I follow questions further than I meant to

You can often feel a change before anyone has found good words for it. A tool becomes slightly more capable, a familiar job starts working differently, or people quietly adjust what they expect from themselves and each other. Months later it acquires a name and becomes a trend. I am interested in the earlier moment, when the change is already affecting ordinary lives but the explanations around it still feel incomplete.

When a question catches my attention, I usually begin by checking the obvious answer. One recent example sounded simple: when something has been explained to you clearly and you feel you understand it, have you learned it? You would expect so. In a Harvard physics experiment, students who sat through a polished lecture left convinced they had learned more, while students who had to work through the same material, with more effort and less comfort, performed better when tested. Clarity still matters, but the feeling of understanding turned out to be a poor witness to what had actually been learned.

That question sat inside a larger programme of work that eventually ran to more than twenty separate research passes and several hundred individual claims. Some led back to primary documents and usable data. Others ended at an old article repeating a number that nobody appeared to have measured. Several conclusions that initially looked convincing fell apart during checking. The finished piece may use only a small part of that material. The rest has still done its job by changing what survives. And what survives does not stay a finding: it becomes an argument, a scene, a line a character can deliver without the writer flinching.

Different questions ask for different kinds of work. An analytical film can take a complicated subject and give it direction, pace and a human stake. An application lets an idea meet real behaviour instead of remaining an argument. A series can place social and technological pressure inside characters whose choices feel familiar. A song can carry an experience that would lose something if it were explained directly. These forms share the same foundation, although the research is rarely visible on the surface. The way in is the story, the image, the character, the melody or the thing itself.

I studied public administration, which taught me to look at how institutions turn decisions into outcomes, and later worked in auditing, where a confident report is only the beginning of the examination. That combination became unexpectedly useful when AI made it possible for one person to attempt work that previously required a team. I use AI across research, software, visual production, voices, music and revision. It expands how much ground I can cover, while the important decisions remain mine: what deserves to be believed, what works for an actual person, what feels emotionally true, and what should be removed even after it has taken days to make.

The work underneath should give the finished piece weight without turning it into homework. You may meet the result on a screen or in your headphones; the work beneath it does not have to appear on the surface. Where a piece makes factual claims about the world, I publish enough of the trail - the sources, the data, the questions that stayed open - for those claims to be checked. That practice keeps me honest, because a claim reads differently when another person can follow it back to where it began. The habit is older than this site: it was formed while I was building software that had to prove it was finished, where a confident report and a true one turned out to be different things.

This site brings that work together. The subjects and formats may change, but the pieces tend to begin in the same place: something ordinary no longer behaves quite as expected, and I want to understand what changed, what it means for people, and what form might make it recognisable to someone else.