I faced a similar problem in academia, but it was in economics. I started my undergrad career as a computer science major, later ending up in economics, and I w

I faced a similar problem in academia, but it was in economics. I started my undergrad career as a computer science major, later ending up in economics, and I went on for my PhD in economics after finishing my BA. For those that aren't familiar, a PhD in economics is a lot like a PhD in mathematics. If you don't already respect economists mathematical ability, you really should, but that's not my point here because math is just a tool, and contains no absolute truths for the social scientist. In an effort to combine solid theoretical research with well built empirical models, I learned how to program Python to scrape data from web sites. Since nobody around me knew anything about computer science, I found myself having to teach myself everything when it came to "how would I write a program that collects X things from Y products in Z Internet markets". In the beginning, when I was learning to use Python libraries for doing HTTP requests (eventually converging to using 'requests'), how to parse that HTML (started with BeautifulSoup and then converged to 'lxml'), and how to aggregate and analyze that data (eventually converging to 'pandas'), I had to spend a lot of time learning the ins and outs of Python. To this day, I still think it was a great choice because Python has changed the way I approach any kind of business question that could be answered with a well defined empirical model. As a research assistant, I would spend 12 hours writing scripts to automate the collection and analysis of data for some project I discussed with a professor. The day after writing that script, I would go and talk to the professor and want to discuss some of the computational issues with the script (say, encoding issues, or even the use of computers on EC2 to help collect massive amounts of data every day), but the economics professors would not care at all to hear about any of that. "You need to be studying the economics, not the computer science" they'd tell me. This would infuriate me, because in my mind, if you wanted to be able to ask all these interesting questions that rely on the data, then you need to spend time to make sure you've done the computation correctly and in a way that is reliable. Maybe I dug my own grave by showing my excitement as I began to become more fluent with Python, and thus more confident in my abilities. Nonetheless, I eventually decided to leave my PhD program because my interests in topics that lie at the intersection of computer science and economics were a bad fit for my program. It was so disheartening to me because I truly loved economics, and was confident that what I intended to do with my PhD research was going to be unique and lay a framework for the field of industrial organization. However, the experience I received in my program was the lack of an adviser that could truly help me achieve what I wanted to do, and a regular negative response to anything I'd bring up that was out of the realm of what my professors were used to talking to graduate students about. Overall, as my tone may signal, graduate school has left me very, very bitter.