Subject: RE: "Biological Open Source"
From: "David RR Webber \(XML\)" <david@drrw.info>
Date: Thu, 16 Nov 2006 09:10:56 -0700

Janet, 

If its patents - then perhaps the solution Sun is using makes sense? 

Sun announced the issuance of a Non-Assertion Covenant (NAC) for UBL.

 The full text of the Sun UBL NAC can be found at

  http://www.oasis-open.org/committees/ubl/ipr.php

and then Jon Bosak's email here: 

 http://lists.oasis-open.org/archives/ubl-dev/200607/msg00005.html 

This could easily be combined with an OSI license then - if some
combination of technology is occurring -  

e.g. software that explores genonomic permutations based on patented
genome sequences...? 

DW

 -------- Original Message --------
Subject: Re: "Biological Open Source"
From: Janet Hope <janet.hope@anu.edu.au>
Date: Thu, November 16, 2006 1:02 am
To: license-discuss@opensource.org
Cc: lrosen@rosenlaw.com

Larry,  

 
Thanks for your response.  I learned a lot by reading your book, by the
way! 

 
I think we talked about this very issue when we met back in 2003.  My
question is this.  Suppose you have a technology that has nothing to do
with software whatsoever -- for example, a method for creating a
genetically engineered plant.  Suppose that the technology is protected
solely under patent law.  In that case there is no copyright licence,
because there is no copyright.   

 
So, I'm talking about a situation in which the assumption that
Presumably software is involved in Biological Open Source or we
wouldn't be discussing this here at all. does not hold.  (As I
mentioned in my earlier post, strictly speaking this discussion is off
topic for that very reason and may end up getting moved to another
list.) 

 
 
Now suppose that the owner of this patented technology wants to
distribute it in a way that achieves the same substantive outcomes in
terms of "technology freedom" and ongoing collaboration as a copyright
owner seeks to achieve by adopting an open source copyright licence --
the same *substantive* outcomes, I say, but inevitably through
different licensing *forms* given that there is no copyright but only a
patent right.  How would he or she go about it?  

 
As I understand it, this is what CAMBIA have tried to do with their BiOS
licence, and they have solved many of the obvious problems
satisfactorily.  I've identified a few areas where I think the
practical effect of the licence is more restrictive than would be
permitted in the case of an open source copyright licence.  What I'd
like to know is, could CAMBIA have built a better (meaning closer)
analogy with the open source copyright approach?  Or are these
differences inevitable given the different legal and/or technical
context in which they are working? 

 
If I understand correctly, this is a slightly different problem to that
addressed by the patent licence grants you mention, because they in a
sense are merely "getting out of the way" of an open source copyright
licence, not doing all the work of implementing an open source-style
approach in the absence of any copyright grant. 

 
On the other hand, I don't see why my hypothetical patent owner wouldn't
just use language very similar to that included in OSL 3.0.  The only
part of that language that wouldn't make sense in the absence of any
copyrighted work is the reference to an Original Work or Derivative
Work.  Nevertheless, CAMBIA have chosen to use quite different wording
in their grant (BiOS PET License Clause 2), and I'm wondering whether
that wording looks open source to you.  Assuming the patent is a
process patent, are they granting every possible freedom to licensees,
or are they holding something back?  If they're holding something back,
can they still call what they are doing "open source"? 

 
Does all this make sense? Or am I still missing something? 

 
Cheers 
Janet  

 

 
 

  
 
   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 
   
   
   
   
   
   
  

 
 
Dr Janet Hope 
Centre for Governance of Knowledge and Development 
Research School of Social Sciences 
Australian National University 
T: +61 2 6125 0172 
F: +61 2 6125 1507 
janet.hope@anu.edu.au 
http://rsss.anu.edu.au/~janeth