4 Responses to The Metadata is Bigger Than the Data

  1. Carl Lemp 2015/10/15 at 00:58 #

    Aren’t these the problems of all natural languages as well? Texts are written where existing words are used in new ways. New words are invented and gather meaning over time. Words fall into disuse and lose their meaning. It seems that the big failing of our meta-data systems is that they are not very adaptable to change.

    • Mark Baker 2015/10/20 at 08:18 #

      Thanks for the comment, Carl.

      I agree completely. Metadata is often used to fix something that by its nature cannot be fixed. The mutability and flexibility of language is essential to its function. Its usage has to mold to our differing needs and interests over time and between domains. Attempting to fix meanings in place does not add clarity, it creates brittleness.

      This does not mean that we cannot create and use metadata effectively, but it has to be done with respect for how language works, rather than an attempt to subjugate language to the database world.

      • Carl Lemp 2015/10/21 at 10:27 #

        Mark,
        Do you have any insights or suggestions into how metadata can be made adaptable to change while still providing some level of added meaning? Topic Maps looked like a good approach for doing that for a while but that technology never seemed to get accepted by the broader market. Are there techniques for using DITA metadata in an adaptable way?
        Thanks,
        Carl

        • Mark Baker 2015/10/22 at 10:42 #

          Carl,

          The first property of metadata is that it must be accurate. The first priority of most metadata systems it to make them universal. These two things are fundamentally at odds with each other. All language is local. Every domain attaches its own meanings to common words, as well as introducing specialized terms. Each domain make distinctions that other domains don’t make. The attempt to make metadata universal results in metadata that is not true to local definitions and local distinctions.

          Underlying this is a naive belief that all that prevents common understanding between domain is agreement on a taxonomy. This is not true and betrays a fundamental lack of understanding of how human communication works. (See: http://everypageispageone.com/2015/08/04/the-economy-of-language-or-why-we-argue-about-words/ on why communication does not work like this, and http://everypageispageone.com/2015/09/28/the-age-of-the-content-manager/ on how database-thinking has come to dominate the conversation about content, and the problems that come with it.)

          So how do you make metadata useful? Simple. You make it true. To make it true, you will have to make it local. The content of a domain requires the metadata of a domain.

          Can the metadata of a domain be used to connect and manage content between domain (and thus across enterprises)? Yes, as long as the semantics are compatible for the type of connection you are making. Even if the vocabulary is different, if the semantics are compatible, as simple mapping will allow you to make the connection.

          What if the semantics are not compatible? Then the connection can only be make by stories, not by metadata. Shoehorning the content into a universal metadata framework will not actually make the connections work, it will just make the CMS work. As far as understanding goes, it will not build the bridge that needs building, just obscure the need to build it.

          All language is local. All semantics are local. All useful metadata is local. Most of us belong to more than one locale. The bridge between locales is stories.