Future of FLOSSE: Interview with Stephen Downes – Part 1

"On controlled metadata taxonomies: My copy of Schopenhauer is at one point of time a resource used in my philosophy class and sometimes a thing that works as my door opener. Both are perfect examples of use of Schopenhauer. How do I classify that, is it a book or is it furniture?"

Listen 1st part (MP3) – 29min – 13Mb

The first interview I conducted for the future of FLOSS in education was with no other than Stephen Downes himself.  It was my first experiment with Skype recording. The connection had some problems and recording peaked at times, so I’m sorry for lesser quality of the sound but the content is excellent.

There was a lot of good stuff so I decided to cut the interview into two separate pieces, 30min each. This is the first part, the second part will be available later (it’s even better than the first).

Stephen Downes works for the NRC (National Research Council of Canada) in Moncton, New Brunswick. He is part of the e-Learning research group which is attached to the Atlantic initiative of Institute for Information Technology.

As Senior Research Officer he is given a fairly free hand to pursue a research agenda in the field of e-Learning and is the author of OLDaily, a newsletter of issues related to educational technology. Previously he specialized in learning objects and in what manner
learning objects are organized, arranged, syndicated, distributed and displayed to
potential learners typically using a LMS (Learning Management System) and the techology
underlying that.

I asked him about learning objects and DLORN (Distributed Learning Object Repository Network). It’s a good start to be the basis of an open content learning object repository which could support shared construction of learning objects.

In comparison to that federated searches and controlled metadata are discussed. He is very critical about closed learning object repositories that are not visible to Google or any other search engine. He brings up the scalability problem of federated searches and the problems assosiated with describing learning objects.

As an alternative method he suggests third party metadata, which generally allows people to describe how a certain object is used instead of what it is. This kind of third party review lacks in federated searches.

Then we discuss the future of his Edu_RSS service. Downes brings up a very interesting idea of combining content production with social networking. This is achieved from the use of FOAF (Fried-of-a-Friend) in association with personalized and shared resource feeds. People could see what their friends are reading and writing. This could potentially help people to find others interested in the same topic in the field of education to work with.

Some questions asked in the interview:

• Who are you?
• Learning objects?
• DLORN?
• Are we going to see distributed learning object repositories?
• Metadata?
• Folksonomies?
• Federated search?
• Is RSS a potential base for the semantic web?
• Edu_RSS?

The second part of the interview is here.

"If you sit down to get started and look at tagging a million objects, you’re looking at a major investment. Only large companies can afford to do this but most of the world is not made of large companies"

"Read more" to see the extracted future events and analysis.

Future events

Here is a list of
fictional future events extracted from the interview with Stephen Downes.
If you want to comment or have additional future events to present
based on the interview, please do so.

Disclaimer: The
future events were constructed from the ideas presented in the
interview and do not represent the ideas of the interviewee. No crystal
ball or time machines were used in the construction of these events.
Bear in mind, it’s the future and everything is possible.

Year 2005

Metadata based search still not useful

The LOM (Learning Object Metadata) exists in order to make learning
objects discoverable so that people can execute a search. Yet this
still doesn’t work in practice as institutions don’t have the resources
to describe every single object they create. The keyword and title
fields are the ones people fill the most. The other properties provided
by metadata searches are next to unuseful.

Year 2006

Amateur audio and video enters learning

Cheaper and more accessible tools for recording audio and video have
enabled teachers and students to produce learning results and teaching
materials with new kind of low-end tools they already possess. For
example mobile phones include cheap and low quality but functional
features for sharing their experiences and ideas in audio and video
form.

Folksonomies used as a basis for new ways of content classification

Researchers approach the problem of controlled vocabularities through
folksonomies, where people can describe any content with words and
keywords that come to their mind. Researchers use a large amount of
users to conduct these experiments. Based on their free-form organic
tagging behaviour they try to come up with insight on how to build
better information classification systems.

Year 2007

Learning objects more like software

Previously learning objects were mainly static combinations of text,
images and illustrations. This was because of the lack of
sophistication in various authoring tools. As authoring tools gained
more functionality, learning objects started to look more like
interactive software pieces.

Learning object repositories based on crawlers appear

In spirit of Google and Technorati, new LO repositories appear. These
new repositories are less controlled and provide a different business
model than their predecessors. Their functionality is the same of the
web, based on bots that crawl various resources for learning objects
and automatically extract metadata out of them. A teacher simply just
releases the LO on a website and bots will aggregate the content. A
second layer is built on the search engine to support reviewing and
third party metadata.

Teachers protest against controlled metadata

When people search, they want metadata results from multiple sources,
not from a single provider. Yet user contributed commentary and review
capabilities of commercial learning object repositories are lacking.
Providers refuse to change the policy because it’s against their
business plan to control the search and the metadata. Extended metadata
capabilities are provided only to paying third party providers.

Content tagging businesses appear

Anyone who has tried to encode learning object metadata knows what a
pain it is. This is a major investment as only large companies and
institutions are able to afford tagging with metadata every object they
create. New businesses have appeared to serve these organizations with
professional tagging services. The cheapest providers are located in
india and other low-cost labor countries.

Year 2008

FOAF combined with search results

Services have appeared that combine Friend-of-a-Friend (FOAF)
information with content being searched. This approach is combining
social networking with content production. Developers have learned from
the previous failed attempts like Friendster that social networking
alone is not enough. These new kind of systems enable users to follow
what their friends read and write. Content based social networking
applications lower the bar to find other people to create open content
with.

Constructivism talked even more

Because of Wikipedia and other similar systems where knowledge is
co-created successfully, the educational philosophy of constructivism
has once again gained a lot of debate and discussion in education. Even
companies look at constructivist approaches to carry out on-job
learning, although they use different terms to describe it.

Communities build their own content repositories

Many online communities and courses have built their own searchable
resource repositories for very specific and focused areas of interest
based on the needs of their community. Software tools to create such
resources are available to everyone. The interesting thing is that
these repositories often carry out live information crawled from the
web. Managers/teachers pick and screen the sources of information they
want. Users/students are able to personalize the resulting feeds of
information to fit their needs.

Federated search faces scalability problems

The dominant model to setup a network of learning object repositories
has faced problems in growth. Only the most resourceful organizations
have been able to enter the network as each one of them has to be able
to carry out every single federated search on the network. The larger
and more popular the network has became, the more there are propagated
searches and bigger server requirements. Distributed open content
repositories based on more economical approaches have grown faster in
popularity.

Year 2009

Open content repositories surpass commercial repositories

Open content based repositories surpass commercial repositories in
popularity, contributions and offering. Commercial providers talk about
lack of quality but they are losing. Educators do not need pretty
looks, they need something that gets the job done. Open content is
easier to publish because of decentralized nature, availability of
certain authoring tools with CC built-in and the fact that federated
repositories require you to setup your own repository which takes a lot
of resources to build.

Structure of metadata still argued

Experts have not yet come into agreement of general purpose taxonomies
for tagging content resources with metadata. This is because the way
you describe an object and the way you classify an object, changes from
person to person and from time to time. Even in very focused
communities there is controversy of how to seamlessly tag the resources
they create. The most successful examples come from cases where the
rules are open ended.

Year 2010

Use considered as the most important property of a resource

Previously learning objects were described with metadata that described
what the learning object is. As the availability of content for various
purposes have increased, the information of what a resource contains is
less relevant to teachers. Teachers want to know how and for what
purpose a resource is useful. This is achieved through purposeful
combination of third party recommendations, comments and use scenarions
with objects.

Source of reliable news changes

When the commercial media started to quote Wikipedia in their articles,
the reliability issue was finally solved. Knowledge of Wikinews’s power
spread to the general public. Anywhere and anytime when something
relevant happens, people who are so into the case have a vast ammount
of time and combined effort to write the news. Enthusiasts and experts
do the job better together. Some media companies changed their business
plan to use independant reporters as a vechile.