🚀 go-pugleaf

In article <RNB0aB22w164w@mantis.co.uk> mathew@mantis.co.uk (mathew) writes:
>So I'm casting doubt on the assertion that the GPL can compete with 
>conventional methods of funding development of special purpose software,
>not asserting that the GPL is entirely useless.

Even Stallman admits that his approach can't really "compete" with
commercial software development organizations.  His proposal is that all
software should be free, and software developers should be paid by the
government out of a software tax.  To prevent some of the traditional
problems of socialism (if one is paid regardless of whether one produces,
there's no incentive to produce), I presume that software developers would
be paid based on their output.  As long as software developers are allowed
to profit more by restricting distribution of their programs and the
source, there isn't going to be a huge market in free software.

Presently, hardware manufacturers are *encouraged* to keep their hardware
design secret so that they control who can develop optimized software.

>>                                                                        HP
>> can get away with what they do because no one has yet had the inclination
>> to write a copylefted version.
>
>Then if we have to rely on "inclination" to encourage software development 
>rather than monetary reward, we'll be waiting rather a long time, won't 
>we? :-)

Well, we always have to rely on inclination.  Even people who get monetary
reward have to be inclined to write the particular piece of software that
we want.

Software vendors are mostly inclined to write general purpose programs that
run on the most computers.  Hardware vendors are inclined to write programs
specialized to their systems because it helps sell the systems.  And, as I
pointed out above, they have incentives to make it hard for others to write
that software.

>> And what about my point about the possible undocumented hardware
>> interfaces?  If this were true, it would be virtually impossible for a
>> third party to write a good X server without licensing technology from the
>> system vendor.
>
>In which case we would see good, free X servers for the documented systems,
>and expensive proprietary X servers for the undocumented systems.  And people
>would buy the former over the latter, would they not?

Most customers don't really care about the source code, so market forces by
themselves won't result in its availability (if there were that many people
who wanted source code, market forces would already have had that result).
In particular, when evaluating workstations, most customers' criteria
include performance rather than available free software.  Thus, releasing
source has to be done out of a sense of "rightness" about it, not greed.
It has to come from the realization that there's a small segment of the
customer base that's better served by providing source.

>Or is there some conspiracy by hardware manufacturers to ensure that their
>systems are badly documented so that we all have to buy their proprietary
>X servers?

I don't know how much of a "conspiracy" it is, but there isn't a whole lot
of incentive for hardware vendors to release their schematics and device
interfaces.  I remember when people first started thinking about turning
old Sun 3/50's into dedicated X terminals, the problem was that the display
interface was not in the documentation; Sun-3 X terminals were eventually
implemented by configuring a small SunOS kernel and running the existing X
server as a process, rather than by replacing the kernel completely,
because the SunOS kernel includes the display driver.
-- 
Barry Margolin, Thinking Machines Corp.

barmar@think.com
{uunet,harvard}!think!barmar