Making Metadata go away:
“Hiding everything but the benefits”
Erik Duval
Dept. Computerwetenschappen, Katholieke Universiteit Leuven, B-3001 Leuven, Belgium
Tel. +32-16-32.70.66
Fax. +32-16-32.79.96
EMail:erik.duval@cs.kuleuven.ac.be
Wayne Hodgins
Strategic Futurist, Autodesk Inc.
Tel. +1-415-507-5759
Fax. +1-707-773-1285
EMail:wayne.hodgins@autodesk.com
Abstract: In this paper, we argue that, in
order to facilitate the ubiquitous uptake of metadata,
and in order to realize their potential for advanced
flexible end user functionalities, the metadata should
become more transparent to the end user. We also
argue that we are technically capable of realizing this
goal, and illustrate the issues involved with some
practical examples.
Keywords: learning objects, learning
object metadata, metadata
1 Introduction
Having spent so much of our time and effort over the
last few years on the development of concepts,
standards, tools and infrastructures for metadata, the
authors are more than pleased with the steadily
increasing attention and focus on this topic, both in the
Research & Development (R&D) world as well as in
the commercial marketplace and not-for-profit sphere.
However, we believe that we are still very early on an
exponential growth curve and that many of the current
developments and efforts are somewhat misguided: in
our view, they often place too much of an emphasis on
the elusive quest for perfection and thus illustrate that
“the perfect is the enemy of the good”
1
. Perhaps even
more disconcerting is our concern that many of these
efforts are perfecting the irrelevant, as they are focused
on the direct and literal use of metadata, thus seeking
to continue historical and current practices, rather than
trying to design, experiment with and implement more
innovative and effective ones.
Moreover, many current developments do so at the
expense of end users, who are supposed to spend
considerable time and effort on the definition of
detailed metadata, using obscure terminologies and
unreadable, machine oriented syntaxes.

1
French proverb.
As the authors have been, and continue to be,
deeply involved in standardization activities around
Learning Object Metadata
2
, we want to point out
that standards are meant to enable developers to
realize interoperable technical components.
Standards are not meant to be visible to end users!
As an example, too many tools and
implementations of LOM and SCORM use the
exact same terminology as in the LOM standard
document. Terms like “catalogue entry”,
“contribute” and “semantic density” are fine and
appropriate for the standard document itself.
However, these terms are not likely to be terms that
are familiar and understood by many audiences and
communities who are creating and using such
metadata. It is understandable that early
implementations of new standards and
specifications focus on the implementation of the
functionality required. However, it is obvious that
evaluations of the actual experience of end users
with these tools will show the failure of this
approach.
Indeed, just like web browsers do not disclose the
hairy details of HTML or HTTP, usable tools
should not expose detailed Learning Object (or
other) Metadata. We should hide those details and
develop tools that do not unnecessarily burden or
complicate the life of the end user.
These observations are meant to assist us in moving
on to the next stage, rather than to be merely
critical or negative. Just as with the early days of
the Web, when the only (or at least most widely
used) tools available were text editors, and authors
provided the actual HTML tags (direct from the
HTML standard) to create HTLM documents, we

2
IEEE Learning Technology Standards Committee
(LTSC), Learning Objects Metadata (LOM)
http://ltsc.ieee.org/wg12

can see how the evolution of HTML tools has
proceeded to the point of making this largely
transparent to end users, content creators and readers
alike. In the usual clarity that hindsight provides, even
the very title of “HTML editor” exposes the risk of
“perfecting the irrelevant”. For most, there is no desire
or need for HTML directly, only for the benefits it very
much provides. Nor is it any coincidence that the
tipping point of the exponential rise in the use and
benefits of HTML matched this evolutionary path of
tools which masked the underlying complexity of the
HTML standard and let creators and consumers keep
their focus on the content. XML is now following a
similar route and the intent of the authors here is to
promote the most rapid migration to this next phase of
metadata implementation. The goal is to make the use
of metadata similarly easy and transparent. This is the
way to achieve widespread adoption of metadata and
reap the inherent and often unexpected benefits.
In this paper then, we try to address some
misconceptions and myths with regards to metadata, in
the hope that this can help to focus R&D on
approaches that will result in powerful flexible
enablers for end users. In this sense, the paper can be
regarded as a follow-up to [3].our previous research
agenda “Learning Objects Revisited”.
2 Basic Message
The basic message of this paper is that, in order to be
successful, metadata should become invisible for end
users. This is just a specific instance of the general
observation that technology achieves success when end
users become unaware of its presence [Norman, 1998].
Indeed, few people are aware of the “user interface” to
tools like cars, telephones, etc. Such tools do not
confront us with the complexities of the underlying
technologies and infrastructures. Consider the
example of the traditional wired telephone; the
complexity of what is required to make it work is
(now) largely invisible and transparent to the end user,
enabling the focus to be on the conversation, This is
only possible because groups of people spend long
hard hours working on such things as a common dial
tone, interoperability of telephone equipment,
agreements between the telephone various telephone
companies on how to use each others infrastructure,
how to seamlessly pass calls back and forth across
their systems, interoperable and common phone
number systems globally (well almost!) and sorting out
the immensely complex cross billing issues to create
simple single billing for customers. More recent
examples include the revolutionary transformation
from analog to digital telephony, which has all taken
place almost invisibly and without much disruption for
most. In contrast, cellular and wireless telephony has
been more of a step backwards as they are often far
too “visible”, difficult to use and get in the way of
the conversation, but we shall leave that for another
time and paper.
Getting back to metadata, our contention is that it is
still in its first stage and as such much too
“visible”, indeed too much in the foreground and
an end unto itself. This is limiting severely the
attainment of a critical mass or penetration of
metadata in the light of the enormous amounts of
digital content out there, despite the uptake and
adoption of metadata by many groups. Moreover, a
large volume of the metadata which does exist is in
the form of "file properties" which is often
proprietary or locked in systems such as LMS,
LCMS, ERP, CMS, etc. Most of the metadata that
reside in these systems remain unknown, and
unexploited. It is a relatively trivial exercise to
resolve this by exposing this resource description
metadata and usage metadata for easy harvesting.
The quiet early success of RSS (Rich Site
Summary) can be traced to exactly these traits of
making it all easy and transparent to the end user,
and enabling the harvesting of such summary
metadata.
3
The authors do not just want complain about this
situation or bemoan it. Rather, we want to
emphasize the need to continuously raise the
common awareness about the possibilities and
power which metadata enables, and to do so NOT
by putting a focus on metadata itself, but rather by
emphasizing what is possible when metadata is
ubiquitously present. This includes demand
creation for metadata, clear understanding of the
business case for metadata, lessons learned, and
recommended practice.
Moreover, even though metadata should remain
invisible for end users, we do believe there is a
serious need for large communities of practice;
professions, etc. to take on the responsibility of
adopting and adapting metadata for their use by
providing the critical elements of relevance and
applicability to their disciplines and constituents.
This includes the hard but necessary work of
defining vocabularies, taxonomies, ontologies,
which are relevant to their specific application,
profession or industry.

3
For a more detailed and descriptive write up
supporting and illustrating these points please
see the ASTD “Learning Circuits” article by
Stephen Downes at
www.learningcircuits.org/2004/jun2004/dow
nes.htm