JSR358-26: Keep or eliminate the obligation to license Essential Patents on FRAND terms for JSRs in which you do not participate

[apastsya]

Jira issue originally created by user pcurran:

Section 6 of the JSPA (Special Patent Considerations) requires that all JCP members, even those who do not participate in the development of a JSR, grant essential patent rights to all licensees of that JSR.

This provision may be a barrier to corporate participation.

Could it even be enforced against individuals or their employers?

Should we keep it, or can we drop it?

Note: our current processes with regard to individuals (Exhibit B) mean that these broad grants are not made by individuals or their employers (Exhibit B refers to a specific JSR only.)

[apastsya]

Comment created by pcurran:

For what it's worth (not much!) my interpretation of this as a non-lawyer is:

• This section makes extremely broad IP grants to the Spec Lead (you are obliged to license your "essential patents" for any JSR - even one that you had no involvement in, and maybe didn't even know existed.)
• We currently do not require equally broad grants from individuals as opposed to organizations (Exhibit B references only a specific JSR.)
• I have doubts whether it would actually be enforceable in a court of law, particularly against individuals or their employers.
• To my knowledge this has never been invoked in the history of the JCP.
• It could be an obstacle to corporate involvement in the JCP (Apple resigned several years ago because of this section of the JSPA.)
• If we don't eliminate this section then it will be effectively impossible to create a significantly simpler subset of the JSPA for members who do not wish to serve as a Spec Lead (something we would very much like to do.)

[apastsya]

Comment created by john_rizzo:

We are voting no for now due to two reasons first it gives Oracle special considerations (in the last subsection). Second is the use of the term FRAND which we have all agreed is problematic due to it widely varied definitions. It is vague at best, at worst it can be situationally defined. I am seeking suggested changes.

[apastsya]

Comment created by ddeutsch:

Oracle supports removal of the requirement that JCP members not engaged in an expert group must commit their essential technology but believes futher discussion is needed to consider everything specified in section 6. In particular, we want to ensure that the JSPA specifies the commitments required of expert group participants for claims on technology that they did not contribute. Unless this is addressed elsewhere in the JSPA, simply deleting section 6 may not be appropriate. A "Term Sheet" should be agreed specifying patent commitments for at least the following situations:

• Spec Lead

• Member of expert group their contributions contributions of others

• members of the JCP not in the expert group

[apastsya]

Comment created by pcurran:

(In response to Don's comment...)

Agreed. It's too simplistic to say we should just delete this section. There's other stuff in it (for example, it explains your rights to withhold a patent grant under certain circumstances.

We really need to take a comprehensive look at the whole IP-flow/IP-grant matter, understand what the current document says, and decide whether this is what we want it to say.

[apastsya]

Comment created by mmilinkov:

My concern is that deleting this section may actually cause more harm than good. As I read the JSPA, if Section 6 is deleted, then there is no obligation for any patent license grant. My preference would be to replace "FRAND" with "royalty free", which would certainly go a long way to helping enable open source implementations.

[apastsya]

Comment created by pcurran:

In response to comments here and discussion within the EG we narrowed this question down to:

Section 6 of the JSPA obliges members to license Essential Patents (unless they proactively opt out of doing so) even for JSRs with which they have no involvement. Should we limit this obligation to those who are members of the Expert Group?

The overwhelming response to a doodle poll on this question was that we need further discussion.

For the record, I will post comments made on that poll to this thread.

[apastsya]

Comment created by pcurran:

Gil Tene said in a comment on the doodle poll:

I think that this item should be discussed in the context of overall patent grant obligations, as well as reciprocity requirements. Specifically, section 6 is currently linked to from section 5.C.1, and losing that could be an issue in the overall IP picture. We need to discuss what will be left at the end - e.g. I believe that at the very least, we need spec IP rights to be able to be licensed under terms that require non-assertion against compatible implementations of the spec...

[apastsya]

Comment created by pcurran:

Mike Milinkovich said in a comment on the doodle poll:

I would like to suggest a royalty-free patent licensing model as a potential solution to this. Both the Apache and Eclipse open source licenses have a very similar model that lays out the requirements that any patents which read on contributions must be licensed royalty-free. This model has seemingly worked very well for those communities, and I would be interested in exploring any shortcomings the model may have in the JCP context.

[apastsya]

Comment created by pcurran:

Steve Wolfe said in a comment on the doodle poll:

We are open to considering all aspects of the IP flows in the JSPA, but need to do so in the context of the complete IP picture. Having said that, we believe one of the JCP's key objectives should be making the Java ecosystem and our customers confident that Java is a safe place to invest, and to achieve that, we believe JCP members should support a model where Java investments are not at risk from unpredictable IP claims.

[apastsya]

Comment created by pcurran:

I should have read Mike Milinkovich's comment more closely when I pasted it! He said:

"I would like to suggest a royalty-free patent licensing model as a potential solution to this. Both the Apache and Eclipse open source licenses have a very similar model that lays out the requirements that any patents which read on contributions must be licensed royalty-free. This model has seemingly worked very well for those communities, and I would be interested in exploring any shortcomings the model may have in the JCP context."

This is exactly what we have today for contributions. The question is about patent rights with respect to stuff that you do not contribute (but without which someone else's JSR cannot be implemented.) Note that the Spec Lead (for everything, whether or not formally contributed) and Expert Group members (for contributions) are already obligated to license royalty-free.

The RAND obligation really is a corner case, and only covers contributions made by others and which result in the inability to implement a JSR without violating your patent. I'd be interested to know how often money changes hands in practice as a consequence of this requirement.

[apastsya]

Comment created by pcurran:

What's in question is not the elimination of Section 6 in its entirety, but rather the elimination of the obligation to license Essential Patents on FRAND terms for JSRs in which you do not participate. I will reword the issue accordingly.