Subject: Re: Cygnus
From: DJ Delorie <>
Date: Tue, 22 Jun 1999 16:07:39 -0400

> Something seems to have been lost in the translation here. If GPL
> software finds buyers at a purchase price of $20M, the buyers have a
> $20M incentive to redistribute widely within their markets.

Not quite.  The incentive to redistribute is that $20M *minus* the
inherent value of that software to the buyer, *minus* the loss of
value due to redistribution.  "Value" means more than price.

* Game console company X pays for custom optimizations to gcc, so that
  their games are faster than their competitors.  Redistributing that
  version reduces its value to X, so they probably won't redistribute.
  When that competitive advantage is no more, the value in
  redistributing is higher - perhaps they're open sourcing their dev
  kits for obsolete platforms - so they redistribute then.

* Accounting company Y pays $5M for software that will save them $10M
  in operating costs.  There is no longer a need to recoup costs by
  redistribution, and such a distribution would *incur* costs due to
  network usage, support calls, etc.

* Software Package Z comes with support, but a redistributed copy
  wouldn't.  There is no incentive to buy such a copy as the
  unsupported version is available for free via ftp, so redistribution
  to recoup costs isn't practical.

* I myself paid $650 for Adobe Photoshop, and I have gotten that much
  use out of it.  Even though I usually use the Gimp now, I feel no
  desire to sell my copy of PS to recoup costs.

> To put it another way, how is the open source market to grow if
> big=bad and small=good? Not all users of GPL software will be
> programmers hawking their wares in the bazaar.

Like I've said before, there's more to value than the code itself.
FSB's need to identify other types of value, else they'll never sell
more than one copy of anything.  Cygnus, for example, charges a high
price for software that's available *on time*.  Last time I checked,
something like 95% of all our contracts were delivered within four
*hours* of the deadline - and that in itself is worth a *lot* of money
for chip companies, where a week delay can mean the difference between
success and failure for their chips.