A Learning Object Manifesto
26 January 2006
Editor: Erik Duval
Acknowledgement: In-depth, detailed comments on earlier versions of this manifesto were provided by:
- Vana Kamtsiou, NCSR Demokritos/ Division of Applied Technologies, Athens, Greece
- Prof. Stefano Ceri, Dipartimento di Elettronica e Informazione, Politecnico di Milano, Italy
- Dr. Peter J. Scott, Centre for New Media, Knowledge Media Institute, Open University, UK
- Wayne Hodgins, Autodesk Inc. & Learnativity, USA
We gratefully acknowledge the support from the ProLearn Network of Excellence (http://www.prolearn-project.org/).
Reactions: Your feedback, comments, statements of support or disagreement, etc. are explicitly solicited and much appreciated. Please send them by email to Erik Duval .
Creative Commons License
A note on style
This manifesto is deliberately written as a sequence of short, numbered chunks, with many footnotes and references. The intent is
- to make the claims and statements as clear as possible,
- to avoid cluttering the main line of the argument with details while still being able to include such details, and
- to facilitate structured discussion of the manifesto.
- Many resources have been spent for decades 
[2-4] on research and development in the area of learning technologies 
- Yet, besides countless publications and a mass of PhD’s, there is a remarkable lack of actual impact of technology on the way people learn, in schools, academic, corporate and military contexts alike
- The purpose of this manifesto is to help focus research on relevant questions and problems 
, so that the results of this research will actually have an impact in the Real World.
- We challenge our readers to develop a clear and testable mission statement for learning technologies. Examples from history include:
- "I believe that this nation should commit itself to achieving the goal, before the decade is out, of landing a man on the Moon and returning him safely to Earth". 
- “The ANSARI X PRIZE will be awarded to the team that designs the first private spaceship that successfully launches three humans to a sub-orbital altitude of 100 km on two consecutive flights within two weeks. All teams must be privately financed.” 
- Turing test for machine intelligence: “a human judge engages in a natural language conversation with two other parties, one a human and the other a machine; if the judge cannot reliably tell which is which, then the machine is said to pass the test” 
- The Human Genome Project (HGP), an international, collaborative research program to map and understand all the genes of human beings.
- The DARPA Grand Challenge on autonomous vehicles, that was awared to Stanford in 2005.
The very fact that we struggle to come up with an equivalent mission statement, that is as clear and simple to test and verify, that enables the same sort of continuous progress measuring, and, most importantly, that is as effective in rallying public and private support, is a sign of the lack of clarity in vision and purpose of the field.
- ↑ Early pioneers include Alan Kay, Doug Engelbart, Andy Van Dam and others…
- ↑ Quantitative data are appreciated: please send info on euros spent on “technology enhanced learning” in the different EC framework programs, estimates of global investments in this area, etc.
- ↑ One could probably argue that the impact has been least in the school context, and minimal in a university setting (with the exception of maybe a few “open” universities). Corporate training, especially so-called “on the job training”, has been using ICT to some advantage, and military training has taken most advantage of the affordances technology provides, especially through high end simulations.
- ↑ In our view, relevant research focuses on problems that are
- not too easy: otherwise, commercial or open source development of tools and methodologies, as well as straightforward practice, will solve the problems without requiring any input from R&D;
- not too hard: otherwise, progress will be so slow it becomes impossible to measure and no corrective feedback loops can be introduced to ensure that the research stays on track.
- ↑ President John F. Kennedy (1961)
- ↑ http://www.xprize.org/about/fact_sheet.php
- ↑ http://en.wikipedia.org/wiki/Turing_test
There is historical precedence for using manifestos to help steer R&D in relevant directions.
- In the field of database research, there is a series of manifestos that have tried to help focus research on object oriented , "third generation"  and active  databases.
- Hypertext: an early paper that was quite influential in steering hypermedia research was based on pioneering work on the Notecards system . A more recent example tries to formulate a research agenda for hypermedia on the semantic web .
- Also in the field of (e-)learning, some attempts have been made at formulating clear goals and principles. However, these seem to have had little impact [1, 10, 12, 13, 15].
- LOM research agenda: Specifically in the area of learning objects and metadata, a research agenda has been formulated about two years ago . Some research papers explicitly refer to this agenda and this manifesto builds on the ideas that have since proven to be the more relevant.
Learning as a root solution
- Learning is fundamental to improving the way we deal with the Grand Challenges of our times (aids, hunger, war, …): if we can improve the way that we learn, then we can effectively get better (at getting better) at tackling problems.
- Learning is also fundamental for personal growth and realizing one’s full potential.
- Despite its fundamental nature and the considerable investments in formal education, corporate training, lifelong learning, etc., current learning support is not very advanced.*
(* We recognize the considerable effort on the part of many teachers, parents and learners. Yet much of this effort is often neutralized (or worse) by the organizational context that often mistakes means for the end… See also Richard Dawkins on what makes a great schoolmaster.)
Learning is personal(ized)
- Learning happens in the head of the learner: it is not an effect that can be created by a third party without the active involvement of the learner.
- Consequently, the learner has a right and an obligation to be co-responsible of his learning:
- this implies the right for the learner to "refuse" what is proposed by either a human teacher or an automated agent,
- this also implies that the learner cannot abdicate responsibility for his learning.
Learning Objects are fluid and dynamic
- Although there has been substantial discussion about the definition of learning objects and, more specifically, about the notion of granularity, there is a deep rooted tendency to equate learning objects with static documents.
- We believe that a more appropriate approach is to consider different content models or content ontologies that range from small "raw media" objects to very large "curricular" objects, with several intermediate classes of objects. After a very early proposal by learnativity, now more than a decade old, there has been some recent progress in this area, relying on a set of ontologies for learning objects of different granularity .
- More fluid and dynamic approaches are relevant beyond the strict learning context: content management in general has been focusing on more flexible notions of reuse and repurposing.
- There is too much focus on text only content: we should move beyond text and improve the ways that we can deal with audio, video, animation, simulation, games, role playing, etc.
- Learning objects do not deal with content only: learning strategies or activities can also be captured in learning objects, and thus be made available for reuse or repurposing.
- However the learning object is represented, stored or transmitted, it is important to separate:
- style & presentation
- structure & navigation
- Learning objects can be automatically dis-assembled and re-assembled:
- "packages" of content can be decomposed, the components can be described, stored and managed individually;
- requests can be answered by dynamically generated learning objects.
- Learning objects contain content fragments and structure.
- In this context, structure does not need to be "hard wired" static interrelations. Structure also includes dynamic choreography of content constellations with late binding of the content fragments. In advanced simulations that integrate visualizations, models, rendering engines, etc. from different sources, structure covers all the machinery that determines what is connected to what.
- There are two broad approaches to specifying structure of learning objects:
- In a document approach, the structure is defined according to some model. Nowadays, this approach typically leads to structure expressed in XML.
- In a black box approach, the structure is captured in the interactions between learning components. Typical technologies deployed for this approach include web services and JavaBeans.
- Separate from both content and structure are style and presentation.
Electronic forms must die
- Technology in learning is adopted in a way well understood from consumer electronics: that means that, in order to "cross the chasm" from early adopters to early majority, it is now time to focus on "packaging" the technology in ways that make sense for end users (learners, teachers, trainers, …) rather than developers.
- Specifically, with respect to metadata, we must hide the complexities of the underlying specifications, and only reveal the added value that metadata can enable. In particular, this means that:
- generation of metadata must be and can be automated : manual metadata should only be added if the contributor is specifically motivated to do so, for instance by submitting a review in a blog-like way;
- queries into learning object repositories should be enriched with metadata about the user and his context (course, work, …) in order to maximize precision and recall of the result set.
- Because we need to move technology to the early majority stage, we should adopt the approach of successful end user oriented services provided by the likes of Tivo, amazon, google and ebay …
There is more to metadata than static creation
- There is a strong tendency to think of metadata as something that is created around "publication" time of a learning object, in order to be stored with the object in a repository, never to be changed and occasionally to be used to determine whether or not the learning object it described is relevant. This tendency is misguided, as we can now deal with much more fluid and dynamic metadata than we could in the days of the library card.
- We expect the vast majority of metadata to be created during the "active lifetime" of an object. Indeed, we encourage adding to the metadata usage metadata,
- every time an object is included in a list of search results,
- every time the detailed description of a learning object is consulted,
- every time a learning object is downloaded from a repository
- every time a learner "uses" the learning object (this may include several types of learner-object interaction!),
- every time a teacher or trainer selects a learning object for inclusion in a "course", etc.
- We believe that especially the usage metadata mentioned above should feed back into the learning object lifecycle:
- Authors of content should be able to learn from the "tracking data" that metadata include;
- Researchers can mine the patterns of use reflected in metadata, in order to better understand how people make use of learning objects and metadata;
- Metadata producers (both human as well as software) can be improved, based on the experiences recorded in the metadata.
- It is important to emphasize the importance of metadata about context: searches for relevant content can be enhanced with metadata about the user and his context – this information is readily available when the user is working in the context of a Learning Management System, where her identity, profile (including information about for instance appropriate languages), the course context, etc. are explicitly maintained – see also 7b.
- It is also important to emphasize that the subjective nature of many metadata is "a feature, not a bug": the subjective appreciation of the usefulness of a learning object by a colleague or peer is probably more helpful than the objective, precise spelling of an author name.
Novel access paradigms
- Another way to make electronic forms die (see 7) is to rely on alternative means for making appropriate material available to interested parties. An important such means is the use of social filtering or recommending techniques.
- In the same way as Amazon recommends potentially interesting material to prospective buyers, a Learning Object Repository can recommend learning objects once a threshold of similar users has expressed interest in such object – by downloading it, including it in a course, learning with it, etc.
- Teachers or trainers can be alerted when new material becomes available that is relevant to their needs: some learning object repositories already offer RSS feeds. However, this facility should be made more flexible:
- professors could be informed once a critical threshold of colleagues start using a particular object in a course similar to the one they are teaching;
- learners could forward relevant learning objects to colleagues or peers that share similar needs or interests with them;
- learners and teachers alike could be informed when a learning object they use has been updated;
- Another alternative means for making material available relies on information visualization techniques to give end users an overview of available material, and tools to zoom in on more relevant sections of the collection, down to the individual learning object. This approach is especially useful to give users an overview of the contents of a collection that they are not familiar with.
Metadata is just metadata
- The field of metadata is rather fragmented at the moment, with initiatives such as Dublin Core (DC), Learning Object Metadata (LOM), the Moving Picture Experts Group (MPEG-7), the Resource Description Framework (RDF) and others.
- Such fragmentation can be intimidating to prospective developers and thus hinders adoption and uptake of the very notion of metadata.
- More importantly, this situation leads to "balkanization" of metadata, where tools typically work with one kind of metadata only, so that end users either need to work with parallel sets of tools, or will have only a portion of the global set of metadata available to them.
- This situation is all the more regrettable as the world of "learning content" is not clearly separated from content in general (targeted by DC) or audio-visual content (the focus of MPEG-7).
- As an intermediate solution, we should develop a better understanding of the ways in which application profiles can allow us to "mix and match" metadata elements from diverse origins. Especially when the architectural assumptions are not evidently compatible, such mixing of elements does not seem to be straightforward to implement.
Standards and Interoperability
- We believe that it is essential that we develop an open infrastructure for learning. In this context, "open" means that the standards, as basic building blocks to enable interoperability between components:
- Are available to the public: this does not necessarily imply that the documents should be available at no cost whatsoever, though the cost should not make it impossible for some to access the standards document;
- Can be used, implemented, referenced, etc. without any restriction: this does imply that there should be not patent or other use restrictions for developers or end users.
- We believe that the standards should enable rather than prescribe ways to put components together and to realize a meaningful service to the end user. As an example, most metadata standards do not impose mandatory status of any metadata elements: that decision is left to tool developers, communities of practice or the end user. Rather, they enable the capturing metadata values in an interoperable way: prescribing which metadata values should be captured is not part of their mission.
- It is not trivial to determine what is a sufficient set of simple enough standards: however, we believe that, for the current generation of learning tools, the basic standards and specifications are in place. Some specifications still need to progress to finalized accredited standards, but the focus should now shift to putting the standards at work in terms of relevant applications for end users that hide the technical complexities.*
(* In this sense, the current state of learning technologies is not unlike the state of the Web around 1994 when the basic web standards (HTML, HTTP, URL) were stable and focus shifted to the development of browsers and authoring tools. Even though additional development of XML, web services and other technologies is not without merit, the breakthrough of the Web is more directly caused by the development of tools like Amazon, Google or eBay. We now need to devote time and energy to the development of equivalents for learning.)
By signing up to this research, you can explicitly enlist your support for the views expressed above. We expect such support to be matched by your actions in research, development, deployment or use of technologies for learning.
1. A Manifesto for Change, 
2. Engelbart, D. and English, W. A Research Center for Augmenting Human Intellect IFJC: Fall Joint Computer Conference, Arlington, VA, USA, 1968.
3. Kay, A., Computer scientists\’ responsibility to educational research: A dynamic medium for creative thought in Fourth SIGCSE technical symposium on Computer science education, ACM Press, 46, 1974, 
4. van Dam, A. The Electronic Classroom: Workstations for Teaching. International Journal of Man-Machine Studies, 21 (4). 353-363, 1984,
5. Halasz, F.G. Reflections on Notecards: Seven issues for the next generation of hypermedia systems. Communications of the ACM, 31 (7). 836-852, 1988, 
6. Stonebraker, M., Rowe, L.A., Lindsay, B., Gray, J., Carey, M., Brodie, M., Bernstein, P. and Beech, D. Third-generation database system manifesto. Comput. Stand. Interfaces, 13 (1-3). 41-54, 1991, 
7. Atkinson, M., DeWitt, D., Maier, D., Bancilhon, F., Dittrich, K. and Zdonik, S. The object-oriented database system manifesto. in Building an object-oriented database system: the story of 02, 1992, 1-20, 1992, 
8. Dittrich, K., Gatziu, S. and Geppert, A., The Active Database Management System Manifesto: A Rulebase of ADBMS Features. in Second International Workshop on Rules in Database Systems, 3-20, 1995,
9. van Ossenbruggen, J., Hardman, L. and Rutledge, L. Hypermedia and the Semantic Web: A Research Agenda. Journal of Digital Information, 3 (1), 2002, 
10. Grand Research Challenges in Information Systems. 35, 2003, 
11. Duval, E. and Hodgins, W., A LOM research agenda. in WWW2003: Twelfth International World Wide Web Conference, (Budapest, Hungary), ACM Press, 659-667, 2003, 
12. Jensen, H.S. Open-Ended Manifesto on Research and Learning. Learning Lab Denmark Quarterly (3), 2003, 
13. Picard, R.W., Papert, S., Bender, W., Blumberg, B., Breazeal, C., Cavallo, D., Machover, T., Resnick, M., Roy, D. and Strohecker, C. Affective learning - a manifesto. BT Technology Journal, 22 (4). 253-269, 2004, 
14. Rehak, D., Good & plenty, googlezon, your grandmother and nike: Challenges for ubiquitous learning & learning technology. in PGL Workshop on E-Learning Objects and Systems, (Orlando, FL, USA), 2004, ,
15. Taylor, J., Rodden, T., Anderson, A., Sharples, M., Luckin, R., Conole, G. and Siraj-Blatchford, J. An e-learning Research Agenda. 2004, ,
16. Zemsky, R. and Massy, W.F. Thwarted Innovation. What Happened to e-learning and Why. Learning Alliance Report, 76, 2004, ,
17. Cardinaels, K., Meire, M. and Duval, E., Automating metadata generation: the simple indexing interface. in WWW2005: 14th international conference on World Wide Web, (Chiba, Japan), ACM Press, 548-556, 2005, , 
18. Klerkx, J., Meire, M., Ternier, S., Verbert, K. and Duval, E., Information Visualisation: Towards an Extensible Framework for Accessing Learning Object Repositories. in Kommers, P. and Richards, G. eds. EDMEDIA05: World Conference on Educational Multimedia, Hypermedia and Telecommunications 2005, (Montreal, Canada), AACE, 4281-4287, 2005,
19. Verbert, K., Jovanovic, J., Duval, E., Gasevic, D. and Meire, M. Ontology-Based Learning Content Repurposing: The ALOCoM Framework. International Journal on E-Learning, 5 (1). 67-74, 2006,