essay
The Process of Curiosity
A personal approach to understanding complex systems.
I approach unfamiliar subjects through curiosity about how things work beneath the surface. Rather than treating something as a single object, technology, or concept, I tend to see it as a collection of interconnected systems, each operating according to its own properties, constraints, and relationships. My instinct is to investigate the underlying infrastructure of whatever I'm trying to understand. I want to know not only how something works, but why it works that way, what makes it possible, and what decisions or circumstances led to its current form.
This approach isn't specific to software or engineering. I apply it to physical systems, mathematics, biology, philosophy, and subjects in which I have little or no formal expertise. The underlying process remains similar even when the subject changes.
Understanding through layers
Consider something as ordinary as a road. A road isn't simply a surface that vehicles travel upon. Beneath it are layers of materials, each chosen for particular properties and purposes. Those layers interact with the ground, environmental conditions, and the stresses imposed by whatever travels above them.
To understand the road, I might investigate how it's constructed, why certain materials are chosen, how those materials interact, how they deteriorate, and what properties make one design preferable to another. If the question involves long-term deterioration, chemistry might become relevant. If it concerns structural loading, I might investigate mechanics or material science. But I wouldn't automatically continue into quantum physics simply because it's another layer beneath the chemistry.
The purpose of investigating deeper layers is to understand the system, not to reach the deepest possible level of explanation.
Depth should serve the question being investigated.
Recognizing connections
Experience contributes something important to this process. Over time, studying different systems builds a collection of mental models: patterns of interaction, dependencies, constraints, tradeoffs, and cause-and-effect relationships.
When I encounter something unfamiliar, I frequently recognize structural similarities to things I've encountered before, even when those things belong to entirely different disciplines. These connections aren't necessarily obvious, and they're not always correct. But they provide starting points for investigation.
A mathematical relationship might illuminate a computer architecture problem. A concept from electrical engineering might suggest a way to think about biology. An observation about physical infrastructure might reveal something useful about software architecture. I don't assume that similar-looking systems operate identically. Instead, I use the resemblance to ask better questions. Sometimes the most valuable discovery isn't a new piece of information. It's recognizing a relationship between things I already understand.
Curiosity guided by relevance
My curiosity is exploratory, but it isn't entirely directionless. I tend to follow questions as they emerge from the subject I'm examining. One explanation exposes an assumption; examining that assumption reveals a dependency; understanding that dependency suggests an unexpected connection. This can continue through several layers of investigation.
However, I don't believe every possible connection deserves equal attention. I follow a line of inquiry because it seems relevant, potentially explanatory, or sufficiently interesting to justify further exploration. I may temporarily depart from the original subject when a promising connection appears, but I want to understand why that departure matters. This means the process doesn't necessarily resemble a linear research plan. It's more like exploring a connected landscape, following paths that offer useful perspectives while retaining an understanding of where the exploration began.
Evidence, uncertainty, and explanation
I'm comfortable considering possibilities without immediately accepting them as facts. I distinguish between what is known, what can reasonably be inferred, what remains hypothetical, and what would require additional evidence.
An explanation can be mathematically elegant without accurately describing a physical phenomenon. An analogy can be useful without being literally true. A hypothesis can be worth exploring even when there isn't enough evidence to support it yet. I also recognize that understanding how something works isn't necessarily the same as understanding what it means or where it might apply.
Someone may discover a mathematical relationship without recognizing its applications. An existing scientific explanation may become useful in an entirely different discipline. Sometimes a breakthrough consists of connecting established knowledge rather than discovering something fundamentally new.
I want to remain open to those possibilities without allowing speculation to quietly become certainty.
Knowing when to stop
I don't need to understand every underlying detail before I can usefully understand a system. The depth of investigation should depend on the problem, the importance of the unknowns, and the value of further understanding. Sometimes I want enough knowledge to explain a phenomenon. Sometimes I want to predict its behavior. Other times, I want enough understanding to design, construct, improve, or challenge something myself. Those goals require different levels of investigation.
I consider an inquiry sufficiently developed when I have a useful mental model for my present purpose, can identify its important assumptions and limitations, and understand which unanswered questions would materially change my conclusions. That doesn't mean the subject is exhausted, it just means I have reached a useful stopping point. Curiosity can always return to it later.
The role of AI
AI has fundamentally changed how far I can pursue this kind of thinking in a single session. Historically, exploring an unfamiliar dependency might require considerable time locating references, learning terminology, and establishing enough foundational understanding to ask the next question. Often, the effort of obtaining an answer interrupted the curiosity that prompted the question. A capable AI companion reduces that friction.
I can ask about a concept, examine the explanation, identify an unfamiliar dependency, and pursue it immediately while the original thought remains fresh. This doesn't eliminate the need for skepticism, verification, or independent reasoning. AI-generated explanations can be incorrect, incomplete, or misleading (i.e. hallucinated). It does, however, allow me to maintain the momentum of an investigation across subjects that would otherwise require substantial preparation.
Importantly, I don't want AI to determine the direction of my curiosity for me. I want it to explain, challenge, connect, and help investigate. I want it to recognize potentially useful relationships and make them available for consideration, rather than continuously expanding the investigation on its own.
AI should extend my capacity to explore, not replace the curiosity that directs the exploration.
The underlying philosophy
I approach unfamiliar problems with the belief that I can usually develop a meaningful understanding of them if I have sufficient time, access to information, and an opportunity to investigate. That doesn't mean I believe every problem is solvable through thought alone, or that expertise and experimental evidence are unnecessary, it means I trust my process of constructing understanding.
I don't necessarily need to begin with expertise in a subject; I need a way to examine its components, investigate their relationships, test my assumptions, and progressively build a more accurate mental model. The objective isn't simply to collect facts or arrive at an answer as quickly as possible.
The objective is to understand enough of the underlying structure that unfamiliar things become comprehensible, connections become visible, and new questions become possible.
That is what I mean by the process of curiosity.