Here is a long ramble about how, over time, open-source is likely to
colonise areas which are initially dominated by closed-source software.
When there is a really big fire, I understand that the first forest type
to re-establish itself is the mono-species eucalypt forest, typically
with some kind of understory of a few smaller species. In the wet
gullies, the rainforest species survive. Slowly, over decades and
centuries, the rainforest encroaches on the mono-species eucalypt
forest. The rainforest's combination of many species - large trees,
small shrubs, vines, staghorns etc. together with their rich variety of
polinating insects and birds, traps more moisture, uses sunlight more
efficiently, retains more nutrients and is less susceptible to fire.
It spreads slowly up to limits imposed by dryness.
In some places, equilibrium is reached. On the Lost World Plateau, in
the western reaches of Lamington National Park in southern Queenlsand, I
found a clearly defined border between the two forest types. To the
north was open eucalypt forest with grass underfoot and occasional grass
trees. To the south was full-tilt sub-tropical rainforest. At one
point were fully mature eucalypts to 20 metres high, with lots of light
falling to the ground. Literally two metres away, it was dark, with a
rainforest canopy, vines, rotted leaves on the ground etc. - it was hard
to take a photo there in broad daylight!
It must have taken centuries to reach that equilibrium.
It was peaceful and wonderful. I was the only person on the plateau.
There are no roads there and hardly anyone even drives to the nearest
road, a few kilometers walk and 2,000 feet or so below. One grass tree
was so old that towered over me and had 13 heads. I trust is is still
there - this was in 1976. A plaque on the goat track up the razor-back
to the plateau commemorated a walker who had lost her footing and fallen
to her death. Very peaceful and pure . . . until an F111 roared
overhead, no-doubt doing the roller-coaster trip with its vacuum-tube
based, 1960 technology, terrain-following radar!
Its early days for software, and if Mozilla is any guide, the
open-source development of really large projects can be painfully slow.
As a devoted bug reporter, what I witness there is lots of thinly spread
developers hacking away at a vast codebase which is not a pretty thing
to look at. Its almost an evolutionary process as features get added,
stuff up other things, those stuff-ups get corrected, which cause
further regression in other things etc. But slowly, the Lizard rises!!
But over time, I think there is a one-way process of people writing open
source for any one of a number of reasons, and then that becoming the
best, de-facto standard, program to use.
It seems impossible to contemplate that a commercial program will ever
become more widely deployed than Apache. Sooner or later, Open Office
and Mozilla will be so strong that their deployment grows to the point
of dominance and perhaps ubiquity. I guess that can happen with the
operating system too, though the habit of one OS is hard to break if you
have to go cold-turkey to adopt another.
Commercial software will always have a promotional advantage over
open-source, because there is money and financial incentive to promote
it. Also, with proper investment and management it will usually be
possible to get a team of full-time developers in a Skunk Works totally
focussed code-fest to create a really good program in a much shorter
time than could be done with a geographically distributed volunteer
effort.
At present, many people are wary of relying on a program that comes for
free - despite not realising that virtually every nameserver of
importance on the Net is open-source, and likewise most web-servers and
the operating systems running them are open-source too. Part of the
problem is that no-one plasters the world with promotional material for
all the open-source software which everyone relies on day-to-day in
ordinary email and web-based communications.
How many non-technical people know that, at least as far as I know, one
of the primary reasons Microsoft's NT and Win2k etc. actually works and
remains stable on a LAN is that they used the BSD TCP/IP code. How
pathetic it is that Microsoft, with all their billions, can't write
clearly defined protocol stack software, uses open-source BSD's, and
then spends millions trying to stymie open-source software!
As people slowly wake up to how a monopoly behemoth like Microsoft can
foist software on the world which is so full of security problems which
cause genuine losses for their company . . . and as open-source
alternatives such as Mozilla and Open Office develop momentum . . . an
general understanding will develop that paying for closed-source
commercial software is not always the best approach. I predict this
process will hasten once Microsoft stops being seen as a raging success
- when its share price start plummeting as their sales and sales
projections diminish and as the over-blown share-market gambling on this
currently prized stock turns into a rush to sell.
Meanwhile, I hope that Linux / X and the Mac OS finally get proper
support for the scroll wheel on trackballs/mice. That is one thing I
really like about Windows! Likewise, I wish Windows drivers and X get
properly adjustable, useful, speed and acceleration capabilities like
the Mac (and to a large extent the Logitech Windows drivers) have always
had.
As particular software needs become relatively mature and stable - such
as word processing, spreadsheets, image-graphic editing etc., the
initially dominant commercial products face a losing battle with
open-source.
There will always be the need for software to perform novel functions.
In fields where people don't exactly do things for fun, such as
accounting, CD-R burners (and all the work of constantly trying to
support new drives), it is likely that commercial software will dominate
initially. But over time, someone will get sick of paying for something
that doesn't do exactly what they want, so they write their own and make
it open source - for the hell of it, and/or to get help from other
programmers and maybe pass it on to someone else in the long term.
There is a danger with both commercial and open-source software that
what becomes standardised may be just barely good enough, rather than
something really brilliant. There are examples of brilliance in the
commercial field, such as Adobe's Postscript (something I think only the
insanely brilliant and focussed would write for fun) and . . . maybe the
initial Visicalc spreadsheet programs, the first vector graphics
programs (Illustrator, Corel Draw, Freehand) and probably the extensive
depth of Photoshop, though I am not sure that it is brilliant and
inspired. The CD (Sony/Philips) and Postscript were, I believe,
elegant, truly inspired, excellent, bold and very early developments
which made any other work in their field pointless - so saving us all a
*lot* of trouble. They were both commercial developments, with the
acutely focused research efforts which substantial money and good
management can sometimes provide, and which to my knowledge has not been
matched on such large-scale projects by any volunteer effort.
But PGP, the GNU compiler etc. Linux, so much of the infrastructure
software of the Net (name-, mail- and web-servers) were open source
based on the work of individuals or small groups of people with funding
from universities etc. Of these, only Linux and the GNU compiler and
its friends were really daunting projects to start with - and it took
years for them to mature.
I am not suggesting that commercial software always establishes itself
first, but I think there is a one-way process by which, sooner or later,
open-source is likely to colonises many or all areas which commercial
software initially dominated by virtue of its rapid, investor-driven,
tightly focussed development approach.
As far as I can see, there's no equivalent in software to the
all-consuming forest fire.
I like to think that the following analogies will generally hold:
Rainforest overcoming single-species forest, to a certain point.
Open source overcoming commercial software, as long as the results
are really robust, clearly conceived and elegant.
Truth overcoming lies and ignorance, but not to the extent to prevent
safely blissful ignorance and fun-and-games.
Love and trust overcoming unjustified hatred and mistrust.
Peaceful and harmonious methods of human existence and co-existence
with other species replacing the more destructive approaches of
the past and present.
Overall, I think there is progress in all these fields, but the
population explosion, HIV, anti-biotic resistance, resource depletion,
climate change and the failure of many educated and bright people in
industrialised nations to reproduce makes me worry about the last one
turning out really badly in the next century or two.
At least, by then, computers running open-source software at tens of
gigahertz will come as promotional items in Weeties packets!
- Robin
----------
For Link list information see http://sunsite.anu.edu.au/link/
This archive was generated by hypermail 2.1.1 : Sat Aug 31 2002 - 03:10:03 EST