Subject: Re: My customers
From: Mike Stump <>
Date: Thu, 27 Oct 1994 16:50:36 -0700

In article <> you write:
>I.e., a manufacturer will pay you a one-time fee to develop a packet
>driver for each architecturally new board?

Yes, that has been Cygnus's experience to some degree also.  I am not
exactly sure how much I can say, so I'll wave my hands a little and
just say, Sun helped us do a SPARC V9 port of gcc, and we have seen
plenty of money from other people that do chips as well.

>If your software is portable, well structured, and well documented,
>why is it to their advantage to do that rather than do it themselves
>starting from your existing free code?

Some try.  I mean, look at Intel, and what they have done over the
years with gcc.  Net result, a waste of money, INHO.  They would have
been better off, doing it the right way, and giving us the money they
invested instead.

We position ourselves to be more efficient at it then they sould be.
And I think we are.  This is our value to them.

In the case of packet drives, the company that builds a chip, won't
necessarily understand that it is beneficial to make the driver really
small.  This could be one advantage.  Also, Russell has tons of real
experience, and this should buy him more efficiency.  Also, Russell's
quality is probably better than what the company could manage inhouse.

>How can you price your services so that this type of business (which
>I would think would be very intermittent and low-volume)

About the intermittent and low-volume part.  Let's just say that
Cygnus has seen the low volume and intermittent jobs, and they take
quite a bit in resources, take a lot of time (measures in years), and
we get more than $100 for a port.

>contracting it out to some third party who will work for less than
>you (in India, say, where salaries are something like 1/5 what they
>are here)?

Well, in our case, if someone wants to try and take business away from
us, by contracting out to India, more power to them!  Guess what we
do?  We turn around and offer a new product to people next quarter
with all the sweet and blood that they poured into the product for us,
and we didn't have to spend a dime.  This doesn't hurt us in our
market place, as we are no where near saturation in the market.  This
is a free gift to us, and we thank them for it.  It helps us compete
against all the non-free software out there.

For example, a company sells a free database package for $100, and
some major corporation wants to add graphics to it, and doesn't mind
ponying up the cash, but they give it to India.  The next release of
the software package from the company will include graphics, and might
sell for $110 now..  Or more people might like it now that is has
graphics, and the company can drop the price to $75, and still double
their profits.

But your right, at some point we have to see cash.  If we never see
cash, or not enough of it, we are doing something wrong, and we go out
of business.  Look at IBM, and DEC.

Let me return now to experience.  If you examine the ChangeLogs for
gdb, g++, gas, binutils, or bfd, you will see that usually the work
done on them, is done by Cygnus.  So, right now, one doesn't have to
worry about outsiders taking food from ones mouth.

In some respect, I wish the future would hurry up and arrive, and
people try and compete against us with our own product.  When that
happens, the games changes slightly.  Will we see a decrease in
revenue?  No.  Wait you say, that can't be.  If someone is out there,
taking our market from us, something has to give.  What gives, I
think, is that other compiler vendors will be less able to compete
against us, now that our costs have dropped in half!  We don't have to
pay to advertise, develop, test the other version of our product, and
since we both benefit, we'll work together with our new found
competition to fight others in the marketplace.  We only stop helping
each other directly, when we're the two last players in the market!
By that time, free software has won in that market, and we can go on
maintenance, or shift gears and conquer some other, and more
profitable market.

With free software, the producer is guaranteed to lose in the end.
This is true.  But as long as the company rides the market for all the
money it is worth initially, and then switches to the next most
profitable market, they can always make money (I hope).