The Hardest Part of Analysis Is the Question
When you build prediction and analytic tools for a living, people who know you tend to reach out when things are happening in the world. At my kid's tennis match recently a Dad came up and excitedly asked what is going to happen in Iran. Someone else at a volunteer event asked if we were going to "buy Greenland."
As a consumer of news, I of course have these same questions. But as someone who gets to work with analysts at think tanks and governments in the U.S. and Europe, I often try to think "what are they asking right now?"
Over time, I’ve come to recognize that when news is breaking, this is when expertise, experience, and wisdom matter most. Not because information is scarce, but because it’s abundant. And because I've noticed the pressure to form an opinion tends to lead me to questions that feel responsive, current, and informed, but that don’t actually guide useful thinking, especially if I was an analyst trying to inform policy or decision-makers myself.
For ARC I am starting to design a guide to help you ask the best possible research or forecast question based on what your goals and objectives are, so I've been quietly doing my own question asking in the background, researching how analysts ask the most "useful" questions, and paying attention to process. One event I tackled was the renewed U.S. interest in acquiring Greenland, and what that episode suggests about the future of U.S. and European economic relations.
The Questions That Felt Right -- But Weren’t
After the Venezuela incursion that netted Maduro, the U.S. Administration turned its attention to Greenland, and my reaction was probably predictable. I started asking myself whether the proposal was serious or rhetorical, how Greenlanders might respond, and what Denmark, the EU, and NATO countries would do if pressure escalated.
Those questions felt natural. They were also, in retrospect, a sort of trap. All I was doing was aggregating existing journalism - cataloging reactions rather than understanding what was actually at stake for the future. If I was a professional analyst and had followed them all the way through, I would have ended up with a summary of opinions and statements (something an AI could easily do too!) and still very little insight into what the implications are to economic relations beyond the news cycle.
At some point I’ve learned to pause and ask a more uncomfortable question: does this event matter at all?
In this case, the answer wasn’t really about Greenland per se. It was about if the base rate of U.S. behavior and policy was truly changing. About alliances. About how power is exercised among partners. About how economic and strategic relationships behave when one side signals a willingness to push against long‑standing norms.
Once I framed it that way, the analysis began to shift.
The Question Beneath the Question
Instead of treating the story as an isolated curiosity, I started to see it as a stress test. Not of Arctic geopolitics in the abstract, but of U.S.–European relations under conditions where policy, sovereignty, economics, and strategic cooperation were becoming increasingly misaligned. What I didn’t know was whether this kind of unilateral signaling was becoming more normalized in U.S. policy toward allies, or how elastic European economic cooperation with the United States really is when core issues of sovereignty are implicated. Is this an early indicator of a broader shift in U.S. behavior?
Those uncertainties were more interesting because they could help anyone with skin in the game understand how serious this all was and how prepared they needed to be for change. They also made me realize that my first attempt at a research question was wrong.
My instinct - especially as someone who has written hundreds of falsifiable forecast questions over the years - was to ask something like: "Will the United States obtain sovereign ownership of Greenland by x date?"
It sounded to the point, but it was also analytically shallow.
Even if we could answer that question with high confidence, it would have limited value. In other words, if I answered this question perfectly today, what would the policymaker reading my work still not know tomorrow?
The real problem wasn’t just about understanding the future of Greenland, it was understanding consequences. What does it mean economically, politically, strategically, if a major power openly pressured an ally over territory? And what would that signal about the future of the transatlantic relationship?
When I applied that test here, the gap was obvious. So I rewrote the question, not just to make it broader, but to make it more structurally useful:
"How would sustained U.S. pressure on an EU member state over territorial or sovereignty issues affect long‑term U.S.–EU economic and strategic cooperation?"
That question immediately did things the earlier one could not. It remained relevant even if the Greenland story faded. It supported multiple plausible futures rather than collapsing uncertainty into a yes‑or‑no outcome. And it pointed directly to the kinds of tradeoffs decision‑makers actually face.
Just as importantly, it imposed discipline on the rest of the analysis. I could now tell which information mattered and which did not. I could ask ARC to author scenarios where cooperation absorbed the shock quietly and the status quo was largely maintained, and others where trust eroded in ways that showed up later in trade policy, industrial coordination, or security agreements. I could ask ARC to generate indicators that would help distinguish between those paths depending on which direction they were heading. Asking a better question early enough shaped everything that followed.
The Discipline I Keep Reminding Myself Of
I’ve noticed that when my perspective starts to feel scattered or overly reactive, it’s almost always because I skipped this step. I let the event define the question, rather than forcing the question to reveal what the event actually tests.
Over time, I’ve learned to watch for a few warning signs. If I’m mostly summarizing what people are saying. If I’m answering a question that no clear decision depends on. If the work would feel stale six months from now even if it seems highly relevant today. Any of those requires rewriting my original question.
While there are analytic methodologies that help, there’s no exact formula that guarantees the right research question. But there is a discipline: anchoring analysis to real decisions, targeting uncertainty rather than events, and elevating the question until it genuinely changes how someone might think about the future.
When an analyst gets that right, the analysis that follows is far more useful. When they don’t, the decision maker is flying in the proverbial dark and having to do what the rest of us in the peanut gallery do: read the papers, listen to the podcasts, and talk about it with their colleagues and friends which is the last place we want them to be!
