Re: Market pressures for more reliable software

["Marin David Condic" <marin.condic.auntie.spam@pacemicro.com>]

"Ted Dennison" <dennison@telepath.com> wrote in message
news:jK3W6.8329$pb1.316095@www.newsranger.com...
> In article <9g8lrk$37c$1@nh.pace.co.uk>, Marin David Condic says...
> When I capitalize "Free", its to indicate that I'm talking about FSF-style
"Free
> Software".
>
Yeah, but even under FSF, I can *sell* you the program - I just have to give
you the source as well. That was my point.


> Nearly all software is already developed as custom software under a system
where
> the deveopers are paid for the development. That includes *every* project
I've
> ever worked on in my career. Even for most of the "licensed" commercial
software
> I deal with, the "support" is the most expensive part. So using Free
Software
> models wouldn't even noticably impact the vast majority of the software
> developed.
>
Custom software isn't where I'm worried about the quality. You get what you
pay for there. If you want something that runs forever and never crashes,
you pay for the quality steps involved. Where I think there is an issue is
with shrink-wrap products that you just either have to buy or ignore and
accept whatever it is you get.


> You bring up games. To answer your question directly, yes people *do* pay
for
> "support" for games. 900-number "hint lines" for adventure games are not
unheard
> of. Also, nearly every game company has some kind of customer support (the
money
> to pay for for that obviously comes from game sales, so you can think of
it as
> sort of a "support tax").
>
I was simply trying to illustrate that there are a lot of off-the-shelf
products purchased that you install, use and never go back to the company
for anything. Not even upgrades. Maybe there is some percentage of the
population that will want some kind of support, but probably most of the
people who have "Space Invaders" or "Pong" aren't paying for that. How does
the company make money if they don't charge for the software? If things like
that need to have higher reliability, there has to be some kind of market
pressure to do so.


> >Now there might be something to the notion of software developed under
the
> >Ada Developer's Cooperative License (or similar) in that the guys who
build
> >the code only get paid if the code gets used and sold. The only way for
that
>
> That might work out OK in a world of mass-produced shrink-wrapped
software. But
> remember paragraph 1: the vast majority of software *developed* does not
fit
> that mold. In the world in which I work, I can't use any nifty routine
that's
> going to add to the system's costs, because *I* don't get to make those
kinds of
> decisions. The folks that do are my managager's manager, who's already
obsessing
> about overruns, and the paper-pushers off in accounting, who require all
sorts
> of paperwork and market research and competitive sourcing and several
weeks to
> make any kind of decision (You should see us trying to buy a PC video card
> before it goes obsolete. Its sad, really). Unless we are talking some
pretty
> significant functionality, its far easier for me to just rewrite it all
myself
> (and what a shameful waste of human effort that is!).
>
Been there. Done that. I understand. I know it is hard to market
components - especially with restrictive licenses - especially to companies
building a custom product with limited sales potential. It typically has to
be some sort of major subsystem with well defined capabilities (databases?
graphics packages?) before anybody wants to include it in their product. I
don't know what the answer is. Perhaps there needs to be a variety of tiers
to the payment scheme that makes it *easy* for people to get into it, but if
there are any profits made from it, they give *something* to the developers
who contributed code to the project.


> So that license might get some lucky package shipped in lots of software
copies.
> But for the vast majority of *developers*, its just going to render the
package
> useless.
>
Even for the high-volume stuff, it gets problematic. Think about it. If I
sell "Ada$oft Works" to the general public with a worldwide distribution of
tens (hundreds?) of millions of copies, the *cost* is no longer in the
engineering - its in the stamping out of disks & manuals. I'm going to give
some geek in a garage 10% of each sale for having contributed 10% of the
code? Guess again! I'll *eat" the whole engineering cost because it falls
below the noise level. OTOH, if I have a product being custom built for some
govet. contract that maybe gets handed over to the govt for one lump sum,
suddenly getting me 10% of the way to completion in one transaction looks a
little more attractive. Especially if it is reliable, proven software.

There are still problems with the pricing structure that I think would need
to be worked out - just because I bill my customer $1m for 10,000 SLOCs and
you gave me 1,000 SLOCs doesn't mean I can still make a profit after giving
you $100,000. And then throw in all the complexities of accounting (ask
movie stars who sign up for a percentage of the profits how the studios can
make sure a picture *never* amkes a profit!). Its not an easy question.

MDC
--
Marin David Condic
Senior Software Engineer
Pace Micro Technology Americas    www.pacemicro.com
Enabling the digital revolution
e-Mail:    marin.condic@pacemicro.com
Web:      http://www.mcondic.com/