A Little History of Ontology from an OG
A critical assessment of Kurt Cagle's write up on best practices about ontologists
Kurt Cagleβs posts a reference on Linkedin to his scorching Substack article:
This will likely piss off more than a few people, but I've become increasingly convinced that there is a growing divide between the Linked Data era and where we are today, enough that we should probably be challenging a great number of best practices and methodologies as either no longer being relevant or in need of significant revision. Thoughts in ... The Ontologist.
Itβs clear Cagle put some good effort into this article. I donβt think anyone would be annoyed by his challenge. Iβve had similar perspectives, but he puts it in words that need quick consideration.
Cagle's article is a valuable contribution to the ongoing dialogue about the future of knowledge graphs and the Semantic Web. His critical examination of outdated practices provides a strong foundation for building more resilient, flexible, and ultimately, more useful information systems. For anyone involved in designing and managing structured information, particularly with a focus on effective classification, his insights offer a clear path forward, emphasizing consistency, contextual relevance, and an openness to new tools and methodologies.
Cagle accurately points out that the initial vision of Linked Data, while revolutionary in its conceptualization, has struggled to achieve its full potential in practice. The "galaxy of stars" analogy for Linked Data, he contends, is misleading. This observation is particularly pertinent for those focused on creating practical and navigable information environments. The reality of a few heavily referenced knowledge graphs and a vast "long-tail" of inaccessible or obsolete ones directly impacts the discoverability and utility of information. For any system relying on the coherent organization and retrieval of data, a fragmented and unreliable network is a significant impediment.
His critique of the static nature of many early knowledge graphs and the primitive state of their ontologies hits home for information architects. Effective classification hinges on a well-defined and consistent structure. If the underlying ontology is ill-equipped for querying beyond "trial and error," then the ability to categorize, retrieve, and make sense of information is severely hampered, leading to an environment where data exists but is largely unfindable or unusable.