The Phantom Menace: Why Hierarchical Classification Became Obsolete
The Ghost of Classification Past: A Horror Story - How the Spectre of Automation Killed the Parent/Child Model #wooooh
The Filing Cabinet in the Cloud: Why Hierarchy is No Longer Fit for the Digital Age
We have a ghost in our machine. It’s the ghost of the filing cabinet, a rigid, hierarchical structure born from a world of paper and manila folders. Despite operating in a dynamic, interconnected digital world, many organisations still cling to this archaic model for classifying their information and data. We build complex digital systems and then try to force them into a parent-child taxonomy, a digital facsimile of a physical cabinet.
This isn't just inefficient; it's fundamentally incompatible with modern reality. The traditional archival models, the linear records lifecycle and the top-down hierarchical classification system, have become obsolete. To effectively manage, govern, and extract value from our digital assets, we must fully embrace a shift that began decades ago: a move to relational, continuum-based models. This means favouring the rich context of ontology over the rigid structure of taxonomy, and accepting the fluid, non-linear nature of information over the paper-paradigm’s lifecycle.
What’s fascinating is that this isn't a new revelation sparked by the internet or the AI boom. The most forward-thinking archival theorists saw this coming over half a century ago.
A Prophet in the Archives
While recently reading Dr. Karen Trivette's doctoral thesis, "DESIGNATION BY DESIGN: DEVELOPING A CONTEXT-PRESERVING STANDARDIZED METHOD FOR PHYSICAL AND DIGITAL ARCHIVES ARRANGEMENT", I had one of those moments of profound validation, stuff I had known intuitively and and from dimly remembered essays at grad school suddenly materialised from the ether into solid form. The wraith became real, and I mentally punched the air in agreement. She powerfully articulates a truth that many of us in the field have been advocating for years, citing theorists who were sounding the alarm long before the first web browser was ever coded.
She states:
"When [in 1978] archival science and theorists David Bearman and Richard Lytle addressed the matter of provenance beyond hierarchy, they declared that the hierarchical approach “is far too simplistic for modern organizations operating in a world of multi-national corporations, inter-governmental units, regulatory organizations, and federal programmes administered by state, provincial, and local governments"... hierarchical schema do not have meaning across organizations.... [and] Peterson [in 1986] noted that “some archivists claimed that automation made hierarchies obsolete and unnecessary”".
Let that sink in. In 1978, Bearman and Lytle saw that the complexity of modern business had already outgrown the simple folder-within-a-folder model. By 1986, the mere presence of automation, not the internet, not big data, but early enterprise computing, was seen as the death knell for hierarchy.
This isn't a reaction to the web. This is a 50-year-old recognition that the fundamental nature of how organisations and information work is not hierarchical, but relational.
Reality is a Network, Not a Pyramid
Organisations may have a hierarchical org chart, but that’s not how they function. They operate as a complex network of relationships. Functions, usually siloed, work on related subjects, collaborating across invisible lines to achieve business outcomes. This is not done in a strict hierarchy it’s messy, it’s unstructured, it’s relational in an organic ecosystem, it’s not born of a highly articulated masterplan, where allk the parts operation at the same level and using the same information model. A project in marketing is related to a system in IT, which is related to a budget in finance, which is related to a contract managed by legal, the org chart may look like its all being orchestrated from the top in a highly regimented mode by an all-seeing architect, but the reality is organisational functions are usually highly autonomous and dynamic.
This is equally true of our IT and data ecosystems. Systems are not neatly nested one inside the other; they are an interconnected web of applications passing data back and forth. A single piece of data, such as a customer ID, a product code, etc. exists in dozens of systems simultaneously, each giving it a different context.
A rigid, top-down hierarchy simply cannot model this reality. It forces you to choose one "primary" home for a piece of information, severing its vital connections to all the other contexts it lives in. It breaks the relationships that give information its true meaning and value.
From Taxonomy to Ontology: Embracing the Relationship
This is where we must shift our thinking from taxonomy to ontology.
A taxonomy is a parent-child classification. An invoice belongs in the "Finance" folder, which is in the "2025" folder. It’s simple, but it’s a one-way street. You lose the fact that this invoice is also related to a specific supplier, a project, and a customer contract.
An ontology, by contrast, is a fully relational model. It doesn't just classify things; it describes the relationships between them. In an ontology, an invoice isn't just an invoice. It is for a specific project, it was issued by a particular supplier, it will be paid from a certain budget, and it is governed by the terms of a master service agreement.
This relational view is the only way to truly make sense of our data. It’s how modern search engines work, it’s how knowledge graphs are built, and it’s absolutely essential for AI. An AI model trying to detect fraudulent invoices doesn't just look at the invoice itself; it needs to understand its relationship to the supplier's history, the project's budget, and the typical payment terms. Without ontology, you have data. With ontology, you have intelligence.
Beyond the Lifecycle: The Information Continuum
This same philosophical shift applies to the outdated records lifecycle model. The linear, unidirectional path of create -> use -> store -> destroy is a relic of the paper world. Digital information doesn't follow such a neat progression.
A single digital object can be, all at once:
An active business document being collaborated on.
A formal corporate record subject to a retention policy.
A data asset being used for analytics.
A piece of training data for an AI model.
The continuum model reflects this fluid reality. It sees information not on a linear path to destruction, but as an asset with value and context that shifts and evolves over time, across different dimensions of the business. It’s a multi-directional, non-linear model fit for a multi-directional, non-linear world.
Conclusion: It's Time to Leave the Filing Cabinet Behind
Clinging to hierarchical classification and lifecycle models in the digital age is like trying to navigate a sprawling, modern metropolis with a map from the 18th century. The core principles are there, but the model is utterly unfit for the complexity of the current environment.
The archival theorists who warned us about this decades ago were not just being academic; they were being prophetic. They saw that the nature of modern organisations and the logic of automation demanded a new, relational way of thinking. To unlock the true potential of our data, to build intelligent AI, and to create genuinely efficient automated processes, we must finally listen. It’s time to abandon the ghost of the filing cabinet and embrace the models that reflect reality: the interconnected web of ontology and the fluid, dynamic world of the continuum.