OpenFiction [Blog]

The granularity paradox

Posted in Uncategorized by scarsonmsm on July 6, 2005

An observation about a technical aspect of sharing digital materials: A lot of the excitement around the sharing of digital materials in the past five years or so has revolved around the concept of the “learning object,” ideally a plug-and-play digital object that supports one or more learning objectives. The idea is that you can combine these objects together like stacking legos to create infinitely customizable learning experiences, based on learner need. Two key challenges in designing learning objects have been deciding on the right size and figuring out how learning object can be found.

Prior to joining MIT OCW, in the period of time I was creating the distance learning course that became tOFP, I got very interested in the issues of the right size (commonly called “granularity”) and the main way that people are trying to make learning objects searchable (adding information, called “metadata,” to an associated file that provides details about the object’s content). I had in fact created a database to be used for storing and organizing metadata. When I joined OCW, I thought that the project wasn’t doing enough to make the materials reusable or to “tag” them with metadata, but over time, I’ve come to see that the OCW approach has significant advantages over the learning object model in resolving what I describe as the “granularity paradox.”

OCW materials, as opposed to traditional learning objects, are larger (so bigger granularity) and less thoroughly tagged with metadata. By the LO model, this would suggest that they are less searchable and less repurposable than the standard learning object. The searchablilty issue would seem to be exacerbated by the poor search engine on MIT OCW. Yet, MIT OCW receives relatively few complaints about searchability of materials. So why is searchability not a big issue? It turns out that a large body of high granularity objects organized as classes are in and of themselves the metadata that guides users to materials of interest. With more granular objects, more metadata is required to allow users to access them. So the paradox, clearly stated, is “The smaller the granularity (which would seem to make objects more repurposable), the more metadata tagging is required.” This seems to be an almost exponential inverse–the smaller the objects get, the larger the body of metadata needed to make them locatable.

I’ve also observed that the more formatting applied to materials, the narrower the audience that can use the material. The formatting presupposes the end use and end users of the materials. No one format can account for a wide range of cultural and academic variations in localized learning environments around the world. Highly preformatted materials then become useful for a smaller and smaller slice of the global audience, the corollary “formatting paradox” that resolves the “granularity paradox.” Smaller, more highly formatted materials may not be more repurposable after all. Plug and play may not be the best model.

So on one end of the spectrum you have highly formatted learning objects which require large amounts of effort to format and tag–and hence fewer can be prepared–which can be used by a relatively smaller audience; on the other end, OCW materials which publish less formatted materials in greater volume, which require more effort for users, but allow better adaptation to local conditions, and are thus available to a wider audience.

2 Responses

Subscribe to comments with RSS.

  1. David said, on July 12, 2005 at 7:03 pm

    Steve – thought you might enjoy this article I wrote as a postdoc a while back. Seems to go along with your thinking here.

    http://rclt.usu.edu/whitepapers/paradox.html

  2. Stephen Carson said, on July 13, 2005 at 2:38 pm

    I am in awe of those who think through these things with 1.1, 1.2, 1.3 precision. Well stated that, in addition to metadata being a lot of work, it also mostly concerns “internal context,” as you put it, (which doesn’t amount to much for very small learning objects) rather than the external context that is provided in an OCW environment. Do you have a paper for every occasion?


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: