Subject: Re: Drafting employment contracts for free-software authors
From: DV Henkel-Wallace <gumby@henkel-wallace.org>
Date: Tue, 22 Feb 2005 08:48:04 -0800

On 21 Feb 2005, at 23:22, Kragen Sitaker wrote:

> I work for a non-profit; as part of our current reorganization, we
> want to hire some free-software developers.  In the past, we've found
> that our "intellectual property and confidential information
> agreement" has caused some concern for free-software developers we've
> tried to hire, since it provides for trade-secret protection for
> things the employee develops as part of their employment, in addition
> to the usual copyright-assignment stuff.

At Cygnus we always had people sign a standard agreement.  This was 
important for several reasons.  One was that we _had_ to be able to 
assign work people did on our nickel to the FSF; because I had written 
a blanket assignment to the FSF this interlocked properly.  Second, we 
had people's confidential info (kernel internals, unreleased machines, 
etc) from customers around and that had to protect it.  Third, we made 
patches and such for people that could temporarily not be 
redistributable, and so didn't want people redistributable[*].  And 
finally later in the company's life we toyed with some variants of 
split licensing with stuff we had written; we couldn't have done that 
had we tied our hands earlier.

Although some people objected, all signed.  After all the three 
founders were all quite committed to the GPL (I still think it's the 
right contract in all but a _very very_ few cases, despite having 
written the first Library License) and in fact the first of our own 
packages we released (bfd) was GPLed.  And if you were working on a GPL 
package (gdb, gcc, etc) well by definition you work was going to be 
free so what worry could you have?  The contract properly supported 
that.

And of course work you did on your own time with your own equipment 
unrelated to the company was yours anyway (calif law guarantees this) 
so there was no conflict there.  In fact we let people work on their 
own GPL projects at work as well but I think we insisted they get a 
sign-off in writing (which was _us_ insisting that it be GPLed).

As a non-profit you may have slightly different constraints, some of 
which are self-inflicted (you guys use other agreements than the GPL) 
but probably your reasoning will turn out to be the same.

HTH,
d

[*] This was a funny situation.  If you do work for hire for someone 
(e.g. custom development) than you are a contractor, and the work 
belongs to them.  Once they give it to a third party of course then you 
can too...but you don't have the right to do that unilaterally.

As a practical matter this generally only mattered to 1> people who 
were developing a new chip architecture or version and 2> short-sighted 
lawyers.  So we supported it in several ways:
  1 - Our sales pitch was built around the benefits of it being free 
software; by the time you got around to the contract stage the customer 
was into it, or else had already thrown us out.
  2 - The contract specified that the work would be confidential until 
either 1> they gave it to _any_ third party or 2> a certain time had 
passed (two years?)  In the early days we didn't think of clause 2 and 
were slightly burned by a project that was cancelled.  Only slightly 
since there was noone who could have used those changes anyway.
  3 - We insisted that non-customer-specific changes were assignable 
immediately.  Since this fed into 1> it was generally no big deal.

Technically 2 wasn't needed (the GPL says it anyway) but by putting it 
in the contract we would avoid a dispute with the company's lawyers -- 
but that never happened; once the agreement was signed the lawyers went 
on to the next thing and we worked with our customer.