Confessions of a Researcher: How I found happiness, time efficiency, and novelty through BLT

Image credit: bonappetit.com

Happiness, time, and novel work...pick two out of three if you’re a researcher, right? Maybe not. There IS a way to get all three which works for me. And, while I can’t guarantee it will work for you too, I’m hoping you and I are similar enough that it just might.

What is this ‘magic bullet’? It’s pretty simple--I call it ‘blocked learning time’. Blocked learning time (soon to be called BLT throughout for brevity and to make you hungry) is a strategy where one sets aside weekly ‘blocks’ of time solely dedicated to learning. By dedicated, I do mean dedicated. Ideally, you are not running experiments during your BLT, you are not analyzing data during BLT, you are not doing a listless literature review, and you definitely are not replying to emails. Your only job during BLT is to read about new fields/topics, reinforce concepts, and think about connections from the new field to your current one.

In what follows, I’ll discuss three tangible benefits of BLT--happiness, time-efficiency, and research novelty. Further, I’ll provide concrete examples of each benefit drawn from my experiences as a computational researcher in the physical sciences.

Time Efficiency

In my view, the greatest benefit of the computer science community is its commitment to making material open-source. Others have written great codes and packaged them in an easy-to-use fashion which we can then use to perform our own tasks more quickly. Such a collection of code is known as a library. There are two common ways to write code using a library. The first way is to start with little knowledge of the library, write a heavily ‘buggy’ script, run the script, and solve each bug by copying code from Professor Google. I call this strategy the “greedy” method. The second way is using BLT to spend time upfront learning the nuances of the library, then write an efficient script with minor bugs, and solve each of the smaller bugs using the knowledge-base you’ve built up during your handy-dandy BLT.

I navigated this exact cross-roads with a library known as “Tensorflow”. Tensorflow is especially useful for building complex neural networks (NNs). In my case, I wanted to build a special NN known as a Variational Autoencoder (ooooh fancy!). I started out with the greedy method, looking online for code snippets that I could copy, paste, and use. However, the code I found online lacked some custom features I needed. To add these features into the copied code, I spent hours searching for solutions, implementing them, seeing them fail, debugging, failing, and repeating. Eventually, I wisened up and found this free online Stanford course teaching the nuts and bolts of Tensorflow. After going through a few lectures during BLT, I was able to efficiently write my Variational Autoencoder along with my desired custom features and quickly debugged it using my brain instead of Stack Overflow.

Novelty

Now that I’m more time-efficient, I have extra time to educate myself on relevant fields. My field is Computational Polymer Science. That is, I use computers to study polymers (i.e. the things that make up plastic) before we do $$$$ experiments on them in the lab. Some time ago, I read about advances in a closely-related field, Computational Drug Discovery. There, similar work is done to study drug-like molecules, which are conceptually similar to polymers (remember, at a microscopic level, polymers are just a bunch of molecules chemically bonded together in “conga line”-like fashion). Exploiting this connection, I was able to adapt cutting-edge work in the Drug Discovery domain to be cutting-edge work in my domain. This work is helping us discover new polymers that power the energy-storage devices of tomorrow. I’m not the most brilliant researcher in the field (obviously...I’m the guy who didn’t know how to code a Variational Autoencoder until 3 seconds ago!), but I am the one who made the effort to connect Drug Discovery methods to Computational Polymer Science, therefore I get to reap the benefits.

Happiness

Boosting time-efficiency and the novelty of my work has itself made my work-life happier. This speaks to the positive synergies that implementation of BLT has had in my life. Outside of work, being time-efficient gives me more time to binge Netflix Limited Series.

Conclusion

That’s BLT, and its benefits, in a bun (Zing!). If I may offer one last bit of advice, flexibility is ok! If you have a deadline one week or you want to take a week-long trip to Panama City Beach (although, as much as it pains this Native Floridian to say, I wouldn’t recommend it with COVID) and can’t dedicate BLT... it’s no biggie! But if too many weeks go by without BLT, it’s time to re-evaluate and take another bite.

I mentioned Computational Drug Discovery and Computational Polymer Science as separate but complementary fields. Comment some other good analogies, relevant to your field, below!