🚀 go-pugleaf

edguer@charlie.CES.CWRU.Edu (Aydin Edguer) writes:
>I am troubled by peoples' perceptions of "FSF's inabilit[ies]".  The
>FSF has the _ability_ to do almost anything _they_ want.  They do not
>always have the _desire_ to do anything _you_ want.

Don't confuse theory and practice, and don't confuse criticism and
desire.  I believe that claims about the ultimate wonderfulness of a
system of Free software are poorly justified, and thus I criticise
them.  I also realize that *in theory* the FSF can write whatever
they want to, but in practice their desires are not at all the same as
the desires of people who purchase (for instance) PCs, Macintoshes,
and Nintendos (actually, I'd expect the greatest overlap for the
Nintendo case).  Thus, no "Free specialized mass market" software.

>These people refuse to acknowledge (or just don't know about) things like:
>
>	sc (a PD UNIX spreadsheet)
>	oleo (the FSF spreadsheet - X11 interface planned)
>	xfig (an X11 drawing program)
>	andrew (an X11 multimedia kit - includes WYSIWYG editing)
>	RHETORICAL (planning ai software)
>	PROTOS (knowledge acquisition ai software)
>	DANCAD3D (a PC CAD system - no source)
>	pcb (a printed circuit board router)
>	magic (a VLSI design tool)
>	JACAL (a symbolic math program)
>	Solve (a PD Mac Mathematical analysis program - no source)

I've used xfig.  Vomit.

I've used Andrew.  Too big and slow, though pretty nifty in design.
  (Though to be fair, I haven't seen a non-Free *Unix* system 
   resembling Andrew in any way that I liked very much, either.)

I'm not too keen on anything based on X11.  Too big, too slow, the
  design decision to be window-manager-neutral doesn't do anything
  good for the user interface.

My general feeling is that this software is prototype quality, but
lacking polish in the details.  In my experience, the details are the
hard part, partly because they aren't fun, partly because there are
always more of them, and partly because sometimes there isn't a clear
right answer.  Also, the details (in good GUI software) are incredibly
important.  Also, good mass market user interfaces run on tiny
underpowered machines and make X11R4 look like a slug on a much bigger
machine.  That tuning is typically not fun work, so there is a good
chance that it won't get done.

>>Who pays for efficient word processors with well-designed GUIs and
>>tastefully presented bells and whistles?

>The people who want them.

THAT is the central problem.  In the mass market case, no small number
of people can afford what they want if they're the only ones paying
for the development.  Read what you quoted:

>>Forming a large enough consortium to distribute the price/programming
>>effort is also not a practical proposition, since I'd expect at least
>>5000 people would be needed to get the price sufficiently low.

>I could find a number of programmers who could produce an X11 based
>WYSIWYG word processor in under a year for that sort of pay ($50,000).
>I could find a number of graduate students who could produce the same
>thing for about $10,000.  I might even be able to find a few undergrad
>students who could produce the same thing for about $2,000.

Don't count on it.  "How hard could it be to write a word processor?
I'll bet they could do it as a [team programming class] project!"  The
friend who said that (a very intelligent and creative person) spent
something like 9 man-years failing to finish his word processor.

You might get a minimal version built, this is true.  For-profit
market pressures force you to do something that is better than what is
already out there.  Start free and small, then spread and grow, may be
a worthwhile strategy.

>Once the FSF has a complete operating system to do development on
>(rather than trying to help support many different flavors) we may
>start to see all the nice "mass market specialized software" but until
>that time, FSF is doing what they set out to do and trying to say that
>this is an "inability" to produce packages is completely misleading.

Note the implicit assumption that the FSF's operating system will be
appropriate for running "mass market specialized software"
applications.  It will probably be very interesting to hackers and
programmers, because they'll build want they want.  Will it run on a 4
megabyte home computer with a '386 or 68020 in it?  Will it
efficiently implement a graphical user interface? (hint: if it is
running X11R4, the answer is "no".) In theory, yes, the FSF can do
whatever they want; in practice, they are unable, because hackers will
write what hackers want to use and play with, and their personal
desires (because of their experience and their prejudices, e.g.,
"there's nothing special about a Mac's user interface") are not the
same as those of the mass market.

And (once again) I don't "want" this software, so there is no reason
for me to work on it.  My time is much better spent earning money and
buying the software.  In the time it takes me to write/improve a piece
of Free software, I could earn enough money to buy the non-Free
version several times over.  Other people think like I do (especially
those who don't know how to program at all), and that's why I think
there won't be much Free mass market specialized software.

David Chase
Sun