In article <C789M2.
...@cscns.com>, m
...@cscns.com (Mike Jones/Mountain Alternative Systems) writes:
> The problems of unrestricted software
> It is tempting to say that the major problem is quality.
> However, after using the Linux system for a while, I find the
> quality very good. So, if the problem is not quality what is it?
> Two things. First, stability. The system changes fast is is
> difficult to keep up with. You have to be a genius and a gorilla to
> keep up with all the goings on about Linux. Second, there is no
> guarantee of getting help when you need it. Just last week there
> was a news posting begging for help a third time. People are
> irritated when someone askes a question that was answered in a FAQ.
> This intimidates the novices real quick. There is also no single
> place to go for answers with any assurance you will get an answer.
> You are depentent on the good will of others for help.
I think you have confounded the idea of "freely redistributable"
software with the idea of "no-money-transfer" software. This
confusion has led you to consider two possible drawbacks of
"unrestricted software": (1) instability, and (2) no guarantee of
support. But this is by no means inherent to free software! In fact,
it isn't even alien to restricted software!
Let us take the GNU+Linux phenomenon as an example. With the possible
exception of the ever-so-popular emacs and gcc/g++, this software has
been a hacker-only enterprise. This is because until recently, the
only means by which it has been distributed is via sources on some
tape or ftp site with a "Good luck! Let us know how everything goes."
thank you note. For the most part this hasn't changed. Programmers
would prefer to use their resources hacking and incorporating
contributed patches into their programs instead of conducting PR
campaigns and answering 10 newbie questions a day on the net.
This is where the money comes in. The emacs distribution comes with a
file ("etc/SERVICE" i believe) which lists people who would be willing
to provide competent and guaranteed support for a fee. And recently a
company called Cygnus (Cygnus, Your GNU Support) has been making its
money by supporting and developing parts of the GNU development suite
for a fee! Apparently they are doing very well. So unrestricted
freely redistributable software and guaranteed support for a fee are
NOT incompatible. They have not been the norm only because it has
been the domain of individual hackers (as opposed to companies with
purchasing dept's) who couldn't afford support they could use even if
it were made available.
The instability of free software can be similarly explained. Until
recently, free software has been used almost exclusively in an
environment where it doesn't matter very much if some random utility
breaks. Why? Because there is usually someone around who can fix it
(even if that means reverting to an earlier version), and because the
operations that the software supports are rarely so critical that it
can't wait an hour or a few days to get fixed (for free). However,
the demands of the real world are unlikely to give computers so much
slack.
This is where the money comes in. If one were willing to pay for it,
I am more than convinced that any free software support house such as
Cygnus or Softlanding would provide a service by which you could be
given as much stability as you could pay for. Take gcc, for example.
gcc-2xx has been out for more than a year, but those of you who want
the stability which gcc-2xx doesn't already provide, you can get
gcc-1.42! And this is freely (as in price) provided at cost to the
FSF! I can't even begin to imagine the potential for making money if
there were a greater demand for a stable selection of free software.
Now let's consider the issues of support and stability in the realm of
commercial but restricted software. The convetional wisdom in the
Linux crowd about real world support is that it is non-existent.
Typically, one will be put on hold and forced to listen to some
aggravating elevator musak for a few minutes until one of the
engineers frees up. Then you explain your problem OVER THE PHONE,
and he'll try to hack up a solution while you wait. This is often
prolonged since you have to do your own trouble-shooting ("Could you
read me your AUTOEXEC.BAT?"), and you have to play a sort of mediator
between the tech-support line and your misbehaving computer. And all
of this is if you're lucky! Often you won't have the luxury of a
toll-free number, or your "free" support will have expired last month,
or the guy on the phone wasn't a developer, or you'll get the dreaded:
"Sorry, we no longer support that."
Stability is not that much better. How many times a day would Win3.0
give you the ol' "Unrecoverable Application Error" before forcing you
to reboot? (Oh sure, just blame it on the apps, won't you!) And how
much money did you pay for the "upgrade" to 3.1 which fixed this bug.
Assuming you do more than start up your copy of WordPerfect each time
you start up DOS and stay there most of the day, how often do you
reboot your DOS box? How often is this the solution proposed when you
call tech support to report a problem? How about the old "Abort,
Retry, Ignore?" deal? (Try this at home: disconnect your printer and
"accidentally" type Ctrl-p at the DOS prompt. Now try to abort.)
The reason commercial software companies can afford to provide such
shoddy support and maintenance to their products is that once sold,
you have only ONE source for it. Although it's not strictly a
monopoly (you can always buy Dr.DOS or OS/2), once you've made an
investment (say 100 copies of Windows) the effect is quite the same.
> Future discussion
> I would like to see some discussion about where boundries
> should be drawn between unrestricted software development and
> commercial development. I would also like to see some discussion
> about how the two systems can compliment each other rather than work
> against each other.
i hope i've done my part to explain why we may possibly not need
restricted software at all. I suggest we consider the possibilities
of *unrestricted commercial* software.
rodrigo vanegas
r...@cs.brown.edu
ps. i've cross-posted to comp.os.linux. hope no one minds.
