Re: Proceeding with [Core] via email discussion (Was Re: Postponed beginning for [Education], agenda for [Core])

Dear jeffe,

First, my belated apologies for missing the call on 18 December. I'm afraid I became confused about the two groups - entirely through my own fault.

Second, I realise that I am joining an existing conversation, so it may not be particularly helpful if I introduce a completely new direction. However - here goes!

Having been involved in formal standardisation in the UK's BSI, ISO/IEC JTC1/SC36 and CEN, I welcome the emphasis on the developers' perspective - something that tended to be missing from these organisations.

However, what is being proposed with the Application Foundations strikes me as being a sort of developer's toolkit or library. While I do not doubt that many would find this useful, I am not sure:
1. why there needs to be a single such library - can't people compete in this area?
2. in what senses these foundations will be specific to education.

In my view, the critical enabler for ed-tech is interoperability. Again, much of the technical and transport layers in any solution to this issue will not be education-specific. The problem is in creating the structured data models that are at the same time 1) standardised and 2) flexible, to encourage innovation.

Much software development occurs in the context of what might be termed a binary topography, consisting of user and software. Educational software development occurs in a triangular topography, the corners of the triangle being software, teacher and learner. This fact alone makes the persistence and interoperability of data much more important.

Another way of making the same point is in my abstract model of how education progresses at http://edtechnowdotnet.files.wordpress.com/2013/11/slide-220.jpg. Educational apps will generally encapsulate some sort of learning activity, shown by the inner, light grey figure of eight loop. But to play a meaningful part in a wider educational process, the individual learning activity is embedded within a learning management cycle, represented by the dark grey loop, representing assignment, reporting and analytics. Unless the individual learning activity shares the data it requires at launch with the learning management system, and the data by which it reports learning outcomes then it becomes meaningless in the context of the programme of study. Launch parameters - and in some cases the sharing of runtime data - are important because of the huge variability of educational contexts (different students, different teacher preferences, different learning objectives). The reporting of learning outcomes is vital to the job of the teacher, and vital too if anyone is to make learning analytics fly.

In the early 2000s, there was a massive effort on formal metadata taxonomies to support search and discovery. The theory was that teachers had difficulty in finding the correct learning resource. In my view, this assumption was wrong - teachers had no trouble finding resources - their problem was in deploying them in the classroom. Similarly, it strikes me that the assumption of this group is that developers are having trouble developing good apps and again, I am not sure this is right. Development has never been easier.. It is the automatic plug-and-play integration that is the problem.

Another way of visualising this problem is by comparison with business software. Businesses tend to use very sophisticated and expensive software suites (e.g. Oracle, SAP) which command monopoly prices and either ensure good in-house integration, or rely on large budgets for consultants to effect bespoke integration. Education's requirement for interoperability and integration is every bit as acute as business', but they do not have the money to afford either the costs of closed, proprietary systems, or for bespoke integration work.

My proposals for priorities are therefore:

1. A schema description language that would allow developers to declare their data models in a way that allowed the sharing and piggy-backing off different schemas. The problem is not a technical one of sharing data - but making semantic sense of that data. So interoperable software needs to be able not only to read the schema but to understand which parts of that schema map to data that it already understands and can use.

2. Management of privacy is a prerequisite for good data sharing. What I think is needed in this area is not a set of rules (which will vary across jurisdictions) but a data handling description language - an ability to codify in machine-readable format the sort of processes and commitments that a parent might agree to in a human-readable consent form, or which might be encapsulated in a regulatory instrument. This will not only provide clarity and the ability to discuss difficult issues (e.g. around anonymisation of data) but will also allow software to help enforce such procedures (e.g. this data item cannot be sent to that person, or this sensitive data item has expired and needs to be deleted).

Finally, I should acknowledge that much ed-tech theory to date has been predicated on the "independent learner" going out onto the internet and finding things out and networking with peers. In such a theoretical model, interoperability is not so important as you are taking the teacher out of the equation and falling back on a standard, binary topography. Nor is there any requirement for longitudinal process management. But I think the theoretical justification for this approach was always deeply flawed and it is for this reason that the research shows that technology has to date had virtually no impact on raising the quality of educational outcomes.

For the same reason, I think that if W3C could provide a platform for interoperability and automatic integration of educational apps, it could have a completely transformative effect on the market for innovative ed-tech software.
That is my opening position - I hope it helps!

Kind regards,
Crispin


On 23 December 2014, Jeff Jaffe <jeff@w3.org> wrote:
> Today, I've also put a more elaborated discussion of these questions in our wiki [1] if people would like to comment there.
> 
> Jeff
> 
> [1] <https://www.w3.org/wiki/AB/2014-2015_Priorities/w3c_most_important#Core_task_force>
> 
> On 12/18/2014 10:17 AM, Jeff Jaffe wrote:
> 
> > On today's scheduled [Core] call there was disappointing attendance and insufficient critical mass to make progress.
> > 
> > We decided to progress via mailing list and wiki for now, until our first phone call a month from now (January 15th).
> > 
> > Here are some relevant points.
> > 
> > 1. The agenda for today's call was comprised of a set of questions below. I invite discussion of these questions on the mailing list as a way to move this forward.
> > 
> > 2. If there are other items that you feel we should explore as part of "what is most important for the core of the Web", please add your thoughts as well.
> > 
> > 3. Early next year, the Team will try to address some of these questions as well. That will provide further fodder for our discussion. Let me know if you have questions/comments for the Team.
> > 
> > Jeff
> > 
> > On 12/9/2014 9:39 PM, Jeff Jaffe wrote:
> > 
> > > 
> > > 
> > > Here is the agenda for the Core call on 18 December.
> > > 
> > > 1. Application Foundations - do we have the right categories?
> > > 
> > > 
> > > * Are there topics that are currently omitted that should be added?
> > > * For example, at the TPAC breakout, Daniel Glazmann proposed that we needed to focus on UI requirements as well.
> > >   
> > > * Is the taxonomy correct? Should the functions be re-factored differently (move things around between categories; subdivide categories)2. Developer input. Most of us work inside organizations that produce infrastructure - we don't have sufficient number of developers in the discussion. How do we get more developer input? What kind of roadshow might work? This would help us answer questions such as:
> > > 
> > > 
> > > * Are we providing developers what they need in each area. For example: are we providing everything needed to ensure secure apps? Are we providing everything that is needed to deal with responsive design? Are we making all the knobs needed available for performance? How should the task force start taking on these questions?3. Next steps:
> > > 
> > > 
> > > * Description of Application Foundations: Is the description in the blog post accurate? Is it at the right level of description? Does it need more detail? Less detail?
> > > * Mapping of existing work: Assemble a list of all existing work and make clear which category it fits in.I don't expect to reach closure on these items, but we should discuss how we make progress during the course of this effort.
> > > 
> > > The next meeting of Core is 15 January; then bi-weekly after that.
> > > 
> > > I will do some work on the wiki in the next few days to reflect our status.
> > > 
> > > Jeff
> > > 
> > > 
> > > 
> > > 
> > 
> > 
>