The case for curiosity

Why curiosity is the analyst's greatest superpower

👋 Hello! I’m Robert, CPO of Hyperquery and former data scientist + analyst. Welcome to Win With Data, where we talk weekly about maximizing the impact of data. As always, find me on LinkedIn or Twitter — I’m always happy to chat. And if you enjoyed this post, I’d appreciate a follow/like/share. 🙂

---

I used to give this arduous interview question to data science candidates:

Help me figure out how argsort(argsort(x)) works.

The raw answer is simple: it returns the rank of each number in the list. But the point of this question wasn’t to get to an immediate right answer: it’s a great problem to see how people think. The best candidates would try to understand why this works. And the mechanism behind this problem is a bit more nuanced: any vector contains two bits of information — the values and the index. Roughly speaking, argsort takes the values and puts that information in the index, and puts the index information in the values. And once you realize this, you’ll find that there are some cute things you can play around with as a result: successive applications of argsort will simply swap back and forth between these two states, for instance.

In asking this question, I wasn’t going after domain knowledge or sheer intellectual horsepower. I was pursuing that special trait that I believe can make up for either: intellectual curiosity. I’ve spoken extensively before on how business acumen matters and how technical skills are overemphasized. But there’s a third axis of proficiency that I rarely talk about. I want to use this week’s post to give curiosity some space.

Why curiosity matters

Curiosity is a strong signal that someone is truth-seeking

It’s occurred to me recently that some folks are far more driven to get to the truth of things than others. So I’ve taken lately to thinking about why this is. What are the motivation pathways that inspire one camp vs. the other? In some cases, we have genuine intrinsic objectives — we want to be the best at our craft, we want to understand the world, we want to test the limits of our minds. But at other times, we focus on proxy signals of success — graduating college, having a successful career, making money. And, consequently, we supplant intrinsic motivation with extrinsic stand-ins. We prioritize looking successful over obtaining the internal character from which success comes as a byproduct. Are we trying to look good? Or are we trying to actually be better? The world is thence cleaved into two camps: the intrinsically motivated and their imposters (and by virtue of the fact that you’re reading this post, I imagine you’re in the first camp).

Why does this matter? Because intrinsic motivation generally comes with intellectual honesty, and intellectual honesty makes for exceptional analysts. The trouble, however, is that imposters are often quite difficult to identify. They prioritize efficient mimicry over substance, and so the mimicry is often quite good. Deep curiosity, however, is anathema to imposters. And as such, curiosity can be a clear sign of intrinsic motivation. That someone has something in their character that drives them to deeply study a thing, not because of some achievement or perception benefit, but simply because they find some delight in discovering something with their minds. They value the journey, rather than the reward — the truth, rather than being right. And what is analytics about anyways, if not truth?

Curiosity is great for data work

In general, any piece of data work is inherently a curiosity-driven endeavor. And in the chain of problem conception to solution construction, curiosity benefits us along every step of the way. When our stakeholders present us with a problem, curiosity pushes us to ask why — to understand the business reason behind a request, rather than to simply mindlessly comply. When we start to explore the data, it is curiosity that leads us to check for odd issues, to discover surprising insights, to provide value when value isn’t obvious.

Do you build a basic funnel analysis, or do you consider how you might redefine what the funnel should look like? And when you deliver your findings, do you write up a stale deliverable that wraps your findings in eloquent fluff? Or do you write because you’re genuinely excited about what roads lie ahead — because you’re excited to hint to others how else your work might be used?

What’s more: we live in an interesting point in time where technological advances have drastically changed analytics and data science, and as such, these domains are rapidly evolving. Tools change, methods change, processes change. And as such, curiosity will help you stay on top of the latest and greatest in the industry.

Final comments

So this is what I look for in the argsort-argsort problem: do you put some numbers in, come up with an explanation, then move on? Or can you think beyond that, trying to understand the mechanism by which this works? Are you willing to engage with a problem in a way that you don’t just get to an answer, but you grok? Do you seek to simply solve or to deeply understand?

Curiosity is a great character trait for any hire, but for analytics and data science, it’s critical. These domains rely heavily on exploration, and fruitful exploration has never happened without a healthy dose of curiosity.