In anticipation of a workshop/discussion meeting we are holding on September 24th, I have produced a short position paper, “Suggested Principles for Structuring Standards” (PDF 580kB), to outline the approach I speculate we should take in future in creating specifications and standards for learning education and training.

Responses, alternative views etc are most welcome before, during or after the meeting. I hope other attendees contribute papers and expect to synthesise conclusions of the meeting shortly afterwards.

At last week’s  Future of Interoperability Standards in Education meeting, one of the issues that came up in the discussion group that I was in was that the creation of “derivative works” was a serious unresolved issue. I summarised this in the plenary feedback as “The ability to create derivative works is an ESSENTIAL issue. There are cases when divergence is damaging but also when [necessary] derivation is prevented. How to resolve this paradox?” This is rather cryptic as it stands so I will expand.

The paradox  is that derivation from one standard (I am using “standard” loosely to include pretty much any documented set of technical conventions) to create another is both desirable and undesirable. It is desirable because communities and applications differ, because standards mature, etc and one size will not fit all. It is undesirable because benefits are realised when more people do something in the same way, not to mention confusion arising from proliferation. It seems that there is a Network Effect with standards. I like describing this as a “paradox” as it conveys the idea that we might not be looking at the problem in the right way. An alternative description might be that there are “conflicting issues” in educational technology standardisation (see Dan Rehak’s position paper).

Having discussed this issue with a couple of people since the meeting and reflected a little, I would like to explore how we might start to resolve to the paradox (I do not aspire to actually resolve the matter into self-evident statements). My thinking has similarities to the capabilities and maturity model in Dan’s paper in trying to separate out tangled concepts.

I believe there are three strands to tease out:

1. “Derivation” covers a multitude of different kinds of use. The term “derivative work” has an overlay of meaning from mainstream writing and publishing that is probably not appropriate for many of these “kinds of use”.
2. There is a spectrum of intellectual contribution to a standard from the development of conceptual models to the creation of the published document.
3. “Standard”  covers a multitude of different kinds of artifact.  Attempts to apply labels such as “formal”, “informal” or “specification” usually lead to fruitless argument.

Kinds of Derivation

I am referring here to derivation of a published document (and again using a loose meaning for “standard”). Looking at the different kinds of derivation, with labels-of-convenience that are not intended to following any conventional definitions, I suggest that some of the kinds of derivation that are relevant to standardisation are:

Ratify (cf. “ratify a treaty”)

The standard is taken as-is from its source. Although it may be re-published or referred to by a new identifier or name it is not revised. This form of derivation might be used to create a national standard that mirrors an international one. There would normally be a standing agreement that ratification can or should occur. It is immaterial from a technical point of view which one is used.

Adopt (cf. “adopt a child”)

The standard is taken on by a new organisation or ad-hoc group and the existing organisation/group relinquishes its ownership. “Ownership” implies full control over the future development, publication, transfer of rights etc. So long as the transfer is properly communicated, adoption should not necessarily lead to negative effects.

Spin-off

A snapshot of the standard is taken by a new organisation and reworked according to its documentation conventions. This is a kind of “re-work” (see below). The new work is compatible at a technical level (syntactic and semantic). The new organisation manages the creation and (editorial) maintenance within the bounds of technical compatibility while the originating organisation can continue to exert full control over its version. It is immaterial from a technical point of view which one is used at the point of departure but the originating organisation must accept more constraints on future plans as they cannot deprecate the spin-off (which will have its separate implementers).

Profile

A new work is created that includes elements of a published standard by reference. The new work may include extensions, value lists (aka vocabularies) and additional constraints. Profiling is only possible when the published standard is both persistently available (as a specific version) and structured in a way to allow for the necessary references to be made. This is not a re-work; it is more like original work with citations. While we may wish to avoid needless proliferation of profiles in the interests of realising a Network Effect, profiles are significantly less damaging than re-works as they make clear the points of reuse and divergence.

Re-work

A new work is created that takes an original and makes changes: additions, modifications and deletions. When both the original and re-work are in circulation confusion is created and the effectiveness of both new and original work is harmed. This is what I would expect would conventionally be referred to as a “derivative work”.

Spectrum of Intellectual Contribution

I am not an expert in intellectual property law and may have committed faux pas: use the comment facility.

The concept of “derivation” as expanded above does not apply equally to all of the stages of activity that underpin the publication of a standard. Here, I try to stereotype four kinds of contribution for which “derivation” is only relevant to the second two. The practical difficulty is that these kinds of contribution are often mixed together in the process. Maybe we should look to separating them into pairs and applying different processes. The stereotypes are:

Development of Conceptual Model

I recognise that following is rather shallow from a philosophical point of view and that I am adopting something of a social constructivist point of view.

Conceptual models are shared abstractions of the world. At some point in time a conceptual model must be documented in the standards process but the conceptual model is a social knowledge-construct independent of its representation/documentation. Hence conceptual models are not subject to ownership or intellectual property assertions. If it is just my idea it is not a shared abstraction: not a conceptual model. The development of a conceptual model requires broad participation and discourse to be accurate and hence useful. Evolution of a conceptual model that is documented in a published work should not be considered “derivation” of that work.

Development of Technical Approach

This would include the creation of information models, decisions on patterns and components to profile, technical trialling etc. This is the solution to a problem independent of its description. It is the knowing-how-to: techne. This kind of contribution is the realm of patent law. Contributors should expect to contribute under RAND or royalty-free terms but not to transfer all rights or they may choose to make public non-assertion covenants. A contributor is free to re-use their contribution (NB not the standard incorporating it) but not necessarily the contributions of others. This re-use is not “derivation” (as above).

Contribution of Prior Work

This category of contribution may be broken down along the same lines as “Kinds of Derivation”.

Creation of Published Document

The creation of content, review and editing of “the standard” as a published work is clearly the most concrete part of the process. Without the precisely documented expression, the underpinning conceptual model and technical approach are not directly useful as a standard. It is at this end of the spectrum that contributors should expect to grant ownership of their contribution to another legal entity. We are in the realm of copyright law and “derivation”.

Formal/Informal or Standard/Specification

I have a hunch that applying any of these labels or trying to define them is liable to cause or contribute to more confusion or argument than it is worth.

Few would argue for privilege over merit in general terms and the idea of “meritocracy” is close to the heart of many in the Open Source Software (OSS) community. How far can the ideal of meritocracy be realised? Are attempts to implement meritocratic principles in the development of open standards (using “standards” to include virtually any documented set of technical conventions) visionary or beset my mirages?

What follows is a first pass at answering that rather rhetorical question. I have avoided links as I’m not trying to point fingers (and you would be wrong in thinking there are any between-the-lines references to organisations or individuals).

A meritocracy requires both a dimension of governance and a dimension of value. The latter, “value”, incorporates both the idea that something should be measurable and that there is consensus over desirable measure and its association with positive outcomes of the endeavour. In the absence of the measurable quantity that could be applied in a bureaucratic way we have a hegemony or a club. The Bullingdon Club is not a meritocracy. I suggest the following questions should be asked when considering implementing a meritocracy:

1. Have we recognised that a meritocracy must be situated in a context? There must be some endeavour that the system of merit is supporting and the suitability of a meritocratic system can only be judged in that context. There is no universal method.
2. Do we understand what success looks like for the endeavour? What are the positive outcomes?
3. Is there a human behaviour or achievement that can be associated with realising the positive outcomes?
4. Are there measures that can be associated with these behaviours or achievements?
5. Can this/these human endeavours be dispassionately evaluated using the measures?

Clear and coherent answers can be provided to these questions for OSS endeavours focussed on fixing bugs, improving robustness, improving performance etc. The answers become rather more vague or contentious if we start to include decisions on feature-sets, architecture or user interface design. Many successful OSS efforts rely on a different approach, for example the benevolent dictator, alongside some form of meritocracy.

So: what of “meritocracy in open standards”? Posing the five questions (above), I suggest:

1. The context is open standards development. There are differing interpretations of “open”, generally revolving around whether it is only the products that are available for use without impediment or whether participation is also “open”. It only makes sense to consider a meritocracy in the latter case so we seem to have a recognisable context. NB: the argument as to whether open process is desirable is a different one to how you might govern such a process and is not addressed here
2. Success of the open standards endeavour is shown by sustained adoption and use. Some people may be motivated to participate in the process by ideas of public good, commercial strategy etc but realising these benefits are success factors for their participation and not of the endeavour per se. I would like to place questions of morality alongside these concerns and outside consideration of the instrument: open standards development.
3. This is where we start running in sand inside an hourglass. Anecdotes are many but simple relationships hard to find. Some thoughtfully constructed research could help but it seems likely that there are too many interacting agents and too many exogenous factors, e.g. global finance, to condense out “simple rules”. At this point we realise that the context should be scoped more clearly: not all areas of application of open standards have the same dynamics, for example: wireless networking and information systems for education.  Previous success as a contributor to open standards may be a reasonable indicator but I think we need to look more to demonstration of steers-man skills. The steers-man (or woman) of a sail-driven vessel must consider many factors – currents, wind, wave, draught, sea-floor, etc – when steering the vessel. Similarly, in open standards development we also have many factors influencing the outcome in our complex system: technical trends, supplier attitudes (diverse), attitudes of educational institutions, government policy change, trends in end-user behaviour…
4. Not really. We could look to measures of approval by actors in the “complex system” but that is not a meritocratic approach although it might be a viable alternative.
5. Not at all. Having stumbled at hurdle 4 we fall.

It looks like meritocracy is more mirage than vision and that we should probably avoid making claims about a brave new world of meritocratic open standards development. Some anti-patterns:  “Anyone can play” is not a meritocracy; it depends on who you know, its not a meritocracy. The latter, cronysim, is a dangerous conceit.

There are many useful methods of governance that are not meritocratic; i.e. methods that would satisfy an “act utilitarian”. I suggest we put merit to one side for now or look for a substantially more limited context.