The future of SCORM isn’t necessarily “SCORM”">The future of SCORM isn’t necessarily “SCORM”
There’s not been a lot going on with LETSI and SCORM 2.0 since our big workshop in October, but it’s not like there’s been nothing. There’s meetings and discussions happening weekly, mostly in the Architecture and the Business groups.
Recently, Avron Barr emailed many of us as the Sponsors’ Meeting approaches this month. LETSI very soon will need to get firm onhow to proceed we’ll proceed with the SCORM 2.0 effort. Avron writes…
“In addition to continuing the requirements gathering effort well into 2009 and possibly not calling what we are doing “SCORM” at all, we’ve also discussed the possibility of an open source software platform for e-learning systems and not just an updated reference model.
The actual open source effort might be a LETSI Working Group, a self-managed SourceForge community, or a Member Section in OASIS. We have a lot more planning to do. I’d like your thoughts about how best to explain why were are moving to an open source or “community source” model and what we might actually see coming out of this effort.”
I think he’s right on. Some of us talked about this very notion at the close of the workshop and going on since then. The way I’ve been thinking of it, LETSI is the organization that shepherds the process of gathering the requirements, shepherding the partnerships and architecture and holds the vision for what needs to happen, in terms of a platform for open-sourced organizational learning and performance improvement activities. Some other organization, even inexorably linked, handles the open-source development of the technologies and tools that implement what LETSI comes up with.
For what it’s worth, while a bunch of us at the workshop bought the domain name as a joke, we do own baqon.org… because SCORM is simply better with bacon. But I digress.
SCORM, per Dr. Eric Roberts (and Philip Dodds before him), solved two big problems back in the late 90s: vendors selling LMSs who were tearing a fledgling industry apart with proprietary implementations that made content and data impossible to transfer (thus driving costs of implementation sky high), and lower the costs for acquisition of learning technology for the government while helping to future-proof their capital investments in both content and the technology to get it to soldiers and employees.
Those problems still exist, and SCORM has worked with glaring efficiency to solve those problems, but what’s happened since then is that a global community has gone through a lot of the same struggles working with SCORM. There’s a lot we want to do with online learning (see those white papers and use cases). But SCORM can’t solve all these problems by itself — and by that I mean that the same vendors with an installed base of clients who’ve specialized in solving specific types of problems over the past ten years now aren’t the only people who need to be involved with how their LMS technology integrates with performance management, customer relationship management, talent management, business application systems, etc. LETSI has to pull in decision-makers, engineers and visionaries with a wider net.
Let me get specific so there’s no confusion: LETSI needs Moodle. LETSI needs Blackboard, too. LETSI needs WebEx. LETSI needs SAP and Oracle. LETSI needs SalesForce. LETSI needs a bunch of partners who aren’t traditionally in the learning space. All the ways in which educators, managers, owners and senior leaders want to see training and learning data mashed up with other types of data in their organizations — we need to work in concert… together. I’m almost positive it’s not just the learning professionals who want to mashup this data — we all need to really think about our APIs because we can imagnineer (if you will) for the problems we can visualize today — but the best thing we as technologists can do for the future problems we can’t imagine is provide a way for people to interoperably combine our data with other data.
My goal over the next week is to enumerate what I’m thinking in detail. My hope is that some conversation stirs up both here and on LETSI.org where we can get the best ideas out in the open.
Feel free to question or disagree with me. I can take the criticism.
an important discussion, indeed. also gargantuan! once you get past basic interoperability (which in the scheme of things is the easy part) the conversation becomes very fluid and can flow in many directions at once.
i also agree that the #1 thing in all of this is access to data. after all, what’s the LMS for? gathering and storing data about events. what is training? whether you’re into Learning 2.0 (geez I hate that phrase) or old-school brick and mortar, you’re talking about events where learning (hopefully) occurs. you’d like to be able to report on the outcomes of these events using the data you gathered during the event. this data helps with ROI, with qualitative evaluation of your course/ID methodology, with budgeting, with payroll, etc.
so to me it sounds like you really want to standardize the data and how it’s shared. this sounds almost like you don’t want to talk about SCORM 2.0 but rather about standardizing LMSs (or at least a big chunk of what an LMS offers). maybe you should call it the “LDM — Learning Data model”.
When you think about it, or at least when *I* think about it, the issue of a common data structure is the same deal that Adobe’s looking for with this latest batch of Captivate feedback (your post about connecting to a database, as an example).
Interoperability of the content in these different systems is, at the heart, a struggle to get these individual systems to interpret the data shared by content in the same way, and to relay it in-kind. I’m looking at how to report out from that collection in a way that can mesh with other data streams, but at the core they are branches of the same trunk.
I would love for others at LETSI to weigh in, but I’m inclined to posit that the issue driving interoperability and the ability to mash-up can be boiled down to the transfer and storage and interpretation of tracking or instructional data. The tracking I’m thinking of is the learner activity with content. The instructional data I’m thinking more of as sequencing and navigational instructions, though in that vein it could cover much more than sequencing of aggregated learning experiences.
You both seem to be on the right path to me. I’m cautious of big high level solutions. Consider the stated need for reporting. Rather than looking at a high level reporting service, we should *enable* reporting through a data model. The RTE data model seems to be the “core” of SCORM.
The JavaScript mechanism we have now in SCORM seems limited. The cross domain “problem” seems like an opportunity to adapt our communication mechanism to something slightly better that enables the mash-ups Aaron envisions. I’m not a Web service expert, but it would seem like we need Web service experts to weigh in. I know BBN produced a working Web service of the current SCORM data model in the Sample RTE. I’d like to see a rigorous discussion about just the RTE. Forget about the dependencies of CAM and S&N for a moment. How should the RTE be changed given what we know now?
Aaron, when you talk about a Linux-type community, I imagine focusing on one application that has never been clear — the Sample RTE. Is it a reference implementation? Is it just an example? Is it a purposefully poor example? I say rip that code apart and experiment with a new RTE to get data moving as effectively and efficiently as possible. Hopefully the designers have said what will make it effective. Hopefully the engineers will say what makes it most efficient as a low-level tool that others can use to surprise us.
Certainly sequencing and navigation would have to join the discussion soon after, but I hope it would join the discussion before content packaging or aggregation (whatever you want to call it). I think we should make as few assumptions as possible about the content (i.e., where it is, what it is, or how it was instantiated). This allows content that doesn’t fit the client-side model to play. Web 2.0 has taught us that we don’t need to hold onto the content.
–Damon
Damon,
First, thanks for chiming in.
When I’m talking about a Linux-type community, I’m looking for the lot of us to become an open-source community that supports the tools and technologies that enable, at a working baseline, the different components of what SCORM will become as a reference model, standard, specification, framework (insert your favorite “compendium” type word here). The RTE, for as much as you and maybe a lot of people don’t like it in its current state — the RTE is something most vendors taking a crack at building a SCORM Run-Time get right.
Sequencing, however, is not something that everyone gets right. Looking back, if ADL would have pushed out a web service (as an example — I’m not a web service guy) or some common-sourced engine in a handful of technologies (java, php, ASP.NET as merely examples), I’m pretty sure it would have catalyzed and accelerated the adoption of SCORM 2004. Maybe ADL wasn’t the right organization to do that, but there was never an open source community that could pull on ADL resources (the mind share) to do it, either.
LETSI might not be suited for putting out technologies either, but as I’ve written before, perhaps a sibling organization could be charged by LETSI for that specific purpose. In the current state, that organization could prototype and act as a lab for the specs and standards LETSI is looking to combine. But such an organization might pull in a lot more mindshare from a broad, international community. The problems we’re looking to solve are “global” challenges for working with distributed learning applications, their tools, data sets and technologies, etc.
One might discount the fact that the Sample RTE, even for SCORM 2004 was a first generation product. If you could see the difference between the current SCORM 2004 RTE compared to the original version, it’s hardly the same. That said, ADL has a responsibility to maintain it. With shrinking budgets, and less people to devote to the task, what more would one expect?
So, in answer to your comment, we should totally look at the Sample RTE — there’s at least one proprietary (but free to use) alternative to the test suite (testtrack.scorm.com). There’s no reason why a group couldn’t build a better sample run-time.
A healthy, committed open source effort will always be able to produce a better product than a government project. It’s not a knock on ADL or an governmental institution — it’s a numbers game. There’s probably nine or ten people total who work (or have worked) on the Sample RTE. Compare that to 20-30 developers a small, committed open source project could yield — especially with the advantage that they already have some idea what an Sample RTE needs to look like. And unlike a government effort, since it’s all open-source from the get-go, there’s really no budgetary constraint on collaboration.
One thing to consider when advocating open source implementations as a solution is that many commercial software vendors are very reluctant to include open source software in their products. Licensing issues aside, lawyers say very scary things about the possibility of an open source contributor introducing (accidentally or maliciously) protected IP that s/he doesn’t have the right to distribute into the open source code base. This possibility makes many commercial software vendors very reluctant to redistribute open source components as part of their products. To that end, open/community source software isn’t a solution for an interoperability problem, however, as you state, it could be a beneficial enhancement to a well defined specification.
Great observation, Mike. I agree open source code helps make a better reference implementation as I see it. It also helps encourage experimentation. All your points about commercial concerns are valid and not conflicting in my mind.
Aaron, I’m not cracking on the RTE. I recognize its tremendous value. Any crack you read on my part was targeted at the lack of clarity in the role of the RTE. As Mike says, I want the RTE to make the spec more clear. Depending on what it’s role is influences the degree to which it helps the spec. However, this isn’t really my point, which I’m afraid was not made clear.
My real point is in agreement with you, Aaron, to focus on the data. I refer to the RTE (not the SRTE) as a concept for data exchange — the technical book. I’m saying we should focus on that book first using the SRTE as code to experiment with.
I had some thoughts today on sequencing after reading a paper by Richard Van Eck. I plan to blog about that tonight (hoping I find time).
I think we are all in agreement about the need for a better spec and the role that code can play in making that happen. When it comes to redefining the substance of the spec, I say start with the RTE data communication concept. I’m tempting to try and become a web service expert, but will more likely just seek out someone I can trust. –Damon
I’d like to think that’s changing to some degree, Mike. Proprietary authors can also introduce, accidentally or maliciously, protected IP that they don’t have the right to distribute. I also predict, but this is only opinion and conjecture at best, that the attitudes from organizaitons about implementing open source implementations is going to change as the conditions on the ground change:
* Companies find economic incentives to implement open technologies and spend their money on customization and integration by hiring specialists, much like they have to with even proprietary software.
* An older generation of IT and C-level leaders starts to turn over and more Gen X/Gen Y decision makers emerge with a different view of risk tolerance.
I don’t know that I’m consciously arguing that a community effort is a solution for interoperability, so much as I’m saying that a community effort is a way to get the transparency necessary to hammer out these different perspectives on how something’s supposed to operate.
CC supposedly solves a lot of interoperability problems with systems that implement it. But I wouldn’t know, because I’m not a member of IMS and conversely I’m not an adopter of any of the systems or content that make use of it. Being somewhat privvy to a little of the politcs of working with more closed standards groups, I’m always open to collaborating with groups like IMS however it happens — I mean, we’re all trying to do the same things, we should be more productive with our time and talents — but, I’m not exactly looking forward to working with IMS.
You have a lot of proprietary code, Mike. I LOVE working with your products. Our personal relationship notwithstanding, I have repeatedly said on this very blog that your products and counsel are clearly best of breed on the available market, and I work with a lot of systems and code.
Would you have developed something better if it was opened up? Arguably not. But could I advocate for your knowledge and services had I no exposure to you or Tim? Also, arguably not.
I may be in the vocal minority, but it’s really hard to trust when there’s no transparency. You can’t have advocacy without access. How does a group of like-minded people come together to give access and lend transparency without the effort being common-sourced? I’m open (no pun intended) to the alternatives.
yes, i agree wholeheartedly. with transparency, even if you don’t agree about something, you can at least understand the decisions being made. context is very important for nurturing understanding and communication.
and i could live without IMS. their standards development process is too closed, their specs are too bloated, and they give me the impression they could really give a rat’s ass about the little guys like me (take a look at their membership signup page to see what i mean).
I’m currently writing a post about the futur of SCORM (in French) and was reading again those posts to trying to summarize the zeitgeist.
So if I understand, you advocate for a sort of framework, like RoR[1] or django[2], or even GAE[3], but specialized for elearning ? This framework should provide tools for building mashups of webservices (storages of content on CDN, chats, virtual worlds, etc.), displaying thoses mashup to learners and finally collecting learning progressions ?
This seems to me a really good idea. Like the Cometd project[4], it would be available in many languages on the server side, so each developer would be able to use its prefered environment.
For the license problem, as far as I know (but I’m not a specialist) the MIT license used by RoR, django or symfony [5] seems a good solution as it allows to build closed or open source software on top of those frameworks.
[1] http://www.rubyonrails.com/
[2] http://www.djangoproject.com/
[3] http://code.google.com/intl/en/appengine/
[4] http://code.google.com/p/cometd/
[5] http://www.symfony-project.org/