Subject: Re: engineering counts
From: "Forrest J. Cavalier III" <>
Date: Mon, 22 Oct 2001 09:35:19 -0400 (EDT)

Stephen J. Turnbull <> wrote:

> "Nearly all of the existing open source software is worth some
> fraction of the face value to a CMM 3+ organization."  That value is
> by definition equal to max{0, (cost of doing it from scratch) - (cost
> of inspecting and upgrading to internal standards)}.

You forgot a term.  

The value should be:
  max{0,(cost of doing it from scratch) - 
 (cost of inspecting and upgrading to internal standards)} - (cost of
 identifying, obtaining, and rejecting packages to find those worthy
 of inspection)

Open source software can actually be worse than useless in some 

How long have we heard and said "Software reuse is coming!  Software 
reuse is coming!"  The problem is always in the two negative terms.  

For CMM 4 and 5, I think the cost of meeting internal standards is
almost always the majority of the cost of doing it from scratch.
(For CMM 3, that is still somewhat true.)  Further, CMM 5 processes
are so rare and unique that I think no CMM 5 is going to want to
risk acquisition of software from any other process, even
if it were CMM 5.  The unknowns are too great.

When you have a small positive term inside the max{}, the negative
term outside bites hard.  Even though the cost of that last term can
be reduced and amortized over a large number of users (that is one
goal of the project) it still is never 0.

Fortunately for open source software, the CMM 1 and 2 projects (most
of the development in the world) have a lot to gain from all
those other CMM 2 or lower open source projects.  

But that CMM 1 and 2 isn't the market Tom is trying to feel out.