Subject: RE: "Biological Open Source"
From: "Lawrence Rosen" <lrosen@rosenlaw.com>
Date: Wed, 15 Nov 2006 21:06:16 -0800
Wed, 15 Nov 2006 21:06:16 -0800
Hi Janet,

 

It is nice hearing from you again, and thanks for writing to license-discuss
with that interesting set of questions. Of course, we're free to ask
questions back.... :-)

 

For an example of the grant of patent rights necessary to practice industry
standards, see the Microsoft Open Specification Promise at
http://www.microsoft.com/interop/osp/default.mspx. There are similar
covenants by Sun and IBM and other companies. These are generally considered
compatible with all FOSS licenses.

 

For an example of the grant of all patent rights for any open source
software (plus experimentation, research and teaching) see the International
Characters Covenant Not To Assert at page 2 of
http://rosenlaw.com/IC-Business-Model.pdf. Other companies are considering
similar covenants and/or indemnification to protect their customers from
patent infringement claims, but none have yet been submitted for
consideration as compatible with the OSD. 

 

For an example of a patent grant in a free software license, see section 11
of Draft 2 of GPLv3 at www.gplv3.fsf.org <http://www.gplv3.fsf.org/> . For
examples of patent grants in open source licenses generally, see almost
every license written in the last ten years. My own personal favorite is the
one-sentence patent grant in section 2 of the Open Software License, OSL
3.0, at www.rosenlaw.com/OSL3.0.htm. 

 

I don't think there's much enthusiasm to come up with yet another way to say
these things about patents. 

 

Presumably software is involved in Biological Open Source or we wouldn't be
discussing this here at all. Are the covenants and licenses I highlighted
above adequate for any patented technology-presumably including biotech-that
can benefit from implementation in copyrightable software? What is different
or unique about Biological Open Source that suggests we need something new?

 

Best regards,

 

/Larry

 

Lawrence Rosen

Rosenlaw & Einschlag, a technology law firm (www.rosenlaw.com)

Stanford University, Lecturer in Law

3001 King Ranch Road, Ukiah, CA 95482

707-485-1242  *  fax: 707-485-1243

Author of "Open Source Licensing: Software Freedom and 

                Intellectual Property Law" (Prentice Hall 2004)

  _____  

From: Janet Hope [mailto:janet.hope@anu.edu.au] 
Sent: Wednesday, November 15, 2006 5:58 PM
To: license-discuss@opensource.org
Subject: Re: "Biological Open Source"

 

 

On 15/11/2006, at 11:39 PM, Michael Tiemann wrote:

 

Janet, I for one welcome your participation.  Moreover, while it is not

likely to lead to a specific license approval, it is very much on topic

for the expertise this list represents.

 

 

Thanks, Michael, for your positive response.  Having received some private
as well as public encouragement and no discouragement, I'll go ahead and put
my question to the list.  (It has been suggested that a new list could be
started for non-software open source licensing issues to cut down on
off-topic posts -- I think that would be very helpful.)

 

This post is long. Most of it falls under the headings "Background" and
"Preliminary licence analysis".  Readers who are interested, but not that
interested, are encouraged to skip those parts.  

 

THE LICENCES

 

The licensor, CAMBIA, has issued two "Biological Open Source" ("BiOS")
licences, one for Plant Enabling Technologies ("PET") and one for Genetic
Resources Indexing Technologies ("GRIT").  Each must be read in conjunction
with its own Technology Support Services Subscription Agreement ("PET TSSS"
and "GRIT TSSS").  All of these instruments are currently in version 1.3 and
are available at http://www.bios.net/daisy. 

 

Let me say up front that I am not affiliated with CAMBIA and have not been
involved in the process of drafting the BiOS licences.  As a matter of
courtesy to the drafters, I am simultaneously posting my comments to
CAMBIA's own discussion forum.  However, as Michael suggests, I suspect that
license-discuss is where much of the relevant expertise is to be found. 

 

THE QUESTION

 

The question is: To what extent is CAMBIA's BiOS licence scheme "open
source" in the software sense? 

 

In other words:  If the BiOS licences were software/copyright licences,
would you recommend them for OSI approval?  If not, why not?

 

SPECIFIC QUERIES

 

In particular (see the rest of this post for more detail on each of these
points):

 

(1) If a licence is primarily a patent licence and the patent grant itself
purports to be open source (not just compatible with an open source
copyright grant elsewhere in the licence), should field of use restrictions
be permitted? (OSD #6)

 

(2) What should be regarded as the equivalent of "source code" in a
biotechnology licence? Should it include materials etc in addition to that
which would be required for disclosure under patent law? If so, on what
terms should it be made available to users? Specifically, is it acceptable
to charge a fee that does more than recover the costs for
storage/retrieval/postage of samples etc (OSD #2)? Is it acceptable for the
amount of the fee to depend on the size and type of the user's organisation
(OSD #5) and the length of the user's "subscription" to ongoing "technology
support services"?

 

(3) What is the nearest patent equivalent of the open source copyright
licence grant to copy, modify, and distribute copies or derivative works
(OSD #1)?  Does it depend on the type of patent (eg process or product)?  Is
the patent grant in the BiOS licences effectively limited by the exclusion
of "Improvements" from the definition of "BiOS Licensed Products"?

 

(4) Is a right to sublicense essential in an open source licence? Is it
acceptable for this right to be limited or absent? If so, what other
conditions should be met to ensure that a licensor cannot impose non-open
source restrictions as a condition of issuing a new licence (eg OSD #7)? How
can these conditions be met in the case of a technology to which the
equivalent of a copyright notice cannot be so easily attached as it can to
software code?  

 

(5) Does the grantback in the BiOS licence go beyond a copyleft-style
provision in restricting users' freedom to fork the innovative process? Is
there any pressing legal or practical reason why the licensor in a
biotechnology/patent context could not simply require licensees to make
defined improvements available on the same terms as the original licence, as
in the software/copyright context?

 

BACKGROUND

 

CAMBIA (formerly an acronym for Center for Application of Molecular Biology
in International Agriculture) is a private, not-for-profit research
institute located in Canberra, Australia.  Founded in 1994 by its present
CEO Dr Richard Jefferson, CAMBIA is a small organisation that -- atypically
-- combines "wet lab" development of biotechnology research tools with
intellectual property informatics and policy development.  CAMBIA is
financed by grants from philanthropic organisations including the
Rockefeller Foundation, by national and international research funding
bodies, by official development assistance and by licence revenue from its
own patented technologies.(See http://www.cambia.org/daisy/cambia/589.html.)


 

In 2005, CAMBIA launched an initiative called Biological Innovation for Open
Society (BIOS: note the upper case "I").  The BIOS initiative consists of: 

(1) Patent Lens: searchable databases containing EPO, US and PCT patent
documents, together with ancillary IP-related information and tutorials. 

(2) BioForge (styled after Sourceforge.net): a portal for protocol-sharing,
comments on patents, and discussion tools in both public and secure
environments; intended to develop into a collaborative technology
development platform.  

(3) Biological Open Source (BiOS with a lower case "i") is the aspect of the
BIOS initiative that is "intended to extend the metaphor and concepts of
Open Source to biotechnology and other forms of innovation in biology". (See
CAMBIA BiOS License for Plant Enabling Technology Version 1.3, Recitals,
first paragraph.)

 

A description of the materials and methods that researchers can obtain from
CAMBIA, including an indication of which materials and methods are available
under which BiOS licence, is at
http://www.cambia.org/daisy/cambia/materials.html.  These technologies are,
of course, quite different from those licensed under OSI-certified software
licences.  Plant enabling technologies and genetic resource indexing
technologies consist not of software code (though software may be one
component), but of heterogeneous methods and materials, including living
biological materials.  Typically, such technologies are legally protected by
a mixture of intellectual and personal property rights, including patents.
By contrast, I understand that most open source software licences are
primarily copyright licences, though some also contain a limited patent
grant.

 

As I mentioned in my first post, the fact that the BiOS licences apply to
plant enabling technologies and genetic resources indexing technologies
means that they cannot be certified by the Open Source Initiative as being
"open source" licences because the terms of the official Open Source
Definition are not broad enough to accommodate the legal and technical
differences just described. In consequence, there is no certification signal
to help potential users determine whether or not signing up to a BiOS
licence will actually give them the nearest equivalent to the "software
freedom" guaranteed by the principles of open source software licensing.
Clearly, the task of translating open source licensing principles from one
technology setting to another is one that creates room for both legitimate
adaptations and distortions of the kind that would lead the OSI to refuse to
certify the relevant licence if the technology in question were, in fact,
software.  By my preliminary analysis, the BiOS licences contain a mixture
of the two.  However, I (and many others) would like to be able to
supplement that analysis with the expert views of people on this list.

 

Note that I am in no way impugning CAMBIA's good faith or suggesting that
the BiOS licences are "bad".  Certainly, CAMBIA has made a serious effort at
a very difficult task.  The licences may well be good licences, and good for
innovation.  However, the question here is whether it is reasonable to
describe them as "open source".

 

PRELIMINARY LICENCE ANALYSIS (WARNING: MAY CONTAIN ERRORS! TINLA!!)

 

For simplicity, I refer here only to the terms of the BiOS Plant Enabling
Technology (PET) agreement. The PET scheme closely resembles the GRIT
scheme. The main difference is that whereas the PET scheme contains a
limited sublicensing right, the GRIT scheme does not contain any right to
sublicense.  

 

1. Overall scheme of the licence 

 

The BiOS licensing scheme encapsulated in version 1.3 of the PET Licence and
corresponding TSSS Agreement contemplates an initial contribution of IP and
technology by CAMBIA that may act as a seed for ongoing technology
development. ("IP and Technology" is defined in Licence cl. 1.8.)  Thus, the
subject of the initial licence grant is a collection of CAMBIA's own
patents, materials and technology data.(See definitions of "CAMBIA Patents",
"CAMBIA Material" and "Technology Data" in Licence cll.1.9, 1.10 and 1.13
respectively.)  Downstream, this initial subject-matter may be supplemented
by "Improvements", "Improvement Patents" and "Improvement Materials"
(Licence cll 1.6, 1.7 and 1.11 respectively) developed by the licensee,
non-exclusively granted back to CAMBIA under Licence cl. 3 and then
redistributed by CAMBIA to all BiOS licensees as part of the grant of "IP
and Technology".

 

2. Parties

 

Although the licence is described on the BIOS forum as a "template", one
party (CAMBIA) is explicitly named throughout the licence text as the
licensor.  I understand that this is discouraged in the OSI approval process
because it contributes to licence proliferation; however, this is a
relatively minor point.  

 

The sense in which the BiOS licences are templates is that they contemplate
a number of licensees all receiving CAMBIA's permission to use the IP and
Technology on "substantially similar" terms. (See Licence Recitals,
paragraph 3. Why are the terms "substantially similar" and not identical?
Two reasons: (1) different licensees may agree to different numbered
versions of the licence (see Licence cl. 7.5 and TSSS Agreement cl. 7.6).
(2) The terms on which licensees can access any materials needed to practise
the licensed inventions vary according to the type and size of the licensee
organisation (see TSSS Agreement cl. 3 and Annex D).) 

 

3. Patent grant

 

The licence grant is contained in clause 2 of the PET Licence, the key part
of which reads: 

"2.1 CAMBIA hereby grants to BiOS LICENSEE under Licensed Patents in the
Field of Agriculture a worldwide, non-exclusive, royalty-free right and
license to make and use the IP & Technology for the purpose of developing,
making, using, and commercializing BiOS Licensed Products without obligation
to CAMBIA...."

 

The nature of the grant is crucial to the question whether the BiOS licences
can be justifiably described as "open source". Note that unlike open source
software licences, the BiOS licence is primarily a patent licence.  While
patent grants in open source software licences must be compatible with an
open source copyright grant, they are not the primary means by which these
licences seek to protect users' rights to access and use the technology.
The question here is not merely whether the BiOS patent licence grant is
\textit{compatible} with open source principles, but whether it is itself
"open source" (whatever that is taken to mean in the biology context).

 

3a. Grant is restricted to a particular field of use

 

The BiOS patent licence grant is limited to a particular field of use,
namely Agriculture. ("Agriculture" is defined quite broadly in Licence cl.
1.3.)  There is a question whether this conflicts with OSD # 6.

 

While a strict formal interpretation of open source principles would
presumably prohibit field of use restrictions in open source biotechnology
licences, the difficulty with such an interpretation is that in the patent
context, field of use restrictions, together with territorial restrictions,
may be a useful tool for making technology that would otherwise be bound up
in exclusive licences more readily available for public interest and broader
commercial use.  The licensing policy of Public Intellectual Property
Resource for Agriculture (PIPRA), a collective intellectual property
management regime for agricultural biotechnology recently established by a
group of land grant universities in the United States, is a case in point.
One of PIPRA's primary aims is to overcome the fragmentation of public
sector intellectual property ownership by identifying residual rights
retained by members who have assigned unnecessarily broad exclusive rights
in important technologies to major commercial firms. One of the architects
of the initiative explains the importance of field of use and territorial
restrictions in this context: "Best practice includes partitioning of
patents: if you go and license something like your agrobacterium technique,
license it just for cotton; or better, for cotton in the US; or even
better... define which varieties, or... constrain it to varieties owned by
the licensee company in the US.  The more you can constrain the space of the
technology grant, the more is still left over [that you may choose] to put
into the commons." (Greg Graff, personal communication.) Thus, the value of
such field of use restrictions in terms of achieving wider access to key
biotechnologies may outweigh the value of keeping to a strict analogy with
open source software licensing. 

 

What do people think about this? I understand that with respect to patented
code, the F/LOSS community objects to field of use restrictions in software
patent licences because such restrictions are perceived as capable of
insidiously undermining freedoms granted in relation to the same code under
an open source copyright licence.  However, the approach of open source
community leaders to this situation appears to have been to campaign for
maximum breadth of patent licences, while remaining willing to compromise
for the sake of workability.  Thus, many open source licences, including the
Apple, IBM and Mozilla licences, have field of use restrictions in their
patent grants. The W3C Patent Policy also represents a compromise.  How far
should this willingness to compromise go in a case where the patent grant is
the main or even the only grant in the licence? 

 

3b. Scope of grant

 

The grant is a "right and licence to make and use the IP and technology for
the purpose of developing, making, using and commercialising BiOS Licensed
Products".  Note that the grant does not explicitly include a right to sell
the IP and Technology itself.  Nor does it permit the licensee to make or
use the IP and technology for purposes other than developing, making, using
or commercialising BiOS Licensed Products.  "BiOS Licensed Product" is
defined in Licence cl.1.4.  That definition is discussed further below, but
significantly, it is restricted to assets that are substantially distinct in
some respect from the licensed IP and technology and are "intended for
commercialisation".  "Commercialisation" is not defined in the BiOS licence.
However, in light of recent US case law concerning patent infringement by
universities (eg Madey v Duke), it may be quite broad. 

 

The OS mantra is "anyone, anywhere, for ANY PURPOSE, must be free to copy,
modify AND DISTRIBUTE the software, for free or for a fee...".  At first
glance, the BiOS patent grant seems considerably narrower than this.
However, patent law grants a patent owner a different set of rights to those
of a copyright owner.  It is therefore not a straightforward question what
the equivalent of the broad open source copyright licence grant should be in
the patent context.  It is arguable that an open source copyright grant
covers all of the exclusive rights of the copyright owner that are relevant
to the licensed technology, and that an open source patent grant should do
the same.  

 

IAAL, but not a patent lawyer.  However, I understand that in Australia at
least, the precise form of a broad patent licence grant that would cover all
of the owner's rights-to-exclude would depend on the nature of the patent
claims.  Section 15 of the Australian Patents Act 1990 gives the patent
holder the right to exploit the invention. The definition of "exploit" in
Schedule 1 of the Act distinguishes between products and processes. For
products, it includes the right to make, hire, sell or otherwise dispose of
the product. For processes, the definition includes the right to make, hire,
sell or otherwise dispose of products resulting from the process. Without
looking at the claims in CAMBIA's patents, it appears that they are process
patents (ie methods of doing plant enabling technology and genetic resource
indexing technology). The licences grant the right to use the methods. Any
products are supplied separately under the support agreement. The licences
actually grant the right to sell products using the methods under licence.
Thus, the licence grant may be broad enough to satisfy open source
principles with respect to the initial IP and Technology.  

 

What do you all think about this?

 

3c. Limited right to sublicense

 

A final point to note regarding the BiOS licence grant is that it includes a
limited right to sublicense, as follows: 

"2.1.1 granting limited sublicenses to third parties... to conduct research
and/or development activities...for BiOS LICENSEE, provided that the product
and/or other results (including all intellectual property rights) resulting
from said limited sublicense are owned exclusively by BiOS LICENSEE, said
limited sublicenses to terminate when such activities cease or such
ownership terminates, and

"2.1.2 granting limited sublicenses to third parties... for the sole purpose
of commercializing BiOS Licensed Products that embody the IP & Technology or
are generated by use of the IP & Technology, said limited sublicenses to
terminate when said commercialization ceases.

"BiOS LICENSEE shall be responsible to ensure... that any Improvements
produced by sublicensees are considered to be Improvements hereunder....
BiOS LICENSEE shall provide a list of sublicensees to CAMBIA in writing at
least once a year....

"BiOS LICENSEE shall further be responsible to ensure that...third parties
to whom a sublicense has been granted are notified... that said third
parties have no right to sublicense absent the execution of a BiOS License
Agreement with CAMBIA, and that CAMBIA may be approached for a BiOS License
Agreement.

"Other than stipulated under this Article 2.1, no further right to
sublicense is granted to BiOS LICENSEE hereunder."

 

I am unclear as to whether a right to sublicense is essential for OSI
certification of new licences.  My understanding from Larry Rosen's book is
that the right to sublicense is considered desirable, but is not a feature
of all OSI-approved software licences.  I imagine that in the absence of a
sublicensing right, OSD #7 becomes very important because unless people are
sure they are entitled to a licence no matter what, the need to go back to
the initial licensor for permission to use the technology would not be a
mere formality and could become too restrictive.  However, it is not clear
exactly how OSD #7 would operate in relation to the exchange of biological
materials, methods etc, as distinct from code or other content to which a
copyright notice can be easily attached.

 

Can anyone clarify the OSI requirements re sublicensing?  (Sorry if this has
been dealt with in previous posts.)

 

4. Source code

 

The freedom to copy and modify open source software programs and to
distribute copies and modifications is only one important aspect of open
source licensing.  Another is the practical enablement of that freedom
through unrestricted access to software source code.  In the case of
patented biotechnologies, it may be argued that the equivalent of source
code is automatically publicly available because of the disclosure
requirement under patent law, and that an open source-style licence grant is
all that is needed to render a patented technology "open source".  On the
other hand, biotechnology licensors typically go beyond this level of
disclosure when attempting a genuine transfer of technology.  This practice
suggests that patent disclosure many not be enough to give full practical
effect to the legal freedom to use and improve a technology that is intended
to be conferred by an open source licence grant.  

 

In the BiOS PET licensing scheme, separate provision is in fact made for
access to materials needed to practise the licensed patents.  This access is
governed by the terms of the TSSS Agreement (see Licence cl. 4) and appears
to be conditional upon (1) a substantial annual payment for at least three
years, depending on the licensee's organisation type and number of employees
and (2) significant reporting requirements regarding new materials,
technology data and project ideas.  If access to materials needed to
practise the licensed payments is regarded as equivalent to access to source
code, these restrictions represent a significant departure from established
open source principles of non-discrimination among licensees, access to
source code for free or at the minimum charge required to recover the
provider's costs, and freedom from ongoing obligations to the licensor,
including obligations that take the form of reporting requirements.

 

5. Grantback

The next phase in the scheme established by the BiOS licences and TSSS
agreements is triggered when a licensee, in the course of exercising the
licence grant, makes or discovers something new.  

 

The BiOS PET licensing scheme distinguishes between "BiOS Licensed
Products", which may be developed, made, used and commercialised "without
obligation to CAMBIA", and "Improvements", which are subject to a grantback
obligation  (see below) that also covers Improvement Patents and Improvement
Materials (Licence cll. 1.7 and 1.11, respectively).  The definitions of
these different categories of potential follow-on innovations are
complicated and tinged with uncertainty.  However, the fundamental intention
appears to be to allow the licensee to appropriate any new asset that has
been generated through use of the IP and technology and is intended for
commercialisation.  Such an asset is a "BiOS Licensed Product" (Licence cl.
1.4), provided it does not fall within the definition of "Improvement"
(Licence cl. 1.6).   

 

The definition of "Improvement" is thus a critical element of the BiOS
scheme. This definition is potentially extremely broad.  Note in particular
that even something as minor as a repeatable observation can constitute an
Improvement, and that an Improvement need not be a Plant Enabling Technology
(itself very broadly defined in Licence cl. 1.12) because of the catch-all
"but for the terms of this License Agreement...".  

 

5a. Does the definition of "Improvement" limit the licence grant itself?

 

A preliminary question is whether the broad definition of "Improvement" in
Licence cl. 1.6 effectively limits the scope of the main licence grant in
Licence cl. 2.  As noted earlier, the purpose of the grant is limited to
"developing, making, using and commercialising BiOS Licensed Products".
"BiOS Licensed Products" is defined in cl. 1.4 to mean "any tangible or
intangible asset of BiOS LICENSEE (including without limitation any material
or method, but excluding Improvements)".  This suggests that anything that
falls within the definition of Improvement automatically falls outside the
definition of BiOS Licensed Product and therefore is not covered by the
grant even for purposes other than commercialisation. 

 

Similarly, any new development that is neither a BiOS Licensed Product nor
an Improvement does not appear to be covered by the terms of the licence.
In either case, further permissions may be required in order to avoid
infringing the licensor's proprietary rights.  CAMBIA appears to be free to
make such permissions conditional upon payment of royalties or other
conventional licensing terms -- or to refuse them altogether.

 

5b. Grantback: copyleft equivalent or licensing "club"?

 

>From an open source perspective, the breadth of the definition of
"Improvements" is not necessarily a problem, although it could be if it
extended the range of follow-on innovations that would be caught by the
grantback well beyond those that would be caught by a copyleft-style
provision relating to "derivative works".  As I understand it, the question
of what constitutes an appropriate incentive for licensees to innovate in a
copyleft-style licence is a separate issue from compatibility with open
source principles.  The broadest examples of copyleft "hooks" in open source
software licences are tailored to catch all derivative works that are
distributed outside the boundaries of the licensee's own organisation; other
open source licences permit varying degrees of freedom to operate with
derivative works. The BiOS grantback appears to fall somewhere in the middle
of this spectrum.  In a 2004 article on open source patent licensing, Sara
Boettiger and Dan Burk point out that copyright confers exclusive rights
only against unauthorised copying or other violations of the specifically
enumerated rights of the copyright owner arising out of contact with the
copyrighted work itself.  By contrast, independent creation is not a defense
to claims of patent infringement. The BiOS licence appears designed to
replicate the copyright situation by excluding from the definition of
Improvement improvements that are "developed without the use of the IP and
Technology".  Similarly, the BiOS arrangement appears intended to mirror
open source software licences by permitting in-house use of improvements
that are actively protected as trade secrets: these are not caught by the
BiOS grantback. (Note, however, that this escape clause may be somewhat
narrower than the positive requirement in open source software licensing for
a derivative work to be externally deployed before it triggers any copyleft
obligation: clearly, there is a difference between active dissemination of a
technology and merely failing to  adequately protect a trade secret.
Further, note that in-house use of an improvement under conditions of trade
secrecy may not be permitted at all under the BiOS scheme unless interpreted
as a form of "commercialisation": see Licence Grant in cl. 2.2 and
definition of BiOS Licensed Product in cl. 1.4.)

 

The point at which the BiOS licence provisions relating to follow-on
innovations appear to me to depart from open source principles is not in the
breadth of what may be caught by the grantback, but in the fact of the
grantback itself.  While the copyleft or "reciprocal" obligation to be found
in some open source licences has been characterised in academic literature
as a type of grantback, this does not appear to me to be an accurate
description.  A copyleft licence does not establish a tit-for-tat
relationship between the licensor and licensee.  Rather, a copyleft licensor
says to the licensee: "Do as you have been done by".  The copyleft
obligation is not to the licensor per se -- though he or she may be
responsible for enforcing it -- but to the whole community of potential
users of the follow-on innovation.  The licensor may or may not be a member
of this community, depending on the scope of his or her activities and
interests.  For this reason, I prefer to describe the copyleft obligation as
a "grantforward" (as in "passing it forward") rather than a "grantback".

 

The BiOS grantback, by contrast, appears to be just that: whether or not the
intention is noble, it still constitutes a type of privilege granted by the
licensee to the licensor in partial consideration of the licence grant.
Under cll. 3 and 4 of the Licence and relevant provisions of the TSSS
Agreement, a BiOS licensee is obliged to grant to CAMBIA a worldwide,
non-exclusive, royalty-free licence (with the right to sublicense to other
BiOS licensees) to Improvement Patents, Improvements, Technology Data, and
any  Improvement Material provided by the licensee to CAMBIA and necessary
to practise Improvements. (Definitions are in Licence cll. 1.7, 1.6, 1.13
and 1.11 respectively.)  Thus, the form of the BiOS licence differs from a
copyleft open source licence: instead of requiring licensees to license a
subset of follow-on innovations on the same terms as those of the original
licence, it builds the follow-on licence into the original licence and
places the initial IP owner, CAMBIA, at the centre of the network as a kind
of gatekeeper. While CAMBIA does have an obligation to make these follow-on
innovations available on the same terms as the initial licence, that
obligation extends only to existing BiOS licensees.  In essence, the
arrangement is a licensing "club".  Given the restrictions on sublicensing
by BiOS licensees (Licence cl. 2), new members can join this club only by
executing a new licence with CAMBIA; CAMBIA is under no obligation to grant
such a licence.

 

Club-type licensing arrangements are actually quite familiar in the
biotechnology context.  Depending on the circumstances, such an arrangement
may be good for innovation, but it seems to me that it is not open source.
Why?  Because an arrangement in which the initial licensor retains a central
position linked to his or her ownership of the seed IP and Technology is
inconsistent with the freedom to fork the innovative process.  

 

The freedom to create a 'code fork' is regarded by some as a defining
characteristic of open source -- recall Eric Raymond's open letter to Sun a
little while back. According to Steve Weber (author of The Success of Open
Source), under the terms of an open source licence, anyone who is
dissatisfied with the conduct of a project leader -- on technical,
administrative, political or even purely personal grounds -- is free to take
the collaborative effort in a new direction. Even though in practice,
forking is rare, the ever-present possibility makes project leaders
responsible to their co-developers and ensures that no individual or group
unduly dominates the process of technology development. At the same time, it
ensures that a promising technology need not be left on the shelf because of
waning interest or incapacity on the part of an initial innovator.  This is
a concern often expressed on this list in the form of: "What happens to the
licence if So-and-so, the licensor, goes bankrupt or disappears in ten
years' time?" 

 

Of course, all else being equal, it is natural for an initial innovator to
remain in charge of ongoing development and to act as a champion of the
technology. However, it seems key to the open source approach that the
initial innovator not use his or her ownership of the intellectual property
in the initial seed technology to retain control over its ongoing
development.  

 

It appears to me that this freedom to fork the innovative process with or
without the approval of the initial licensor is lacking in the BiOS scheme.
Why?

(a) The grantback incorporates stringent reporting requirements to CAMBIA
(Licence cll. 3.2 and 3.4) and obliges the licensee to provide CAMBIA with
any materials necessary to practise Improvements (Licence cl. 3.2). 

(b) Discussions of technology data between BiOS licensees must be channeled
through CAMBIA (TSSS Agreement, cl. 2.3).  

(c) CAMBIA retains the right to impose an obligation on licensees to include
CAMBIA's trademarks in downstream technologies (Licence cl. 7.9).  (This
provision closely resembles one that I thought was removed from an early
version of the open source BSD software licence because it imposed too much
of a restriction on the freedom of downstream developers -- is this
correct?) 

(d) Under Licence cl. 7.5 and a similar provision in the TSSS Agreement,
CAMBIA "may publish revised and/or new versions of the BiOS License for
Plant Enabling Technologies from time to time. Such new versions will be
similar in spirit to the present version, but may differ in detail to
address new or newly identified issues. Each version will be given a
distinguishing version number. ... Where there is... any controversy between
the parties respecting the interpretation or application of the terms of
this Agreement, the latest... version of the Agreement published on the BIOS
website shall be controlling." 

(e) Perhaps the most extreme instance of CAMBIA's apparent attempt to keep
its hand on the reins of future development is in the TSSS Agreement, cl. 6,
which reads in part:

"From time to time BiOS LICENSEE may have interest in exploring solutions to
certain problems in its field of business. In such case BiOS LICENSEE shall
[emphasis added] send a written project proposal to CAMBIA. To the extent
that such project proposal is consistent with the terms and intent of the
BIOS License, which consistency shall be determined by CAMBIA in its sole
discretion, CAMBIA shall use its best efforts to... bring the proposal to
the attention of other parties within the BIOS Initiative...".  

In other words, the licensee is obliged to send a written proposal to
CAMBIA, but CAMBIA is not obliged to do anything with it.  The effect
(whether or not intended) is that CAMBIA gets a private preview of any new
project.

 

Thank you for your input.

 

Janet

 

 

 

 

 

 

 



Hi Janet,

 

It is nice hearing from you again, and thanks for writing to license-discuss with that interesting set of questions. Of course, we're free to ask questions back.... :-)

 

For an example of the grant of patent rights necessary to practice industry standards, see the Microsoft Open Specification Promise at http://www.microsoft.com/interop/osp/default.mspx. There are similar covenants by Sun and IBM and other companies. These are generally considered compatible with all FOSS licenses.

 

For an example of the grant of all patent rights for any open source software (plus experimentation, research and teaching) see the International Characters Covenant Not To Assert at page 2 of http://rosenlaw.com/IC-Business-Model.pdf. Other companies are considering similar covenants and/or indemnification to protect their customers from patent infringement claims, but none have yet been submitted for consideration as compatible with the OSD.

 

For an example of a patent grant in a free software license, see section 11 of Draft 2 of GPLv3 at www.gplv3.fsf.org. For examples of patent grants in open source licenses generally, see almost every license written in the last ten years. My own personal favorite is the one-sentence patent grant in section 2 of the Open Software License, OSL 3.0, at www.rosenlaw.com/OSL3.0.htm.

 

I don't think there's much enthusiasm to come up with yet another way to say these things about patents.

 

Presumably software is involved in Biological Open Source or we wouldn't be discussing this here at all. Are the covenants and licenses I highlighted above adequate for any patented technology—presumably including biotech—that can benefit from implementation in copyrightable software? What is different or unique about Biological Open Source that suggests we need something new?

 

Best regards,

 

/Larry

 

Lawrence Rosen

Rosenlaw & Einschlag, a technology law firm (www.rosenlaw.com)

Stanford University, Lecturer in Law

3001 King Ranch Road, Ukiah, CA 95482

707-485-1242  *  fax: 707-485-1243

Author of "Open Source Licensing: Software Freedom and

                Intellectual Property Law" (Prentice Hall 2004)


From: Janet Hope [mailto:janet.hope@anu.edu.au]
Sent: Wednesday, November 15, 2006 5:58 PM
To: license-discuss@opensource.org
Subject: Re: "Biological Open Source"

 

 

On 15/11/2006, at 11:39 PM, Michael Tiemann wrote:

 

Janet, I for one welcome your participation.  Moreover, while it is not

likely to lead to a specific license approval, it is very much on topic

for the expertise this list represents.

 

 

Thanks, Michael, for your positive response.  Having received some private as well as public encouragement and no discouragement, I'll go ahead and put my question to the list.  (It has been suggested that a new list could be started for non-software open source licensing issues to cut down on off-topic posts -- I think that would be very helpful.)

 

This post is long. Most of it falls under the headings "Background" and "Preliminary licence analysis".  Readers who are interested, but not that interested, are encouraged to skip those parts. 

 

THE LICENCES

 

The licensor, CAMBIA, has issued two "Biological Open Source" ("BiOS") licences, one for Plant Enabling Technologies ("PET") and one for Genetic Resources Indexing Technologies ("GRIT").  Each must be read in conjunction with its own Technology Support Services Subscription Agreement ("PET TSSS" and "GRIT TSSS").  All of these instruments are currently in version 1.3 and are available at http://www.bios.net/daisy.

 

Let me say up front that I am not affiliated with CAMBIA and have not been involved in the process of drafting the BiOS licences.  As a matter of courtesy to the drafters, I am simultaneously posting my comments to CAMBIA's own discussion forum.  However, as Michael suggests, I suspect that license-discuss is where much of the relevant expertise is to be found. 

 

THE QUESTION

 

The question is: To what extent is CAMBIA's BiOS licence scheme "open source" in the software sense? 

 

In other words:  If the BiOS licences were software/copyright licences, would you recommend them for OSI approval?  If not, why not?

 

SPECIFIC QUERIES

 

In particular (see the rest of this post for more detail on each of these points):

 

(1) If a licence is primarily a patent licence and the patent grant itself purports to be open source (not just compatible with an open source copyright grant elsewhere in the licence), should field of use restrictions be permitted? (OSD #6)

 

(2) What should be regarded as the equivalent of "source code" in a biotechnology licence? Should it include materials etc in addition to that which would be required for disclosure under patent law? If so, on what terms should it be made available to users? Specifically, is it acceptable to charge a fee that does more than recover the costs for storage/retrieval/postage of samples etc (OSD #2)? Is it acceptable for the amount of the fee to depend on the size and type of the user's organisation (OSD #5) and the length of the user's "subscription" to ongoing "technology support services"?

 

(3) What is the nearest patent equivalent of the open source copyright licence grant to copy, modify, and distribute copies or derivative works (OSD #1)?  Does it depend on the type of patent (eg process or product)?  Is the patent grant in the BiOS licences effectively limited by the exclusion of "Improvements" from the definition of "BiOS Licensed Products"?

 

(4) Is a right to sublicense essential in an open source licence? Is it acceptable for this right to be limited or absent? If so, what other conditions should be met to ensure that a licensor cannot impose non-open source restrictions as a condition of issuing a new licence (eg OSD #7)? How can these conditions be met in the case of a technology to which the equivalent of a copyright notice cannot be so easily attached as it can to software code? 

 

(5) Does the grantback in the BiOS licence go beyond a copyleft-style provision in restricting users' freedom to fork the innovative process? Is there any pressing legal or practical reason why the licensor in a biotechnology/patent context could not simply require licensees to make defined improvements available on the same terms as the original licence, as in the software/copyright context?

 

BACKGROUND

 

CAMBIA (formerly an acronym for Center for Application of Molecular Biology in International Agriculture) is a private, not-for-profit research institute located in Canberra, Australia.  Founded in 1994 by its present CEO Dr Richard Jefferson, CAMBIA is a small organisation that -- atypically -- combines "wet lab" development of biotechnology research tools with intellectual property informatics and policy development.  CAMBIA is financed by grants from philanthropic organisations including the Rockefeller Foundation, by national and international research funding bodies, by official development assistance and by licence revenue from its own patented technologies.(See http://www.cambia.org/daisy/cambia/589.html.) 

 

In 2005, CAMBIA launched an initiative called Biological Innovation for Open Society (BIOS: note the upper case "I").  The BIOS initiative consists of:

(1) Patent Lens: searchable databases containing EPO, US and PCT patent documents, together with ancillary IP-related information and tutorials.

(2) BioForge (styled after Sourceforge.net): a portal for protocol-sharing, comments on patents, and discussion tools in both public and secure environments; intended to develop into a collaborative technology development platform. 

(3) Biological Open Source (BiOS with a lower case "i") is the aspect of the BIOS initiative that is "intended to extend the metaphor and concepts of Open Source to biotechnology and other forms of innovation in biology". (See CAMBIA BiOS License for Plant Enabling Technology Version 1.3, Recitals, first paragraph.)

 

A description of the materials and methods that researchers can obtain from CAMBIA, including an indication of which materials and methods are available under which BiOS licence, is at http://www.cambia.org/daisy/cambia/materials.html.  These technologies are, of course, quite different from those licensed under OSI-certified software licences.  Plant enabling technologies and genetic resource indexing technologies consist not of software code (though software may be one component), but of heterogeneous methods and materials, including living biological materials.  Typically, such technologies are legally protected by a mixture of intellectual and personal property rights, including patents.  By contrast, I understand that most open source software licences are primarily copyright licences, though some also contain a limited patent grant.

 

As I mentioned in my first post, the fact that the BiOS licences apply to plant enabling technologies and genetic resources indexing technologies means that they cannot be certified by the Open Source Initiative as being "open source" licences because the terms of the official Open Source Definition are not broad enough to accommodate the legal and technical differences just described. In consequence, there is no certification signal to help potential users determine whether or not signing up to a BiOS licence will actually give them the nearest equivalent to the "software freedom" guaranteed by the principles of open source software licensing.  Clearly, the task of translating open source licensing principles from one technology setting to another is one that creates room for both legitimate adaptations and distortions of the kind that would lead the OSI to refuse to certify the relevant licence if the technology in question were, in fact, software.  By my preliminary analysis, the BiOS licences contain a mixture of the two.  However, I (and many others) would like to be able to supplement that analysis with the expert views of people on this list.

 

Note that I am in no way impugning CAMBIA's good faith or suggesting that the BiOS licences are "bad".  Certainly, CAMBIA has made a serious effort at a very difficult task.  The licences may well be good licences, and good for innovation.  However, the question here is whether it is reasonable to describe them as "open source".

 

PRELIMINARY LICENCE ANALYSIS (WARNING: MAY CONTAIN ERRORS! TINLA!!)

 

For simplicity, I refer here only to the terms of the BiOS Plant Enabling Technology (PET) agreement. The PET scheme closely resembles the GRIT scheme. The main difference is that whereas the PET scheme contains a limited sublicensing right, the GRIT scheme does not contain any right to sublicense. 

 

1. Overall scheme of the licence

 

The BiOS licensing scheme encapsulated in version 1.3 of the PET Licence and corresponding TSSS Agreement contemplates an initial contribution of IP and technology by CAMBIA that may act as a seed for ongoing technology development. ("IP and Technology" is defined in Licence cl. 1.8.)  Thus, the subject of the initial licence grant is a collection of CAMBIA's own patents, materials and technology data.(See definitions of "CAMBIA Patents", "CAMBIA Material" and "Technology Data" in Licence cll.1.9, 1.10 and 1.13 respectively.)  Downstream, this initial subject-matter may be supplemented by "Improvements", "Improvement Patents" and "Improvement Materials" (Licence cll 1.6, 1.7 and 1.11 respectively) developed by the licensee,  non-exclusively granted back to CAMBIA under Licence cl. 3 and then redistributed by CAMBIA to all BiOS licensees as part of the grant of "IP and Technology".

 

2. Parties

 

Although the licence is described on the BIOS forum as a "template", one party (CAMBIA) is explicitly named throughout the licence text as the licensor.  I understand that this is discouraged in the OSI approval process because it contributes to licence proliferation; however, this is a relatively minor point.  

 

The sense in which the BiOS licences are templates is that they contemplate a number of licensees all receiving CAMBIA's permission to use the IP and Technology on "substantially similar" terms. (See Licence Recitals, paragraph 3. Why are the terms "substantially similar" and not identical?  Two reasons: (1) different licensees may agree to different numbered versions of the licence (see Licence cl. 7.5 and TSSS Agreement cl. 7.6). (2) The terms on which licensees can access any materials needed to practise the licensed inventions vary according to the type and size of the licensee organisation (see TSSS Agreement cl. 3 and Annex D).) 

 

3. Patent grant

 

The licence grant is contained in clause 2 of the PET Licence, the key part of which reads:

"2.1 CAMBIA hereby grants to BiOS LICENSEE under Licensed Patents in the Field of Agriculture a worldwide, non-exclusive, royalty-free right and license to make and use the IP & Technology for the purpose of developing, making, using, and commercializing BiOS Licensed Products without obligation to CAMBIA...."

 

The nature of the grant is crucial to the question whether the BiOS licences can be justifiably described as "open source". Note that unlike open source software licences, the BiOS licence is primarily a patent licence.  While patent grants in open source software licences must be compatible with an open source copyright grant, they are not the primary means by which these licences seek to protect users' rights to access and use the technology.  The question here is not merely whether the BiOS patent licence grant is \textit{compatible} with open source principles, but whether it is itself "open source" (whatever that is taken to mean in the biology context).

 

3a. Grant is restricted to a particular field of use

 

The BiOS patent licence grant is limited to a particular field of use, namely Agriculture. ("Agriculture" is defined quite broadly in Licence cl. 1.3.)  There is a question whether this conflicts with OSD # 6.

 

While a strict formal interpretation of open source principles would presumably prohibit field of use restrictions in open source biotechnology licences, the difficulty with such an interpretation is that in the patent context, field of use restrictions, together with territorial restrictions, may be a useful tool for making technology that would otherwise be bound up in exclusive licences more readily available for public interest and broader commercial use.  The licensing policy of Public Intellectual Property Resource for Agriculture (PIPRA), a collective intellectual property management regime for agricultural biotechnology recently established by a group of land grant universities in the United States, is a case in point.   One of PIPRA's primary aims is to overcome the fragmentation of public sector intellectual property ownership by identifying residual rights retained by members who have assigned unnecessarily broad exclusive rights in important technologies to major commercial firms. One of the architects of the initiative explains the importance of field of use and territorial restrictions in this context: "Best practice includes partitioning of patents: if you go and license something like your agrobacterium technique, license it just for cotton; or better, for cotton in the US; or even better... define which varieties, or... constrain it to varieties owned by the licensee company in the US.  The more you can constrain the space of the technology grant, the more is still left over [that you may choose] to put into the commons." (Greg Graff, personal communication.) Thus, the value of such field of use restrictions in terms of achieving wider access to key biotechnologies may outweigh the value of keeping to a strict analogy with open source software licensing.

 

What do people think about this? I understand that with respect to patented code, the F/LOSS community objects to field of use restrictions in software patent licences because such restrictions are perceived as capable of insidiously undermining freedoms granted in relation to the same code under an open source copyright licence.  However, the approach of open source community leaders to this situation appears to have been to campaign for maximum breadth of patent licences, while remaining willing to compromise for the sake of workability.  Thus, many open source licences, including the Apple, IBM and Mozilla licences, have field of use restrictions in their patent grants. The W3C Patent Policy also represents a compromise.  How far should this willingness to compromise go in a case where the patent grant is the main or even the only grant in the licence?

 

3b. Scope of grant

 

The grant is a "right and licence to make and use the IP and technology for the purpose of developing, making, using and commercialising BiOS Licensed Products".  Note that the grant does not explicitly include a right to sell the IP and Technology itself.  Nor does it permit the licensee to make or use the IP and technology for purposes other than developing, making, using or commercialising BiOS Licensed Products.  "BiOS Licensed Product" is defined in Licence cl.1.4.  That definition is discussed further below, but significantly, it is restricted to assets that are substantially distinct in some respect from the licensed IP and technology and are "intended for commercialisation".  "Commercialisation" is not defined in the BiOS licence.  However, in light of recent US case law concerning patent infringement by universities (eg Madey v Duke), it may be quite broad.

 

The OS mantra is "anyone, anywhere, for ANY PURPOSE, must be free to copy, modify AND DISTRIBUTE the software, for free or for a fee...".  At first glance, the BiOS patent grant seems considerably narrower than this.  However, patent law grants a patent owner a different set of rights to those of a copyright owner.  It is therefore not a straightforward question what the equivalent of the broad open source copyright licence grant should be in the patent context.  It is arguable that an open source copyright grant covers all of the exclusive rights of the copyright owner that are relevant to the licensed technology, and that an open source patent grant should do the same. 

 

IAAL, but not a patent lawyer.  However, I understand that in Australia at least, the precise form of a broad patent licence grant that would cover all of the owner's rights-to-exclude would depend on the nature of the patent claims.  Section 15 of the Australian Patents Act 1990 gives the patent holder the right to exploit the invention. The definition of "exploit" in Schedule 1 of the Act distinguishes between products and processes. For products, it includes the right to make, hire, sell or otherwise dispose of the product. For processes, the definition includes the right to make, hire, sell or otherwise dispose of products resulting from the process. Without looking at the claims in CAMBIA's patents, it appears that they are process patents (ie methods of doing plant enabling technology and genetic resource indexing technology). The licences grant the right to use the methods. Any products are supplied separately under the support agreement. The licences actually grant the right to sell products using the methods under licence. Thus, the licence grant may be broad enough to satisfy open source principles with respect to the initial IP and Technology. 

 

What do you all think about this?

 

3c. Limited right to sublicense

 

A final point to note regarding the BiOS licence grant is that it includes a limited right to sublicense, as follows:

"2.1.1 granting limited sublicenses to third parties... to conduct research and/or development activities...for BiOS LICENSEE, provided that the product and/or other results (including all intellectual property rights) resulting from said limited sublicense are owned exclusively by BiOS LICENSEE, said limited sublicenses to terminate when such activities cease or such ownership terminates, and

"2.1.2 granting limited sublicenses to third parties... for the sole purpose of commercializing BiOS Licensed Products that embody the IP & Technology or are generated by use of the IP & Technology, said limited sublicenses to terminate when said commercialization ceases.

"BiOS LICENSEE shall be responsible to ensure... that any Improvements produced by sublicensees are considered to be Improvements hereunder.... BiOS LICENSEE shall provide a list of sublicensees to CAMBIA in writing at least once a year....

"BiOS LICENSEE shall further be responsible to ensure that...third parties to whom a sublicense has been granted are notified... that said third parties have no right to sublicense absent the execution of a BiOS License Agreement with CAMBIA, and that CAMBIA may be approached for a BiOS License Agreement.

"Other than stipulated under this Article 2.1, no further right to sublicense is granted to BiOS LICENSEE hereunder."

 

I am unclear as to whether a right to sublicense is essential for OSI certification of new licences.  My understanding from Larry Rosen's book is that the right to sublicense is considered desirable, but is not a feature of all OSI-approved software licences.  I imagine that in the absence of a sublicensing right, OSD #7 becomes very important because unless people are sure they are entitled to a licence no matter what, the need to go back to the initial licensor for permission to use the technology would not be a mere formality and could become too restrictive.  However, it is not clear exactly how OSD #7 would operate in relation to the exchange of biological materials, methods etc, as distinct from code or other content to which a copyright notice can be easily attached.

 

Can anyone clarify the OSI requirements re sublicensing?  (Sorry if this has been dealt with in previous posts.)

 

4. Source code

 

The freedom to copy and modify open source software programs and to distribute copies and modifications is only one important aspect of open source licensing.  Another is the practical enablement of that freedom through unrestricted access to software source code.  In the case of patented biotechnologies, it may be argued that the equivalent of source code is automatically publicly available because of the disclosure requirement under patent law, and that an open source-style licence grant is all that is needed to render a patented technology "open source".  On the other hand, biotechnology licensors typically go beyond this level of disclosure when attempting a genuine transfer of technology.  This practice suggests that patent disclosure many not be enough to give full practical effect to the legal freedom to use and improve a technology that is intended to be conferred by an open source licence grant. 

 

In the BiOS PET licensing scheme, separate provision is in fact made for access to materials needed to practise the licensed patents.  This access is governed by the terms of the TSSS Agreement (see Licence cl. 4) and appears to be conditional upon (1) a substantial annual payment for at least three years, depending on the licensee's organisation type and number of employees and (2) significant reporting requirements regarding new materials, technology data and project ideas.  If access to materials needed to practise the licensed payments is regarded as equivalent to access to source code, these restrictions represent a significant departure from established open source principles of non-discrimination among licensees, access to source code for free or at the minimum charge required to recover the provider's costs, and freedom from ongoing obligations to the licensor, including obligations that take the form of reporting requirements.

 

5. Grantback

The next phase in the scheme established by the BiOS licences and TSSS agreements is triggered when a licensee, in the course of exercising the licence grant, makes or discovers something new. 

 

The BiOS PET licensing scheme distinguishes between "BiOS Licensed Products", which may be developed, made, used and commercialised "without obligation to CAMBIA", and "Improvements", which are subject to a grantback obligation  (see below) that also covers Improvement Patents and Improvement Materials (Licence cll. 1.7 and 1.11, respectively).  The definitions of these different categories of potential follow-on innovations are complicated and tinged with uncertainty.  However, the fundamental intention appears to be to allow the licensee to appropriate any new asset that has been generated through use of the IP and technology and is intended for commercialisation.  Such an asset is a "BiOS Licensed Product" (Licence cl. 1.4), provided it does not fall within the definition of "Improvement" (Licence cl. 1.6).  

 

The definition of "Improvement" is thus a critical element of the BiOS scheme. This definition is potentially extremely broad.  Note in particular that even something as minor as a repeatable observation can constitute an Improvement, and that an Improvement need not be a Plant Enabling Technology (itself very broadly defined in Licence cl. 1.12) because of the catch-all "but for the terms of this License Agreement...". 

 

5a. Does the definition of "Improvement" limit the licence grant itself?

 

A preliminary question is whether the broad definition of "Improvement" in Licence cl. 1.6 effectively limits the scope of the main licence grant in Licence cl. 2.  As noted earlier, the purpose of the grant is limited to "developing, making, using and commercialising BiOS Licensed Products".  "BiOS Licensed Products" is defined in cl. 1.4 to mean "any tangible or intangible asset of BiOS LICENSEE (including without limitation any material or method, but excluding Improvements)".  This suggests that anything that falls within the definition of Improvement automatically falls outside the definition of BiOS Licensed Product and therefore is not covered by the grant even for purposes other than commercialisation.

 

Similarly, any new development that is neither a BiOS Licensed Product nor an Improvement does not appear to be covered by the terms of the licence.  In either case, further permissions may be required in order to avoid infringing the licensor's proprietary rights.  CAMBIA appears to be free to make such permissions conditional upon payment of royalties or other conventional licensing terms -- or to refuse them altogether.

 

5b. Grantback: copyleft equivalent or licensing "club"?

 

From an open source perspective, the breadth of the definition of "Improvements" is not necessarily a problem, although it could be if it extended the range of follow-on innovations that would be caught by the grantback well beyond those that would be caught by a copyleft-style provision relating to "derivative works".  As I understand it, the question of what constitutes an appropriate incentive for licensees to innovate in a copyleft-style licence is a separate issue from compatibility with open source principles.  The broadest examples of copyleft "hooks" in open source software licences are tailored to catch all derivative works that are distributed outside the boundaries of the licensee's own organisation; other open source licences permit varying degrees of freedom to operate with derivative works. The BiOS grantback appears to fall somewhere in the middle of this spectrum.  In a 2004 article on open source patent licensing, Sara Boettiger and Dan Burk point out that copyright confers exclusive rights only against unauthorised copying or other violations of the specifically enumerated rights of the copyright owner arising out of contact with the copyrighted work itself.  By contrast, independent creation is not a defense to claims of patent infringement. The BiOS licence appears designed to replicate the copyright situation by excluding from the definition of Improvement improvements that are "developed without the use of the IP and Technology".  Similarly, the BiOS arrangement appears intended to mirror open source software licences by permitting in-house use of improvements that are actively protected as trade secrets: these are not caught by the BiOS grantback. (Note, however, that this escape clause may be somewhat narrower than the positive requirement in open source software licensing for a derivative work to be externally deployed before it triggers any copyleft obligation: clearly, there is a difference between active dissemination of a technology and merely failing to  adequately protect a trade secret.  Further, note that in-house use of an improvement under conditions of trade secrecy may not be permitted at all under the BiOS scheme unless interpreted as a form of "commercialisation": see Licence Grant in cl. 2.2 and definition of BiOS Licensed Product in cl. 1.4.)

 

The point at which the BiOS licence provisions relating to follow-on innovations appear to me to depart from open source principles is not in the breadth of what may be caught by the grantback, but in the fact of the grantback itself.  While the copyleft or "reciprocal" obligation to be found in some open source licences has been characterised in academic literature as a type of grantback, this does not appear to me to be an accurate description.  A copyleft licence does not establish a tit-for-tat relationship between the licensor and licensee.  Rather, a copyleft licensor says to the licensee: "Do as you have been done by".  The copyleft obligation is not to the licensor per se -- though he or she may be responsible for enforcing it -- but to the whole community of potential users of the follow-on innovation.  The licensor may or may not be a member of this community, depending on the scope of his or her activities and interests.  For this reason, I prefer to describe the copyleft obligation as a "grantforward" (as in "passing it forward") rather than a "grantback".

 

The BiOS grantback, by contrast, appears to be just that: whether or not the intention is noble, it still constitutes a type of privilege granted by the licensee to the licensor in partial consideration of the licence grant.  Under cll. 3 and 4 of the Licence and relevant provisions of the TSSS Agreement, a BiOS licensee is obliged to grant to CAMBIA a worldwide, non-exclusive, royalty-free licence (with the right to sublicense to other BiOS licensees) to Improvement Patents, Improvements, Technology Data, and any  Improvement Material provided by the licensee to CAMBIA and necessary to practise Improvements. (Definitions are in Licence cll. 1.7, 1.6, 1.13 and 1.11 respectively.)  Thus, the form of the BiOS licence differs from a copyleft open source licence: instead of requiring licensees to license a subset of follow-on innovations on the same terms as those of the original licence, it builds the follow-on licence into the original licence and places the initial IP owner, CAMBIA, at the centre of the network as a kind of gatekeeper. While CAMBIA does have an obligation to make these follow-on innovations available on the same terms as the initial licence, that obligation extends only to existing BiOS licensees.  In essence, the arrangement is a licensing "club".  Given the restrictions on sublicensing by BiOS licensees (Licence cl. 2), new members can join this club only by executing a new licence with CAMBIA; CAMBIA is under no obligation to grant such a licence.

 

Club-type licensing arrangements are actually quite familiar in the biotechnology context.  Depending on the circumstances, such an arrangement may be good for innovation, but it seems to me that it is not open source. Why?  Because an arrangement in which the initial licensor retains a central position linked to his or her ownership of the seed IP and Technology is inconsistent with the freedom to fork the innovative process. 

 

The freedom to create a ‘code fork’ is regarded by some as a defining characteristic of open source -- recall Eric Raymond's open letter to Sun a little while back. According to Steve Weber (author of The Success of Open Source), under the terms of an open source licence, anyone who is dissatisfied with the conduct of a project leader -- on technical, administrative, political or even purely personal grounds -- is free to take the collaborative effort in a new direction. Even though in practice, forking is rare, the ever-present possibility makes project leaders responsible to their co-developers and ensures that no individual or group unduly dominates the process of technology development. At the same time, it ensures that a promising technology need not be left on the shelf because of waning interest or incapacity on the part of an initial innovator.  This is a concern often expressed on this list in the form of: "What happens to the licence if So-and-so, the licensor, goes bankrupt or disappears in ten years' time?"

 

Of course, all else being equal, it is natural for an initial innovator to remain in charge of ongoing development and to act as a champion of the technology. However, it seems key to the open source approach that the initial innovator not use his or her ownership of the intellectual property in the initial seed technology to retain control over its ongoing development. 

 

It appears to me that this freedom to fork the innovative process with or without the approval of the initial licensor is lacking in the BiOS scheme.  Why?

(a) The grantback incorporates stringent reporting requirements to CAMBIA (Licence cll. 3.2 and 3.4) and obliges the licensee to provide CAMBIA with any materials necessary to practise Improvements (Licence cl. 3.2).

(b) Discussions of technology data between BiOS licensees must be channeled through CAMBIA (TSSS Agreement, cl. 2.3). 

(c) CAMBIA retains the right to impose an obligation on licensees to include CAMBIA’s trademarks in downstream technologies (Licence cl. 7.9).  (This provision closely resembles one that I thought was removed from an early version of the open source BSD software licence because it imposed too much of a restriction on the freedom of downstream developers -- is this correct?)

(d) Under Licence cl. 7.5 and a similar provision in the TSSS Agreement, CAMBIA "may publish revised and/or new versions of the BiOS License for Plant Enabling Technologies from time to time. Such new versions will be similar in spirit to the present version, but may differ in detail to address new or newly identified issues. Each version will be given a distinguishing version number. ... Where there is... any controversy between the parties respecting the interpretation or application of the terms of this Agreement, the latest... version of the Agreement published on the BIOS website shall be controlling."

(e) Perhaps the most extreme instance of CAMBIA's apparent attempt to keep its hand on the reins of future development is in the TSSS Agreement, cl. 6, which reads in part:

"From time to time BiOS LICENSEE may have interest in exploring solutions to certain problems in its field of business. In such case BiOS LICENSEE shall [emphasis added] send a written project proposal to CAMBIA. To the extent that such project proposal is consistent with the terms and intent of the BIOS License, which consistency shall be determined by CAMBIA in its sole discretion, CAMBIA shall use its best efforts to... bring the proposal to the attention of other parties within the BIOS Initiative...". 

In other words, the licensee is obliged to send a written proposal to CAMBIA, but CAMBIA is not obliged to do anything with it.  The effect (whether or not intended) is that CAMBIA gets a private preview of any new project.

 

Thank you for your input.

 

Janet