跪拜 Guibai
← Back to the summary

File Management for Associative Brains: The Case for Tags Over Hierarchies

Mosaic Pieces -- Brushing Off Thoughts, Building Tools, and Self-Consistency: An Order for Everyone in the AI Era

Preface

In my youth, I favored Huineng's "fundamentally there is no thing, where can dust alight," finding it eloquent and of higher zen. As I gained experience, I came to realize that pure idealism often lacks a footing in reality; body and mind are deeply entangled with external things, and Shenxiu's "constantly brush and wipe, lest dust alight" better fits the actual world.

This is like a metaphor for the second law of thermodynamics: an isolated system naturally slides toward disorder and chaos, and to maintain local order, external energy must intervene. Even Maxwell's "demon," which tries to reverse entropy by distinguishing fast and slow molecules, modern physics has ruled: its observation, recording, and forgetting all come at the cost of consuming energy and emitting entropy outward. "Diligent brushing" is never a one-time fantasy; it is the life that guards the gate, soberly paying the price for order in a long dissipation.

However, if the energy cost of maintaining order spirals out of control, the demon at the gate will eventually collapse from exhaustion.

For a long time, I considered myself "somewhat clever but extremely lazy." In school, I marveled at teachers who managed lesson plans, tests, student analyses, and schedules neatly in their computers; at work, I genuinely admired some old-school engineers who could keep vast engineering assets and even branch structures meticulously organized. I could only roughly categorize into a few broad groups, and I blamed my laziness for failing to "constantly brush and wipe."

Only recently, upon re-examining, did I realize this is actually a cognitive architecture impedance mismatch. They can do it because their mental chains naturally fit the hierarchical encoding of "tree directories," while my thinking is leaping and divergent—what emerges is interwoven high-dimensional features and networked connections that cannot be condensed into a single, one-way absolute path.

The hidden danger of this cognitive load erupted on my several computers recently.

I wrote a script to scan the entire disk, and the result was suffocating: on just one device, there were over two hundred scattered Git repositories. To truly "brush and wipe," I would have to enter each one. But for a leaping thinker, every repository is not a dead folder but a frozen slice of thought—merely moving between paths triggers a storm of context switches; any subtle change in information or weight fluctuation instantly connects to an entire associative network, igniting unstoppable mental tendrils.

My brain's rated power cannot sustain this level of cache thrashing and interrupt storm; forcing organization would only drain my mind through excessive work. Common tools on the market, without exception, are built on hierarchical indexing logic. They don't help; they push me to do more futile, useless work.

Since the problem is a fundamental mismatch between processor architecture and data organization, the solution is not to overclock yourself with willpower but to rebuild the underlying interaction protocol.

I stopped the futile self-blame for "neglecting to organize" and instead dissected my cognitive model: my brain is essentially a high-dimensional, networked feature-matching system that naturally rejects the serial addressing of tree structures. Based on this pain point, I reimagined a management tool for feature-sensitive thinking: replacing hierarchical paths with multi-dimensional feature projections, setting up a "context firewall" to block unnecessary mental provocation. It is no longer a supervisor urging work but an external co-processor that shares the work cost with the gatekeeping demon.

Dear readers, this mosaic piece arises from the birth of this small tool, but if it stops at code and engineering implementation, it would betray the original intention of this mental examination. In this piece, I want to discuss with you several deeper problems:

  1. The essence and branches of brushing: Does the "organizing" we cling to in daily life brush away dust or blind the mind?
  2. Three deliberations and generalization: Can this decomposition scheme suitable for leaping thinking be extended to documents, knowledge, and even universal asset governance?
  3. Self-knowledge and empowerment in the AI era: With the technical threshold gradually lowering, what is the core prerequisite for constructing personal order?

Brushing Off Thoughts: From Moral Self-Reflection to Cognitive Verification

Frankly speaking, when most people face chaotic order, their first reaction is often moralistic judgment.

In traditional education and workplace contexts, a tidy desktop and well-ordered documents are seen as symbols of self-discipline and maturity, while clutter and fuzzy categorization are often labeled "lazy," "careless," or "lacking organization." For a long time, I was also captive to this standard. Whenever I saw my messy hard drive, an inexplicable internal drain arose—blaming myself for not "constantly brushing," for not having the willpower to overcome entropy.

However, the first insight from this examination was: the first step of "diligent brushing" is not to brush external things, but to brush off your own thoughts.

The dust on thoughts is precisely the obsession that "the standards others can meet, I must also adapt to by cutting my feet to fit the shoes."

When I truly stopped and dissected my mind with an objective, cold engineering perspective, I found that leaping thinking is not a physiological defect but another computational architecture:

If you are a neural computing unit naturally suited to run graph databases or vector retrieval, but you insist on using willpower to simulate a traditional B+ tree unidirectional index, then every file search triggers massive internal exceptions and cache replacement.

Author's note: Based on my personal experience, the "procrastination" and "laziness" I often blamed on myself are actually protective mechanisms of the mental system when it encounters severe cognitive impedance. Wiping away the dust of moral self-blame and recognizing your cognitive strengths and weaknesses is the first step to clarifying real needs.

Investigating Things and Building Tools: Smooth Construction and Spatial Leap

After recognizing my cognitive model, the solution came naturally. Since I am not suited to layer-by-layer traversal of physical directories, why force myself to catalog files like a librarian?

I conceived and implemented a local workspace called igit (open-sourced on GitHub: leobert-lan/igit). In designing it, I did not adopt any market logic of "teaching users how to manage properly," but entirely around the traits of leaping thinking, breaking it down into a complete lifecycle of smooth transition and spatial operation:

Phase One: Smoothly Building Space from Physical Directories

Many tools are hard to promote because they demand users to "transform completely" at the first step—before adopting them, you must spend days reorganizing and categorizing all directories. For leaping thinkers, this is like igniting an interrupt storm at the starting line.

Therefore, igit's first phase pursues zero-impedance smooth cold start:

You don't need to set aside a block of time for a transformative deep clean; the dynamic space naturally grows from each casual "light brush" in daily work.

Phase Two: Free Flow and In-Place Actions in Dynamic Space

When business groups (Group, N:1) and attribute tags (Tag, N:M) gradually take shape through daily flow, the tool naturally enters its second phase:

  1. Replace hierarchical paths with multi-dimensional semantic projection: On the physical disk, code is scattered; but in the cognitive view, through a set of free Tags and Groups, all assets are instantly flattened and gathered according to the current thought. Want to see core business, see core business; want to find a prototype with a specific attribute, aggregate through a tag funnel.
  2. Space is context, verbs are intent: Once you enter a single repository node, all subsequent operations take effect in place by default. No longer do you need to repeatedly type lengthy paths or memorize tedious command-line arguments. Want to commit, type commit; want to see changes, type tig. Missing arguments are gently handled by option interactions, never throwing a cold syntax error.
  3. Context firewall blocks provocation: Objective Git commits and branches are the underlying facts, while the developer's subjective governance conclusions are permanently deposited. Repos that physically exist but are marked as folded or hidden never pop up to provoke meaningless thoughts.

Extended Value: From Passive Retrieval to Actively Guiding Thoughts

After this space runs stably, I experienced an unexpected gain: It not only passively accommodates my leaping thinking, but also becomes a mental scaffold that actively guides my thoughts.

When people are extremely focused or extremely busy, they often fall into tunnel vision of a single path; when idle and wanting to start new thinking, facing a blank screen, they often don't know where to begin.

At that point, I can pull out related project slices under a specific Tag. Looking at these nodes that freeze explorations from different past periods, I try to induce myself:

It evolved from a mere "repository indexer" into an external resonance box that saves my mental breakpoints and actively activates thoughts when needed.

Entering Three Deliberations: The Essence of Asset Management is "Cognitive Mapping"

After building the tool, I entered the "three deliberations" stage and began to examine a broader proposition:

Is this decomposition scheme suitable for leaping thinking applicable only to Git code repositories? Do daily document management, knowledge notes, and even more generalized personal digital assets also suffer from the torture of hierarchical structures?

The answer is clearly yes.

Dear readers, consider your daily life: For example, you receive a commercial cooperation agreement with a confidentiality clause in an instant messaging app:

Similarly, you download a classic paper on distributed system consistency protocols in your browser:

Single-parent tree directories force a person to assign a unique "lifelong birthplace" to information at the moment of its creation. But in reality, information's attributes are always multi-dimensional. Leaping thinkers fear organizing documents precisely because, when deciding this "birthplace," the brain is forced to rehearse future retrieval paths, exhausting mental bandwidth.

In my view, future generalized asset management will inevitably move toward a path consistent with this philosophy:

Shifting from "physical container classification" to "cognitive intent projection" is not laziness; it's a paradigm shift in information management from matching machine disk addressing to matching human thought flow.

Reflections on the AI Era: Self-Knowledge, Essential Insight, and the Threshold of Tool Building

Writing this, I cannot avoid the dramatically changing technological cycle we are in.

Five years ago, building a fully adapted workspace for a unique cognitive habit would have been extremely costly: you needed deep knowledge of system APIs, terminal interaction, cross-platform compatibility, and performance tuning. For non-programmers, this was like a castle in the sky—most could only squeeze into off-the-shelf software.

But with AI rapidly advancing, the threshold for achieving technology is gradually lowering. Even a person without a deep engineering background, if they can clearly articulate the problem, can use large language models to build a handy tool that meets specific needs.

However, the lowering of tool-making barriers does not automatically equate to mental liberation.

Implementing a tool that truly fits oneself has extremely strict prerequisites. These prerequisites are not about how to write prompts, but:

  1. Can you, amidst lengthy external noise and judgment standards, constantly brush your heart to perceive your truest cognitive traits, thinking boundaries, and rigid needs?
  2. Do you possess the abstract ability to see through appearances and grasp the essential problem behind them (e.g., mathematical mechanisms, information theory laws, data structure and algorithm essence)?
  3. Do you have the courage to step out of the secular tool paradigm and independently define an interaction model and life order that fits your temperament?

In the AI era, for function units and boilerplate code with clear boundaries, clear design, and standardized specifications, writing them is rapidly degrading into machine-executable labor; but without clear architectural definitions and boundary division, fully automated writing of complex systems remains an unattainable fantasy.

Furthermore, if you lack the ability to see through appearances to recognize essence, even with AI in hand, you often just spin at the surface, creating another mediocre toy. If you haven't brushed off your hearts' obsessions, what you produce with AI is just a slightly more complex "hierarchical directory supervisor" that continues to whip your spirit using mass-industrial standards.

This is the captivating side of AI: it greatly reduces the resistance from "conception" to "artifact," allowing people to break out of the cage of ready-made software and materialize their unique cognitive paradigm into reality.

Conclusion: Everyone Has the Virtue of the Dragon

Returning to the Zen verses of Huineng and Shenxiu at the beginning.

We finally understand that "constantly brush and wipe" is never about futile suffering to meet secular tidiness standards, but about consciously tending to your thoughts and emotions at all times, brushing away the biases and internal depletion that cling to the cognitive surface.

First know yourself, then investigate things to extend knowledge; penetrate the essence of things, and only then can you use tools to cultivate yourself.

In Confucian Yijing thought, there is a grand cosmic view and humanistic concern—the "Qian" hexagram's "a flock of dragons without a leader, auspicious" anticipates a world where "every person is like a dragon." That absolutely does not mean all people should erase their individuality and vie for the same peak; rather, each person can illuminate their true nature, align their position, and extend their innate potential to the fullest. Each can strengthen themselves according to their own rhythm, establish their own order, and never be tamed by any single standard.

The era is placing the chisel of lowered-threshold technology into everyone's hands. You don't have to be a slave to any off-the-shelf tool, nor force your leaping inspiration to conform to any rigid hierarchy or secular mold of thinking.

Brush away the dust in your heart, recognize your true self, and establish your own order in work that aligns with your mind. On this path of self-rescue, everyone has their own light, and everyone can be like a dragon.

Related works: Smashing the Cage -- Design Ability / Mosaic Pieces (Six)

Comments

Top 1 from juejin.cn, machine-translated. The original thread is authoritative.

AI向善

I've pasted the AI's error message before. The root cause was inconsistent encoding; adding a transcoding layer fixed it. Just for reference.