Let's set the scene. I'm from Texas, and for the first time ever, I'm interning for a summer in a brand new location, Chicago. I'm also a foodie, so my goal is to be able to eat as much good food as possible, but I can't do that without first discovering the best places to eat. Of course, once I find the best places to eat, I shouldn't be venturing out to new restaurants, only to have an inferior meal. The challenge is simple, then: how do I know when to explore the area around me for a new restaurant versus exploiting the information I already have?
If we were able to quantify the utility from previous choices, and the expected utility from future choices, and how much a good meal tomorrow is worth in terms of a good meal today, then this would be fairly straightforward, using a variety of explore/exploit frameworks that the field of computer science has developed. We might look into the Gittins Index, which is well suited to just this type of quantitative analysis. But our tastes change, it's hard to quantify just how good a restaurant is, and doing so would be quite a bit of work, so we need a more "gut-feel" solution.
Jeff Bezos, when he started Amazon, said that all he thought about was how much he'd regret not doing it. He termed it a "regret minimization framework," and it's a perfectly valid way that we as humans can solve the explore/exploit problem. Let's simply do whatever we would most regret not doing. We could spin this into a similar (though not quite identical) formulation: do whatever you think will end up turning out the best, a framework which others called the Upper Confidence Bound.
The interesting thing, though, is that there's so much more that you can think of as an explore/exploit problem than just restaurants. I've approached the search for a full-time position much in the same way. With a career, there are certain factors that feed into job satisfaction: location, coworkers, salary and benefits, work/life balance, promotional opportunities, the work itself, etc. An internship is a great way to learn as much as possible about how those factors look, and how much you might personally weigh those factors.
But that's just for one particular job. If the goal is to not just get a job, but to get the right job, you need to explore.
I'm still in college, with another three years to go. That's a fair bit of time. Nevertheless, with my current internship and the two that come afterwards, I have three questions I need to answer:
Which industry do I want to work in?
Where do I want to be?
How much I do weigh one factor over another?
I'm currently enjoying a really amazing internship at Akuna Capital. The salary is amazing, Chicago is great, the work is stimulating… there's nothing that I could complain about, other than the impact that my work has. And if I were to pick one job to do for the rest of my life, I'd probably pick trading at Akuna. It's unlikely that I could find anything that really beats out working with Australians.
But I still want to explore some more. I don't yet know how I feel about data science or consulting - while some of the niceties might not be as ideal, it's unlikely that I find those jobs to be worse across-the-board. Even if I end up deciding to stick with trading, I'll only have had more confidence in my decision.
P.S. I was first exposed to the explore/exploit framework in a book, Algorithms to Live By. The book is one of my all-time favorites, one of the few that I recommend without reservation. It's by Brian Christian and Tom Griffiths.