Why Your Component Categories Are Working Against You

My first serious attempt at organising components used eight categories: resistors, capacitors, ICs, transistors, sensors, modules, connectors, and other. I was fairly pleased with this. Spent a Sunday afternoon labelling everything and putting it all in the right drawer. The bench looked clean. The system felt right.

Within six months, the "other" drawer had become the largest drawer. Within a year, I had two sensor drawers — one official and one that had accumulated under the bench when the original filled up. I had modules in three locations because "module" turned out to mean something different depending on what I was thinking about when I filed it. The ESP32 is a module. So is a motor driver board. So is the GPS breakout. These live in completely different mental contexts even though they're all technically modules.

The taxonomy problem: categories that work for a collection at one size don't automatically work as the collection grows. As you accumulate more things and more variety, the categories you built start pulling in opposite directions.

Why physical categories degrade

Physical categories have a hard constraint that digital ones don't: a component can only be in one place. If you're uncertain whether a relay module goes in "relays" or "modules" or "power control," you have to pick one. Then you have to remember which one you picked, every time you look for it.

Most people build their categories in a relaxed, organised state. But components get retrieved under pressure — mid-build, time-constrained, not thinking at full capacity. The category that seemed obvious on a calm Sunday is the one you're trying to remember at 11pm on a Wednesday when you're already annoyed.

This is why physical categories degrade. They're designed by the organised version of you for the disorganised version of you, and the disorganised version doesn't reliably reconstruct what the organised version decided.

The deeper problem is that categories capture how things are similar, but not how you think about them in context. When you're building a motor controller, you think "I need something to drive the gate of that MOSFET" — not "I need something from the ICs drawer." The category and the use-case description don't map cleanly.

Search beats categories for retrieval

If you can describe what you're looking for in natural language — "the small sensor that measures distance without contact" or "the blue regulator module, adjustable, I used it on the power supply board last year" — and get a result, you don't need to remember which category the item was filed under.

This is why AI-based inventory search is genuinely more useful than spreadsheet categories. A spreadsheet requires you to match your mental description to the category name someone chose when calm. AI search lets you use whatever description comes naturally.

The implication is that what matters for an inventory entry isn't the category you assign — it's the quality of the description. A well-described component is findable through search even if it's miscategorised. A poorly described component (just a part number, no context) is unfindable even in a perfect categorical system, because the part number is only useful if you already know it.

Good descriptions include: what the component does, where you'd typically use it, any distinguishing physical characteristics, and any common names beyond the part number. "HC-SR04 ultrasonic distance sensor, blue module, 2cm to 400cm range, commonly used for obstacle detection in robots" is findable from about six different starting points. "HC-SR04" is findable only if you already know you're looking for an HC-SR04.

A better approach: coarse physical, fine digital

What's worked for me is separating the two functions. Physical organisation uses broad, tolerant categories — not eight, not twenty, more like six. Passives. Discrete semiconductors. ICs and drivers. Sensors and modules. Power components. Connectors and cables. That's it. These categories have enough slack that most things land somewhere reasonable, and something landing in a slightly wrong drawer isn't catastrophic.

The fine-grained taxonomy lives in the inventory system, which supports search. When I'm looking for something, I search. When I'm putting something away, I use the broad physical category and log a descriptive entry in the digital system.

The physical system is forgiving enough to absorb imprecision. The digital system is descriptive enough to find things regardless of physical placement.

The goal was never a perfect taxonomy. The goal is finding things when you need them. Those aren't the same problem, and treating them as the same problem is where most component organisation systems go wrong. Stop trying to make the categories perfect. Make the descriptions good enough to search.