Subject: back to topic (was Re: a stocks and dividends question)
From: Tom Lord <lord@emf.net>
Date: Thu, 9 Jan 2003 17:11:46 -0800 (PST)



No corporate taxes -- very interesting idea.  We need Greenspan on the
list to tell us what the models say about that.

But back to topic:

I asked for refutation or confirmation of my amateur belief that stock
valuations have to be rooted in dividends and got (paraphrasing)
"Yeah, that's economics 101".

So, the FSB reason why this came up was that I was thinking (as usual)
about how to fund practical free software R&D.   In particular, I was
thinking about the proprietary model we say in the 80s/90s (the
VC-funded start-up) and contrasting it with what I think makes sense
(R&D is a service;  vendors should be customers for that service).

Here's some of my fiscal reasoning that free software R&D companies
make sense, bringing an off-list thread back on-list:



           "Moreover, R&D in this form doesn't have to cover a VCs
            ROI -- it _only_ needs to cover the sum of the burn rates
            -- so it's a lower cost approach to R&D for that reason
            to."

   Isn't this a non-sequitor?  Every customer of this service
   has their own opportunity cost of capital just like a VC.
   And you'll still stuck at selling a product (attention and
   risk-taking) to somebody, then opening the fruits of that
   effort to everyone -- which may be a win (pie is bigger, etc),
   but it's still a harder sell.


Suppose we have N free software practical R&D projects.  Some will
succeed;  others fail.  Assume for the sake of the example that when
one succeeds, it will generate revenue almost exclusively for some of
the big vendors (HP/C IBM, Red Hat, Sun, ...).

The average operating cost for each project for one time unit is K.
So, the total cost of the projects is N*K.

The average cost of filtering proposals and choosing one project from
among them is J, totalling N*J.

So the unavoidable cost of the research is N * (J + K).

There are two options: 

a) following the silicon valley model of the late 80s...mid 90s, we
   could assign that unavoidable cost to VCs.  They have the job of
   picking the projects, they pay the operating expenses.  They
   want a return on their investment, R * N * (J + K) where R > 1,
   otherwise they don't make any money.

   Therefore, the total cost to the big vendors for these projects
   is R * N * (J + K), traditionally paid as licensing fees for the
   products of the successful projects.   Note that the factor R
   is competing against interest rates, municiple bonds, and other
   investments available to VCs, so it's going to be fairly high.
   [This is why I asked about stock valuation -- since VCs usually
   get their return by selling stock, not collecting dividends.]


b) following a model better suited for free software, the big vendors
   can do the filtering themselves, and regard "start up development"
   or "practical R&D" as a service in and of itself.   Rather than 
   letting VCs pay the operating expenses, the vendors can pay them
   directly.


So for the same end result, (b) is cheaper by (R - 1) * N * (J + K)
dollars.  (Kind of "cutting out the middle man".)

In fact, (b) is probably cheaper than that, too -- since the code is
all free and there's no reason not to share news of progress and the
results of "failed" projects,  projects can fail a little sooner, labs
can help each other out, code can be reused more ....  In other words 
the factor `K' in the above is smaller under a free software scenario
(b) than under a proprietary software scenario (a).  (Just the usual
"network effects".)

The opportunity cost to the vendors of ponying up the N*(J+K) early
is a wash.  If they aren't spending that amount forward on new free
software R&D, they're spending more than that amount backwards on
licenses to pay for completed proprietary R&D.   In other words, over
time, they don't have that money to spend on anything other than R&D
either way.

-t


[postscript]

There's a sometimes repeated myth that free software R&D happens "for
free", because universities do it, and because hobbiests do it.

The analysis suggests that people who are doing R&D "for free" are
themselves incuring the expense (J + K).

Given an ethically palatable situation in which they could do the same
work, but get a return S * (J + K), ideally S > 1 but even S > 0
provides incentive.

That suggests that academic-and-hobbiest-friendly proprietary licenses
will, by the "rational economic actor" hypothesis, soon kill nearly
all voluntary free software R&D.   The University of California at
Berkeley, for example, has been slowly moving in this direction.

Sure, people like me might try to do R&D anyway -- but, hey, it was
just 3 days ago that I got my "3 Day Notice to Quit" for a late rent
payment, and only by the grace of a long-term loan from an arch user
that I still have a roof.   Clearly I'm an "economically irrational
actor" -- but surely that's a socially constructed situation that 
some of the people on this list can help to fix.

And what about the ethics?  Corporations are not people.  They have
no natural rights.  It is at least arguable that they do not deserve
"software freedoms".