# Engineering research domain knowledge

Nothing on this page is assessed. There is no assignment, no checklist, no submission, and it
will never appear in your milestone record.

It is here because every model carries commitments it does not state. A discount rate is a claim
about how much future people matter. An objective function decides whose welfare counts. A system
boundary decides whose costs are invisible. None of those are bugs — they are choices, they are
defensible, and the difference between a competent researcher and a good one is largely whether
they know they are making them.

The twenty-eight graded skills are what every researcher needs. These are what would make you
unusually good at this particular kind of work. A perfectly competent researcher can go a whole
career without them. The best ones don't.

Read one when you are stuck, curious, or procrastinating on something worse. **Read [Where We
Work](sear-lab-domain-knowledge.md) first** — this page will land better once the domain has stopped feeling
new.

---

---

## Philosophy of science

**Why it matters here:** you build models of systems you cannot run experiments on. You will
never have a control group for the Texas grid. So the question of what counts as evidence for a
claim about a system you can only simulate is not abstract for you — it is your entire
methodology, and it has a literature.

Underdetermination, model validation, and what it means for a model to be "right" when every
model is wrong. Start with Popper on falsification, then Kuhn on why fields resist it. If you
read one thing: Box's "all models are wrong, some are useful" essay, which is four pages and
better than its famous sentence.

## The normative content of your modelling choices

**Why it matters here:** a discount rate is a claim about how much future people matter. Three
percent and seven percent are different moral positions, and the choice between them was argued
out by economists and philosophers, not engineers.

The same is true of your objective function, which decides whose welfare counts; your system
boundary, which decides whose costs are invisible; and your spatial aggregation, which decides
whose congestion shows up in the results. None of these are bugs. They are choices, they are
defensible, and you should know you are making them.

Start with the Stern Review's discounting chapter against Nordhaus's response — the two sides
of the most consequential parameter argument in climate economics.

## Policy and institutional literacy

**Why it matters here:** you model ERCOT. Somebody decided ERCOT would be an energy-only market
with no capacity payment, and that decision was political, recent, and revisable. Knowing who
decides, through what process, and on what timescale changes which of your results are
actionable and which are interesting but inert.

PUCT, FERC, the ERCOT stakeholder process, and how a protocol revision request actually moves.
Read one PUCT docket end to end; it is more informative than any summary.

## History of your own field

**Why it matters here:** the grid looks the way it does for reasons, and most of them are
historical accidents that hardened. Rural electrification, PUHCA in 1935, the 1978 PURPA
opening, restructuring in the 1990s, and Texas's decision to stay outside the interconnections.
You will model constraints that exist because of a law passed in 1935.

## The economics of what you model

**Why it matters here:** externalities, market failure, and why a carbon price is a policy
instrument rather than a moral statement. If your optimisation minimises cost, you have adopted
a welfare framework — worth knowing which one, and what it does not capture.

---

*If one of these turns out to matter for your actual work, tell me. Things move from this page
into the curriculum when they stop being optional.*

*See also [SEAR Lab Domain Knowledge](sear-lab-domain-knowledge.md) — the lab's own three layers and three areas, and
which course goes deep on each.*