Re: Open Source (was: Re: the economics of transition to nanotech)

From: Robert Bradbury (
Date: Sat Jan 15 2000 - 11:26:00 MST

On Thu, 13 Jan 2000, Rik van Riel wrote:

I agree with most, if not all of the things you mentioned
regarding the advantages of open source. I'll just pick
a few nits...

> If the complexity is too great, the original design was wrong.

The complexity can be too great because (a) the underlying
hardware base keeps evolving & expanding (ala the PC platform),
or (b) the people really didn't understand the problem when
the started and a lot of "R&D" had to be done before the
complexity became apparent. I'll cite two cases where
I have extensive knowledge -- compiler design and databases.
In the first case we are *still* evolving compilers 30+
years after their development and in the second case we
are still learning about optimal query languages, data
organization, indexing approaches, object clustering, etc.

The complexity is inherent in compilers and databases because
of the variety of tasks they may be asked to perform. As you
continue to "add-on" to these programs they get more and more
complex until you have to throw out what you started with
and rethink things. I'd cite the human genome as another
example, though humans didn't have their hands in that one...

> Open Source can, and does, take the time for this.

Now you point out one of the down sides which is the long
lead time until you see working code. I think Perl 6 has
been in cooking for 2+ years now. While having open source
produced primarily by a single individual with the big picture
(or groups if the problem can be sub-divided), in the real
world there are people who need the problem solved yesterday
and are willing to go pay someone to solve it. When you
are paying for it, thats when you start to get into the
discussion of "ownership" and "property rights". What
eBusiness startup wants to do a Java application for order
invoicing and have it open source for the next competitor
to use?!?

> Problems that cannot be solved by the normal support people are
> fixed by the people that wrote the software in the first place!
This is a very important point, but you always have the problem
of people moving onto other things and becoming unavailable.

> True for applications, not true for the operating system or the
> tools themselves. The kernel, gcc and glibc people (to name a
> few) are still programming at a fairly low level.

I'd suspect most of the kernel work is going into external interface
components (e.g. filesystems, device drivers, installation scripts,
etc.). Gcc has got a new architecture to contend with coming down
the road. I doubt very much work is being done with glibc.

> ... hindered by timezones and no longer by company-mandated secrecy
> or geographical distances.

While I agree you can do a lot over the net now that you couldn't do
before, there are many problems that are best solved by brainstorming
with all the people in the same room. There may be people who can
do this effectively over the net, but I have *yet* to see a public
net multi-time-zone interactive white-board videoconference that works.
Only if those individuals work in locations where you have some very
pricey hardware can this be made to work currently.

But otherwise a nice commentary on the open source approach.


This archive was generated by hypermail 2b29 : Thu Jul 27 2000 - 14:02:17 MDT