What I'm Up To - Grad School

In the last few weeks, I’ve met a lot of people. As a recent college graduate tends to do, I’ve fielded questions about the next steps in my life. So when I say I’m pursuing graduate school in “analytics” I can, at this point, rattle off a shallow-but-not-inaccurate explanation of what “analytics” is: a combination of data science (think “big data,” “AI”, and “machine learning”) and Operations Research (a set of quantitative methods used for making decisions).

In explaining this, I tend to draw on some canonical examples (like the Traveling Salesman problem) and some from my own experience (like modeling pickup basketball wait times) to clarify what I’m really studying.

While this answer is typically satisfying enough for people’s curiosity (or enough for them to confirm their boredom), it doesn’t really get to the root of why I want to spend another year learning about this stuff. That’s going to be one purpose of this post.

The two elements of the analytics program I’ll be in next year are Data Science and Operations Research. These related and intertwined fields are both focused on coming up with the best answer to a question. But these are the kinds of questions where there may not be a single “correct” answer, just one that’s better than others. Here’s an example.

Let’s say it’s a rainy Saturday in October. You’re a college student with an itch to play basketball and three midterms to study for. On one hand, you don’t want to waste too much time. On the other hand, you know playing basketball could help you focus and will generally help your mental health. Let’s say you’re a 25 minute walk from the gym, and you’ll play for an hour and a half, or 90 minutes (not including initial waiting time). Lastly, let’s just say you expect the wait time to be 15 minutes, but it could be much longer (45 minutes) or much shorter (no wait time at all). You have a few options:

• Take some time to walk to the gym (25 minutes) and see if anybody is playing, open to the option of walking back (another 25 minutes) if the wait to play is “too long.” Result: you waste 50 minutes walking back and forth, get wet, and don’t even get to play basketball.

• Walk to the gym and play basketball no matter how long the wait is, then walk back. Result: you take somewhere between 140 and 185 minutes, but you scratch your itch to play basketball.

• Assume the wait is too long and don’t even bother going to the gym. Result: you’re still distracted at home, but at least you have some more time to study.

What’s the best option here? This is an example of a decision that has a “best” option, but just not one that’s incredibly clear. Let’s say you use a rigorous and well-justified method to make your decision, and commit to playing basketball no matter how long the wait is. You’re confident that your method has yielded the best possible decision. In this case, you’ve found an “optimal” decision.

For a long time, I’ve believed that there are optimal strategies for solving problems. Part of it comes from having to come up with “true” arguments in debate, a larger part from solving open-ended engineering problems, and still more from my everyday experience. My commitment to an optimal strategy underpins a lot of my fundamental beliefs and informs the way I try to live each day.

But finding this optimal strategy isn’t always easy. Let’s imagine our problem becomes a little more complicated: instead of studying alone, you’re studying with four other people, who you also often play pickup basketball with. You’ve agreed that you’ll all spend the day together, whether that’s in studying or other endeavors. Each of you has an itch to play, but at varying levels.

What do you do?

A convenient place to start would be a vote. Everyone picks their best option and votes for it, and you all agree to whatever that is. Let’s think about the worst case scenario: a 2-2-1 tie. Then you go to a tiebreaker, where you eventually pick an option that was only the best for two people. This option could be terrible for the other three, and still get picked. There’s probably an intermediate option (or one not listed) that would, on the whole, be better. But because you’re choosing to vote, you’ve eliminated the chance to come to a good compromise.

This brings me to a very common tradeoff in decision-making: convenience vs. rigor.

This is the tradeoff you have to manage when deciding on a place to eat for dinner. It’s faced by leaders choosing what tasks to delegate (and who to delegate them to). Because this tradeoff deals with our own decision-making, we have to strike a balance every time we make a decision.

Like any tradeoff, the balance for this depends on the decision. Choosing between a set of t-shirts may have some very rigorous method to pick the very best one each time, but it likely won’t make a significant impact on your overall happiness, and the time saved by picking randomly can be worth the “sub-optimal” outcome. On the other hand, deciding how to finance a mortgage may have a bigger impact on your life, and that sub-optimal strategy can get in the way of your long term financial goals. And if you’re planning out a series of smaller decisions that have some kind of impact on each other, a rigorous method may be necessary to stop negative effects from compounding. In essence, your convenience-rigor balance depends on how much sub-optimality you’ll accept. As I’ve gone through my life, I’ve found myself increasingly intolerant of less-than-optimal decisions, especially in situations where the optimal choice isn’t any more difficult than a sub-optimal one.

Through that journey, I began to see how OR and (more broadly) Analytics can be the key to getting closer to these optimal strategies. In situations where there’s a tractable (feasible) solution, OR and Analytics offer a set of methods that can be used to make difficult decisions and develop close-to-optimal strategies. While doing so is rarely convenient, I do believe it’s both worth it and a vast improvement on the way we make decisions now. As data continues to reshape our lives, these methods are an incredible opportunity to make better decisions and make the world a better place.

Over the next few days, I encourage you to start noting where your decisions fall on the convenience vs rigor tradeoff. It’s an interesting thought exercise, and a good way to start to identifying common tradeoffs we see in our lives.

And on the subject of common tradeoffs, I’ve decided I’ll be blogging about common tradeoffs we find in our lives. Like the convenience-rigor tradeoff, each of these tradeoffs takes a similar form in many different situations. To end this post, I’ll leave you with a preliminary list of some of these tradeoffs. If there’s one you think is particularly interesting, shoot me a message via your preferred method and I’ll keep it in mind.

• Sustainability vs. well, everything - Making something “sustainable” requires us to hold resources back.

• Ease of Use vs. Capabilities - For tools we use, ease of use trades off with the amount of tings you can do with that tool.

• Novelty vs. Establishment - The newest and coolest things are also the least established, lending themselves to bugs and small mistakes.

• Short term vs. Long Term Safety - Long term safety requires growth, which in turn requires stomaching more short-term risk. Personal and organizational life is about striking a balance between the two.

• Custom vs. Standard - the more custom something is, the closer it fits to your need. But you sacrifice all the benefits of a standardized part/process/technique when it comes to maintenance.