Please note that this discussion has been continued on the page: Archetype licensing - the case for CC-BY-SA

Introduction

Background

The licenses governing the Intellectual Property (IP) of the openEHR Foundation need to be modernised. Commercial users wishing to base products upon openEHR have raised this matter with the Board and it is accepted a comprehensive revision is needed. The IP status of archetypes has also been raised as a potential issue in standards circles.  The main needs are:

• Clear copyright and license for all types of openEHR IP;
• Flexible, easy to understand rights are needed for commercial users;
• The integrity of original specifications, derived schemas, and knowledge artefacts needs to be protected while allowing derivative works to be made.

To achieve the necessary results, there are various tools available for controlling and managing IP, including:

• Copyright – identifies original author(s) for purposes of asserting rights, which are defined on a national basis, although some pan-national copyright law consolidation is starting to occur;
• Explicit license – defines specific terms of use rather than defaulting to copyright law;
• Trademark registration – defines terms of use of trademark, service marks;
• Digital signing of artefacts – establishes origin and integrity;
• Public issue trackers – encourages public to report problems with IP, e.g. bugs, errors (helps avoid use of viral licenses).

Resolution process

A final date of June 30 is proposed for resolving the license and copyright questions above. Comments can be made on this wiki page or on the openehr-technical list. Private submissions can be sent to Prof David Ingram (d.ingram AT ucl.ac.uk).

Requirements

The table below describes the currently known requirements for each kind of openEHR IP. The main aims are to determine what license to use, and who the copyright should be to.

Terminology-related IP Issues

Archetypes, templates and terminology 'ref sets' contain codes from published terminologies. Codes are routinely used in archetypes in the 'bindings' section, and terms (i.e. the rubric of codes) are used in a similar fashion inside templates. Clearly this needs to be doable without violating any IP restrictions imposed by the relevant terminology publishers. It is proposed that the most sensible approach is for terminology publishers to allow:

• A limited number of codes & terms to be used for any archetype or template, and
• No relationships to be included, since this would then act as a copy of terminology content that may go out of date.

Terminology ref sets are another matter. The current IHTSDO model of a ref set is essentially a list of code points. Ref sets as implemented elsewhere include relationships as well, since the idea is to act as an extract from the terminology with relationships intact, in order to allow browsing and potentially reasoning on the refset.

The openEHR Foundation will work with IHTSDO on this issue in order to determine the rights of use for SNOMED CT, LOINC and ICDx terminologies.

Discussion summaries regarding CC-BY versus CC-BY-SA for content models

Summary of views by Erik Sundvall (preferring CC-BY)

There has been mail discussions on (and off) the mailinglists regarding what license to use for content models (exemplified above by openEHR.org archetypes, templates, archetype-based queries).The concrete suggestion from the openEHR foundation board is tu use CC-BY-SA with some homegrown additions. In my view the argument right now boils down to:

• The main difference between CC-BY and CC-BY-SA is that CC-BY-SA in addition to the CC-BY also adds the requirement that derived works that are distributed must be distributed under an open license. The board now for some strange reason wants to use CC-BY-SA with a loosely formulated addition that derived works "do not have to be openly distributed". I have not heard any reasonable motivation what this would gain compared to just using CC-BY from the beginning. In open source communities homegrown additions to (or modifications of) licenses are known to be potential problem sources. CC-BY and CC-BY-SA (without homegrown additions) have been scrutinized by lawyers and have been translated and developed to fit legislation in many countries, it would take a lot of work to do the same with a homegrown addition to CC-BY-SA - and if there is no well motivated gain, then why not just use CC-BY in the first place?

Some previous mailing list posts regarding this, commented from Erik's perspective:

• January 2006 - The importance of having very well understood open licenses for archetypes was very well formulated by Tim Churches http://www.openehr.org/mailarchives/openehr-technical/msg01774.html
• 1 September 2009 - David Monér starts an interesting thread: http://www.openehr.org/mailarchives/openehr-technical/msg04573.html Some excerpts from the discussion thread follow:
  • Tim Cook says the board should make a statement: http://www.openehr.org/mailarchives/openehr-technical/msg04577.html
  • Erik Sundvall discusses hard-to-interpret situations if using SA and quotes himself from off-list discussions from 2008: http://www.openehr.org/mailarchives/openehr-technical/msg04579.html
  • Sam Heard says -SA will "ensure that specialised or adapted archetypes based on openEHR archetypes remain freely available" and that other derived works won't be a big issue http://www.openehr.org/mailarchives/openehr-technical/msg04622.html
  • Richard Dixon Hughes (a lawyer) says and exemplifies that derived works might be a tricky area http://www.openehr.org/mailarchives/openehr-technical/msg04623.html
  • Erik Sundvall responds to Sam Heard, especially regarding the expressed fear of modified redistribution of archetypes: http://www.openehr.org/mailarchives/openehr-technical/msg04624.html
  • Sam Heard responds to Erik Sundvall http://www.openehr.org/mailarchives/openehr-technical/msg04628.html
  • Thomas Beale says copyright does not always have to be assigned to openEHR http://www.openehr.org/mailarchives/openehr-technical/msg04632.html
  • Erik Sundvall finds Sam Heard's reasoning changing and contradictory, asks for clarification and responds to Sam's other points http://www.openehr.org/mailarchives/openehr-technical/msg04631.html
  • Sam Heard explains parts of his thought process and then says things like "Your arguments for not using SA are well put" and "BY alone is clearly a choice, SA adds a major condition that we need to consider carefully." http://www.openehr.org/mailarchives/openehr-technical/msg04639.html
  • The continued discussion then shifts focus away from Archetype licensing (to CKM and other things).
• February 2010 - Erik repeats his fears that SA will hurt trust in openEHR: http://www.openehr.org/mailarchives/openehr-technical/msg04892.html and a reply where Tom B finds the basic argument solid http://www.openehr.org/mailarchives/openehr-technical/msg04893.html

The fact that objections to using CC-BY-SA were openly declared on the mailing lists, but not properly answered before the decision by the board to adopt CC-BY-SA, makes me wonder how the decisions in the board are made with respect to response from the community. In a (mostly off-list) mail debate from August 2008 and in the (on-list) debate from September 2009 (referenced above) I sensed that Sam Heard (who is part of the board) had some strong undefined feeling that CC-BY-SA would be more beneficial to the openEHR community, but he could not make a strong case for it when confronted. Is the case that Sam or somebody else has later presented strong arguments to the board that could not and will not be presented to the community in the mailing lists or on the wiki? If arguments can not be presented openly, then the risk increases that people suspect the board or some of its members to have hidden agendas, and that is a bad thing for an open project like openEHR.

Summary of views by Sam Heard (preferring CC-BY-SA)

Thanks to Erik for the excellent summary. My personal concerns are best expressed in the following way:

• openEHR archetypes express (and may refine) clinical knowledge in widespread use
• The openEHR specifications mean that sharing of archetypes is the basis for interoperability, the fundamental motivation for this work. Ensuring that archetypes derived from shared archetypes are also freely available seems a sensible approach to avoid any licensing issues for archetypes in use
• CC-BY-SA is more restrictive than CC-BY and a shift from CC-BY-SA to CC-BY is far less problematic than moving from CC-BY to CC-BY-SA
• Erik may be correct that specific provisions are not easy to express. By limiting the derived works to which this license applies to openEHR archetypes and templates would appear to avoid this issue.

Summary of views by Thomas Beale (preferring CC-BY)

I think that the key things we need to preserve with respect to archetypes, & templates are:

• A - allowing an entity or individual to create a private archetype or template (it is all the same thing in ADL 1.5 anyway). This is clearly needed for

• • purposes of research (where some kind of local sharing / distribution was happening)
  • purposes of local use, e.g. hospital-authored and -used templates
  • commercialisation of equipment whose data may be described by templates
  • building of application and other software based on archetypes and templates that may or may not be published openly
• B - preventing an entity or individual from patenting or otherwise 'occupying' part of the archetype / template space, i.e. preventing others from authoring (nearly) the same archetype or template by their own means.
  • this point is roughly equivalent to the problem of patented software, which is by and large not an accepted practice (see http://en.wikipedia.org/wiki/Software_patent).
• C - encourage openness, cooperation and innovation
  • note that, as with other technology, the latter might be best encouraged in some circumstances by private competition rather than single open source efforts.

CC-BY-SA would seem to prevent all cases of the point A above, where the artefact in question was derived from some openly published CC-BY-SA artefact. It might technically prevent B, but B isn't likely to be supportable even with CC-BY for the same reason as software patents not being widely accepted. C is not a function of the licence, it is a function of the technology, formalisms, governance mechanisms, and general health of the community. On the other hand, CC-BY will allow A, probably make no difference to B, and won't get in the way of C. The incompatibility of CC-BY-SA with A) above could lead to a competitive archetype-authoring effort.

Summary of views by Andrew Patterson (preferring CC-BY)

• Patenting should not even enter into our thought processes when it comes to looking at licenses for archetypes and templates. Even in a jurisdiction with
  broad definitions for patentable subject matter, there is no possible way an artifact such as an archetype could possibly be patentable. This is for a whole bunch
  of reasons, including novelty (archetypes are essentially documentation of common medical knowledge), but mainly the fact that it is neither
  a manufactured item, nor even an algorithmic process. So don't even worry about it. And even if you want to worry about it, the reality is that
  it doesn't matter what copyright license you put on an archetype - if someone wanted to try to patent something in this area (perhaps a generic implementation
  algorithm for archetypes) - they can do it without any reference to specific archetypes, hence the license is irrelevant.
• You can't solve everything with the law - commercial entities should want to maintain compatibility with offically published archetypes because
  the benefits of sharing/interoperability/community involvement outweigh any possible benefits they might get by keeping them secret.
• Custom changes to common licenses are evil. Either CC-BY-SA or CC-BY - not CC-BY-SA with changes.

Please note that this discussion has been continued on the page Archetype licensing - the case for CC-BY-SA