File Management for Associative Brains: The Case for Tags Over Hierarchies
A concrete open-source alternative exists for developers whose brains resist hierarchical indexing, and the argument recasts disorganization as cognitive impedance rather than a character flaw. The AI-era takeaway is that bespoke tooling is now cheap enough to build around your own mind, but only if you first understand how that mind works.
Hierarchical file organization assumes a linear, tree-structured mind. That assumption breaks for developers who think in high-dimensional, networked associations. The author recounts finding over 200 scattered Git repositories on a single machine, each one a frozen slice of thought, and realizing that forcing them into neatly nested folders triggered mental cache thrashing and cognitive overload. Instead of guilt-ridden self-discipline, he designed a tool that matches his actual processing model.
`igit` starts with zero-resistance bootstrapping: it scans code directories without reorganizing them, lets the user tag or group repos with single-word verbs as they naturally encounter them, and hides noisy directories. Over time, tags and groups become a dynamic semantic projection where the user can pull up all assets by intent, not path. The tool treats context as a firewall: once inside a repo, actions like commit or diff apply in place, and hidden or folded repos never surface to provoke mental tangents. The author reports an unexpected benefit: the workspace becomes a thinking aid that surfaces past explorations under a chosen tag and prompts new questions.
The author generalizes this beyond code: documents, notes, and other personal assets suffer the same violence from single-parent tree structures. He argues for storage decoupling, dynamic views, and in-situ consumption as the future of asset management. In the AI era, the barrier to building such bespoke tools has dropped dramatically, but that alone does not liberate the mind. The real prerequisite is self-knowledge, the ability to abstract the essence of a problem, and the courage to define a personal interaction model rather than conform to industrial defaults.
The moralization of cleanliness and order in professional culture penalizes minds that don't work hierarchically; reframing it as a cognitive style rather than a defect could reduce needless self-blame.
The idea that procrastination and "laziness" might be protective responses to cognitive impedance is a useful lens, but it risks excusing avoidance unless paired with concrete tooling.
The generalization from Git repos to documents and notes is plausible in principle, but the tool only proves the concept for code; the broader claim remains unverified.
The insistence that AI-generated tools still require clear problem decomposition and self-knowledge is a sharp counter to the assumption that AI alone will make everyone productive.
I've pasted the AI's error message before. The root cause was inconsistent encoding; adding a transcoding layer fixed it. Just for reference.