Engineering and Career: Start Here

Engineering is the discipline of making something work, and of knowing why it works. This wing treats it as a liberal art because the core skills are the old ones: learn an unfamiliar system, reason from evidence, and make something you can explain to another person. Tools change every few years. Those three habits do not.

The premise

A learner in this wing is asked to do three things, again and again.

1. Learn an unfamiliar system. Read its documentation, draw it, and say what each part depends on.
2. Reason from evidence. Say what an observation establishes and what it leaves open.
3. Make something you can explain. If you cannot explain a piece of work line by line, at the level that matters, you do not yet own it.

The Greeks had a word for craft knowledge, techne, and Aristotle drew a line between making a thing and acting well (Nicomachean Ethics VI.4-5). Engineering sits on that line. You make artefacts, but you also decide what is worth making and what counts as done, which is a question of judgment, not just skill. See The Philosopher-Engineer: Study and Workshop Together for the larger ideal this wing serves.

Fact versus analogy. The claim that engineers need evidence-based reasoning is ordinary professional practice. The claim that this makes engineering a liberal art is an argument from analogy to the trivium and quadrivium (see Liberal Arts and Reasoning: Start Here). Treat it as a useful frame, not a theorem. Provisional

Capabilities that outlast tools

Products, versions and exam syllabi expire. A capability is something you can demonstrate with whatever tool is current. The table lists eight, each with a demonstration worth keeping. The full argument is in Building a Career on Capabilities, Not Tools.

How the pages fit together

The pages form a loop rather than a ladder.

• Method. Deliberate Practice: Choose Targets You Can Observe turns vague goals ("study networking") into things you can observe, and gives the practice loop the rest of the wing reuses.
• Troubleshooting. Evidence-Based Troubleshooting: Separating Explanations trains the central skill: asking what an observation establishes and picking the next test that splits the competing causes. Its companion, Stale Information and Dependable Reports, is about information that was true once and may not be now.
• Quantitative habits. Quantitative Reasoning with Stated Assumptions shows that arithmetic is only as good as the assumptions you state.
• AI agents. Working with AI Coding Agents: A Disciplined Loop is a disciplined loop for coding agents. From Assistant to Agent: Tools, MCP and Evaluation goes one step further, to bounded tool-using agents judged against a non-AI baseline.
• Writing. Technical Writing as Reasoning is where reasoning becomes something others can use.
• Career design. Building a Career on Capabilities, Not Tools applies the same discipline to choosing certifications, courses and projects.

Suggested reading orders

If you are new to the field. Read this page, then the method page, then troubleshooting. Do the exercises on paper before you touch a lab. Add the quantitative page and the writing page next. Read the AI pages last, because judging an agent's output requires the habits the earlier pages build.

If you already work as a technician or developer. Start with troubleshooting and the stale-information page; they will probably name habits you already half-have. Then read the AI workflow page, because that is where unreviewed shortcuts do the most damage. Finish with the career page and treat it as an audit of what you are currently studying.

Connections to the rest of the library

• Practice is practice everywhere. The weekly loop on The Weekly Loop: Read, Reconstruct, Make, Defend, Log and the guitar practice material share a structure: a target you can hear or observe, a repair of the first real error, and a return visit. The monochord is a small example of a physical system you can measure and predict.
• The academy shows the oldest versions of these habits: Euclid's postulates are an exercise in stating assumptions before deriving anything, and the elenchus is cross-examination of a claim, which is what a good diagnosis does to a hypothesis.
• The science section applies measurement discipline to sound; the Pythagorean comma is a case where exact arithmetic exposes a mismatch that intuition hides.
• The sourced quotations are a reminder of how much of this wisdom is old.

A rule for when to stop

Perfectionism is the quiet failure mode of technical learners. A practical rule of thumb: stop polishing when the acceptance checks pass and someone else can use the result. The acceptance checks must be written before you start, otherwise you will move the goalposts. "Someone else can use it" means they can run it from your instructions and undo it if it goes wrong. Past that point, further polish is usually a way of avoiding the next, harder problem.

Try this

1. Pick one thing you are currently "studying". Rewrite it as a target someone could check: "explain X to a beginner and answer one changed case", not "learn X".
2. List the eight capabilities above and mark each one: demonstrated, studied only, or untouched. Be honest; the point is to find the gap, not to feel good.
3. Choose the reading order above that fits you, and write down what you expect to be able to do after the first two pages. Check against that prediction later.

Further reading

• Aristotle, Nicomachean Ethics, Book VI (technical skill, practical wisdom).
• Ericsson, Krampe and Tesch-Romer (1993), cited above, for the origin of the "deliberate practice" idea.
• Beyer et al., Site Reliability Engineering (O'Reilly, 2016), freely readable at sre.google/sre-book.
• Anthropic, "Building Effective AI Agents" (2024), for the case for simple, measured designs.