Subject: Re: "Biological Open Source"
From: Janet Hope <janet.hope@anu.edu.au>
Date: Thu, 16 Nov 2006 17:02:47 +1100
Thu, 16 Nov 2006 17:02:47 +1100
Larry,

the way!

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

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
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
> 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.
>>
>>
>
>
> 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
>
>
>
> 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
> 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:
>
> 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
> 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.
>
>
>
> 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
> 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
>
>
>
> 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.
>
>
>
>
>
>
> 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
> 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
>
>
>
> 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
> 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
> 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 needed to practise the licensed payments is regarded as
> significant departure from established open source principles of
> 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
> 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
> 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
> 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
> 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



Larry,

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:
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.

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.

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.

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"?

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.

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