------------
A few people commented on the apparent contradiction in my claiming that
developing a modern desktop operating system is a huge task, and then
describing the development of Linux as a relatively straightforward task,
due to operating systems being well defined and understood.
With the first point, I was commenting on the view that replacing Windows
might provide an opportunity for Australian developers. I think some
parliamentary supporters of open source have or had this view. I was making
the point that it's not feasible.
With the second point, I was commenting on open source as a development
methodology. Public open source methods worked fine for Linux, Apache and
other platform products, but they would not work for, say, a new transport
system for NSW.
7. An Important Distinction
---------------------------
A significant part of open source enthusiasm, particularly recent enthusiasm
at political levels, is simply for using publicly available free software
such as Linux. The availability of the source code for that software is
peripheral to that enthusiasm, even if understood as being important. I have
no philosophical objection to this at all, although I do think there will be
practical problems.
The part that concerns me is the expectation that source code should be
provided for all government software, including valuable specialist
applications, and the accompanying views that this is part of a superior way
of developing software. I contend that this won't work and is dangerous.
Robert H did point out that commercial software can still be provided for
Linux platforms, while retaining the source code, in the normal commercial
manner. Parliamentarians exploring open source policy, if they agree with
this principle, should make it explicit.
8. Trusting Telstra
-------------------
I pointed out that concern about software is inconsistent given that
government routinely trusts many other parties. It trusts Telstra not to
intercept phone calls. It trusts couriers not to photocopy confidential
documents, and it trusts accounting, law and PR firms not to divulge
confidential information.
Robert H claims accounting and other firms are audited. But that's of little
use in stopping disclosure of confidential information. In any case, frauds
at Enron, disasters at HIH and findings by the US Supreme Court have shown
us just how trustworthy accounting firms such as Arthur Andersen really are.
"Phone-home" programs are cheap junk things easily avoided by informed
people. They are not the type of program bought by government buyers.
Government has software acquisition procedures that verify that programs
work as required, and it has network managers able to detect untoward
activity, so I actually think risk from funny code is a furphy. Indeed, the
diversity of open source code, and the greater ease in providing a package
for distribution and installing it on a production server, could actually
represent a bigger threat to government.
However I do make an exception in the case of voting software. Just as the
business procedures associated with an election employ rigorous supervision,
well beyond the scope of normal business of government, so too I believe
that all parts of a system used in recording and counting votes should be
open to scrutiny. In fact, I believe that level of scrutiny should be much
higher than just open source. Such systems should incorporate additional
public viewports so that, for example, a remote user could follow the path
of any randomly selected, anonymous, vote, and inspect various aggregate
data about polling booths and so on.
9. Summary
-----------
The fundamental points in this discussion are that open source does not
really provide protections for software developers, and thus it destroys
valuable business opportunities for Australia. The debate generally fails to
acknowledge important distinctions, particularly the difference between
deciding to using public software and then mandating open source as a
development methodology for all software. Finally, parliamentarians must be
much more careful in analysing competing interests in the technology
industries.
--
1. Federico Heinz and Oscar E. Heinz: Proprietary Software and
Less-Developed Countries - The Argentine Case
http://www.vialibre.org.ar/index.php/article/articleprint/15/-1/8/
2. John Blau: High Growth rates for open source in Germany, IDG News
Service, July 02, 2003
http://www.infoworld.com/article/03/07/02/HNgermanopen_1.html?platforms
3. Villanueva Nuñez: Reply by Villanueva Nuñez to Microsoft, Lima, April 8,
2002. Translated by Graham Seaman April 28, 2002
http://www.aful.org/politique/perou/english/rescon_en.html
Regards, Tony Healy
-----------------------------------------------------------------
Aus-Innovate, Sydney
Aus-Innovate is interested in the development of Australian R&D and
software, and the furthering of professional software engineering interests,
including workplace issues. Info: thealy at siliconchic.com
_______________________
http://www.oekonux.org/