Bioperl: UCSF Discussion, Public V. Private Genome Efforts

[Andrew Dalke]

This is reference to BobbyO <BobbyO@cgl.ucsf.EDU>'s message to
the bioperl mailing list.

The discussion of how to deal with software patents is an ongoing
issue in the free software world (see http://lpf.ai.mit.edu/ and
http://www.freepatents.org/).  There's much less agreement on how to
use software license restrictions against non-software patents.

Your proposal, while noble in intent, has several difficulties when
applied to some realistic scenarios.  I have covered some of them
below.

You said:
> A license attached to these tools which restricted their use in
> producing patentable results, would bring much force to the
> academics' requests at the information-exchange bargaining table.

It would be very hard to define that restriction since there may be
many causal relationships whose effect are hard to define.  If I use
software with this restriction could others in my company be prevented
from submitting a patent.  If I start a service which answers peoples'
genomic questions, must I ask them if the results will be used to make
a patent?  If I publish my results using said software, are others
prevented from using those results for a patent?

The only way around it is to produce software so inticing that others
will use it even with that restriction.  (Granted, for those not
interested in getting patents, this is not a problem.)  You'll also
have to worry about the many developers who disagree with your
position and have no problem writing code used by others to make
patents.  I, for one, do not mind if my code is used by others to
get a patent.

> Legal tools which protect fair use and open distribution of software
> have been pioneered for some time, and provide a ready model for
> developing a license for publicly funded bioinformatics
> software. For example, the widely-known GNU copyleft is a document
> which need merely be attached to the sourcecode of any program to
> insure that software's free & open-code distribution forever.

  Unlike your proposal, the GPL does not restrict what you do with the
output of (most) GPLed software.  You can use the GNU C compiler to
generate completely proprietary software.  You can use the GNU
calculator (bc) to do the math for a patent application.  Why should
bioinformatics software be any different?  Or is my analogue incorrect?

  I'll also point out that every other license is incompatible with
the GPL, so your proposed "OASUL" could not be used with GPL code.
(That's why I prefer LGPL or other licenses over GPL.)

  Another nit, the GPL only applies if you distribute the software.
It says that if the software is distributed in source or binary form,
it must be distributed under the GPL.  It does not ensure the software
remains available.  If I provide a service which someone else uses,
the GPL does not ensure that the source be made available.  Otherwise
every time you use, say, a GPL'ed web server you would have to have
access to the source code.

  (Digressing for a moment, the GPL is out of date in the sense that
it doesn't deal well with mechanisms like shared libraries, perl
modules or distributed CORBA services.)

> 1) Develop a OASUL license-document which guarantees academics'
> rights on commercial use of their software. This license must be
> publicly and widely available, legally sound, and activated simply
> by attaching the license to the software before distribuiting it.

  Academics don't already have "rights on commercial use of their
software."  In many cases they have to work through their university
for that, and the university takes a cut.  Software I wrote at UIUC
said "Copyright The Board of Trustees of the University of Illinois"
not me or the research group.  Commercialization of the software would
require working with the university, and it's been a hard struggle
just to keep the distribution available with source and gratis.


> 2) Request a mandate from one or more major funding sources that the
> OASUL license be attached to sofware developed with public funds.

  Oooo, I would love to see the politicing for that to occur.  If the
funding sources are anything like the ones I know, they would love for
the research group to make some money from the software from
commercialization.  For one thing, the current Congressional mood is
to fund ideas which have relatively short term relevancy.  The NSF
likes it when one of the projects they've funded goes commercial
because then they can better justify their work.  Secondly, if the
research work does because, in essense, self-funded, then in addition
tl calling it a success, the funding agency can start funding others.

  The viewpoint of the US government has been that software developed
by the government is, for the most part, in the public domain.  That's
why you can do pretty much whatever you want with BLAST (one of the
informatics software you mentioned as being supported by public
funds).  If a funding agency like the NIH were to mandate the
restrictions you proposed, NCBI could no longer take NIH money to work
on BLAST because they are legally restricted from doing so.

  I don't see that happening.  So there will have to be different
forms of licenses depending on who gets the grant and their legal
status.


> 3) Allow optional appendices to the OASUL which, on attachment to
> the software below the primary OASUL document, would remit certain
> rights for each software tool to be remanded by proxy to academic
> negotiating groups such as the NIH, International Human Genome
> Project Board of directors, or relevant University System housing
> the software developers' laboratories.

  Universities don't like having existing rights removed.  Few do, but
universities have lawyers and time.  This would be hard to do.

  Consider also the group I worked in, which received research funds
from the university, NIH, NSF and a private fund.  Who gets the
rights?

  At a company I worked we wanted to use software which was funded by
two sources; both of which had commercial redistribution rights.
After about 4 months work trying to get permission to use the software
(we would even pay for it) we gave up.

  So, I think your proposal would end up making things much more
complicated and frustrating than they should be.


> 4) A version of the license which applies to bioinformatics
> algorithms, in addition to software, which are coverable by US &/
> international intellectual property law.

  Why should it bioinformatics specific?  (For that matter, what's a
definition of "bioinformatics" which distinguishes it from chemical
informatics, molecular modeling and computational chemistry?  :)

  A better argument is that new fields often go through a period with
a very fast growth in knowledge.  The premature introduction of
patents into this environment can stymie the growth of knowledge,
which is contrary to the Constitution justification for having
patents.  (Yes, this is a US-centric viewpoint.)  And bioinformatics
seems to be in this phase.

> 5) Crafting of these licenses such that they obligately guarantee
> free distribution & use of software for any non-profit use.

  I constantly try to point out that the profit/non-profit distinction
is not appropriate.  There are non-profit groups which hold and
restrict use of held patents (a lot of universities).  There are
for-profit companies which do not (eg, AT&T's patent on the unix
"sticky bit").  The major, popular "open source" licenses (GPL, LGPL,
NPL, BSD, etc.) do not make a distinction.

  Also, I'll also ask you be more specific in your use of the word
"free."  There are two definitions, "no cost" and "no restrictions."
The GNU distinction is "free beer" vs. "free speech" and only requires
the latter.  It is hard for me to justify that the software must be
gratis.

  You also need to address the belief many companies have that that
need to protect themselves by having a patent portfolio which they
can cross-license to other companies in exchange for not being sued.
(The lpf.ai.mit.edu page describes the reasons for and against this
better than I can.  It also proposes a way for "free" patents to
become more popular at http://lpf.ai.mit.edu/Patents/mutual-def.html)

> As a student, my voice & position are below the notice of leaders
> who could apply these ideas for the benefit of our community. If
> yours is not, and you still agree with this proposition now that it
> is more carefully stated, then I hope this letter is of use.

  As you can tell, I disagree with your ideas.  However, I can be
proven wrong.  There's an expression attributed to Linus Torvalds of
Linux fame: "He who writes the code chooses the license."  The easiest
way to prove me wrong is write a useful program with your proposed
license and see what happens.  You don't have to be a "leader" to do
that.

						Andrew Dalke
						dalke@bioreason.com
=========== Bioperl Project Mailing List Message Footer =======
Project URL: http://bio.perl.org/
For info about how to (un)subscribe, where messages are archived, etc:
http://www.techfak.uni-bielefeld.de/bcd/Perl/Bio/vsns-bcd-perl.html
====================================================================