embedded devices and open source

[Jon 'maddog' Hall]

I have been half-watching this thread because I perceive a tremendous
time-sink of my time, but I will add this one observation to it:

All of you are correct in one way or another.  Bear with me while I
explain this, and I have a feeling this thread will end quickly....

The end user typically does not know or care if the functionality they
have comes from hardware, firmware or software.  They buy functionality.

In the old days a modem (for instance) might be all hardware.  Then
people found out that you could simulate the functions of a lot of
hardware with software, put it in a ROM and call it "firmware".  A
little bit later (as the price of processors dropped) this firmware
could be something supplied in a driver, and the functionality could be
improved dramatically at relatively little variable cost on the unit.

But to the customer, for the most part, they bought the functionality of
the product.  And that is what makes them buy THAT product over the
product of THE COMPETITION.

Of course if you have put a lot of effort into your own product you do
not want your competition to simply take it and put it into theirs.  You
protect it any way you can.  Patents, obscurity, trade secrets, etc.
And you also may buy technology from people who also expect you to
protect THEIR intellectual property.

All of this has been covered before.  Now comes the new stuff.....

Everyone knows that the competition is sophisticated and can probably
reverse engineer everything you have done.  I have been in meetings
where everyone acknowledges that.  But if you have taken "reasonable
measures" to protect your (and others) IP, then that is fine with:

	o your engineers (well, some of them)
	o your technology partners
	o your boss
	o your stockholders
	o the courts

Of course it is all an illusion, but when you get to the last bullet, it
may mean a lot, and it keeps the other bullets happy.  If, on the other
hand, you expose your IP to everyone through FOSS, it makes all of these
other bullets unhappy.  You first have to convince these other bullets
that FOSS will make them more money, get more fame, etc.  Sometimes that
is hard to do.  Particularly when the customers that you are trying to
woo with this new philosophy are 1-2% of your current marketplace.

To illustrate this, we had an engineer who worked on Digital's dumb
frame buffer X-server.  A "dumb frame buffer" X-server is one where
there is virtually no hardware acceleration, it is all done in software.
Great engineer, really bright, and Digital's was the fastest dumb frame
buffer on the market.  He refused to allow Digital to ship the code for
our X-server.

Over the years the other frame buffer X-servers almost caught up, and
the hardware accelerators were becoming so cheap that few people cared
about the dumb frame buffer X-servers any more other than the really low
end.  But because we would not release the sources for our X-server,
various of these customers would not buy from us.

So I had to fly out to California, sit down with the engineer and walk
through his code.  Each step of the way we would come to some "crucial
feature" that our competition could copy if they had our source code.
I would ask the same two questions:

	o does any of our competition have this algorithm?
	o do they release their source code?

The answer in 59 of the 60 points was "Yes" and "Yes".

The last point was an algorithm that was used in dithering, but it was
only used in a very, very special case, and gained the X-server a 2%
speedup when it was used....so I asked the final question:

"Is it worth us losing millions of dollars of sales because of that
feature not being released as source code?"

At that point the engineer saw the point, and gave in.  Any of our
competition could have implemented our "trade secrets" if they wanted
to, since all of the other X-servers out there had implemented what
our engineer had.

But it took me, a fairly technical person, to sit down with the engineer
and go over this.  It was a political gamble for me, and the payoff (for
me) was not that great.

So the summary of this letter is:

        o In most cases the obscuration is not a real deterrent in
        reverse engineering, but that does not matter to a court of law,
        management and stock holders.  It is an illusion that has been
        accepted as reality.
        
        o Unless you can show a significant increase in revenue *against
        your competition* by going FOSS, you probably will not win the
        battle
        
        o this is not an issue of pure logic, but of law and emotions.
        The purely logical person is at a disadvantage in this argument.
        
        o It comes down to the bottom line that if all customers refused
        to buy products that were not FOSS, everything would be FOSS
        tomorrow.
        Well, the day after tomorrow, as it would take a few days to
        undo all of the "we will not show you ours if you do not show us
        yours"
        licenses.  But since most customers don't care, closed source
        goes on.  That is the bad news.  The good news is that more and
        more customers do care.
        
Warmest regards,

maddog