How Do You Turn an Architecture Idea Into a Strong Research Question?

Most architecture dissertations don't collapse because the writing is weak. They collapse because the question underneath the writing was never strong enough to carry a year of work. A student notices something on a site visit, reads a couple of articles that seem to agree with them, and starts writing before checking whether there's actually a question there at all. I've watched this happen more times than I can count, and it's rarely a lack of intelligence. It's a lack of pause.

An interest is not a question. A topic is not a question. Even a strong opinion isn't one, however confidently it's held. A research question has to point towards an answer that doesn't already exist, using a method you can actually carry out in the time you have. That's a narrower thing than simply caring about a subject, and it's worth sitting with that discomfort before racing past it.

Why an Interesting Idea Isn't the Same as a Researchable One

Say you're drawn to how informal settlements in Lagos adapt their materials over time corrugated sheeting swapped for concrete block as families save enough to upgrade in stages. That's a real interest, and a good one. But "informal settlements and material adaptation" is a subject area, not a question. Subject areas absorb endless reading without ever producing an argument.

The trouble is that reading feels like progress. Collecting images, annotating case studies, filling notebooks it all feels productive. I once sat with a student who'd brought forty pages of notes to a supervision and had no answer when I asked, simply, "What are you trying to find out?" The material was there. The question wasn't. No amount of extra reading was going to fix that gap; only stopping and asking the right question would.

What a Question Needs That an Interest Doesn't

A workable question needs a boundary and a way of being tested. Miss either one and the whole project drifts.

The boundary matters because architecture touches almost everything: climate, politics, material science, how people actually behave in a room. Robin Evans' writing on the corridor is a genuinely huge claim about domestic life, but it holds together because it's tied to one spatial element and one historical shift, not to "the home" in general. If you're chasing that kind of ambition, borrow the discipline that comes with it, not just the scale.

Testability is a separate matter. Wondering whether Oscar Newman's defensible space ideas actually reduced crime on a specific estate is a legitimate question, but properly testing it needs decades of crime data most students can't get near. Better to find that out in week two than month four. And it's worth asking, honestly, whether anyone beyond you would want the answer. A question can be perfectly specific and still only close a gap in your own reading, rather than one that exists in the field.

Narrowing Without Losing the Point

The instinct when a topic feels too big is to bolt on more qualifiers until it looks specific. This rarely works, because narrowing means deciding what to leave out, not decorating what stays in.

Take housing density. Broad enough to carry ten different dissertations, vague enough that most of them would be weak. Narrowing it means a run of real decisions: which city, which period, which typology, which outcome you're actually measuring. Each decision removes territory; it doesn't add detail to what's already there.

Here's the part that surprises people: the narrower version is usually more interesting, not less. "Does housing density affect community?" tells a reader nothing they didn't already assume. Asking why residents of Vienna's mid-rise social housing report different satisfaction with shared courtyards than residents of perimeter-block schemes built in the same decade gives someone an actual picture to argue with.

Testing Whether the Question Survives Contact With Reality

Before committing, push the question a little. Can it be answered with evidence you can genuinely reach? Does it quietly assume the outcome rather than asking for one?

That second trap catches good students constantly, because a strong conviction can feel like open enquiry from the inside. If a question can only be answered one way, it's an argument wearing a question mark. Real research questions carry some risk of turning out differently than you hoped. If that risk isn't there, ask yourself why.

Where Good Questions Actually Come From

The mistake many students make is hunting for something that sounds suitably academic rather than paying attention to what they've already noticed and found odd. It's worth spending a proper hour working through a good set of architecture research topics before you settle on anything, not to copy one, but to see the shape of what makes some ideas sharper than others; the difference is usually more instructive than it looks at first glance. Scrolling half-heartedly through the first five results you find usually lands you somewhere safe, well-worn, and a little lifeless. The thing that didn't match your expectations is nearly always more useful than the thing that confirmed them.

Peter Zumthor's Therme Vals gets written about constantly, almost always in glowing terms. A student who instead noticed that the building's celebrated tactility seemed to matter less to actual bathers than the sequence of light and dark spaces a claim you could check against visitor accounts rather than architectural criticism alone would have a far sharper starting point than one who simply wanted to write about "materiality" because it sounded appropriately serious.

I can't promise a supervisor will spot borrowed enthusiasm within a paragraph. Sometimes it takes weeks. But it tends to show eventually, usually when a student can't explain why the question matters to them beyond "it seemed like a good topic."

Recognising When the Question Needs to Change

Committing to a question doesn't fix it permanently. Early fieldwork often shows the original question was aimed at the wrong scale, or asking about the wrong thing entirely.

A common sign: your evidence keeps answering a slightly different, adjacent question more clearly than the one you asked. Or the phenomenon you wanted to study barely shows up in the case you chose. Adjusting at that point isn't a failure, even though it can feel like one when you've already told your supervisor what the dissertation is "about." Most first attempts are a little wrong. That's not something to be embarrassed about; it's closer to what research is actually for.

Getting From Question to Proposal

Once a question holds up under this kind of pressure, the rest of the proposal tends to follow on its own, because your method and case selection take their shape from what the question genuinely demands, not from what sounds thorough on paper. A vague question produces a vague method, because there's nothing precise to build one around.

That's really the whole case for spending real time on the question before writing a single other word. It isn't a formality to get through before the actual work starts it more or less is the work. Everything after it is mostly the business of answering something specific enough to deserve the effort you're about to give it.