How I Track and Actually Learn New Things

Stevie Chancellor
5 min readSep 5, 2019
Stacks in a library, with many shelves of books
So many books, so little time! Credit: University of Illinois Library

When I was a new PhD student, one of the skills I struggled with most was absorbing lots of information from many sources. One of the major differences between grad school from undergrad is the sheer volume of facts, formulas, and theory you need to understand to become an expert in your research and field in general (coming from papers, textbooks, quals/comps, class, seminars, advising meetings, just to name a few).

How do you manage all of these sources and actually learn it? What about remembering the techniques and approaches that just came out in the field? And retaining it later, even two or three years down the road? Without the scaffolding of a formal classroom to guide you, this can be pretty tough!

I started out by trying to read and with my citation manager and looking concepts up as needed, but the content never really stuck. I had trouble recalling facts and concepts, especially when I was self-guiding my own learning, even from things I had read the month before.

In the 2nd year of my PhD, I adopted a technique to help manage the information PhD students should be absorbing. It’s a simple, structured way to begin processing and retaining. It takes 15–20 minutes a day and a notebook. It’s been exceptionally useful for me in conquering the details and becoming an expert in my field of study. The people I’ve pitched this to and who have started it have also found it powerful, and I hope you find it useful as well.

The Notebook Technique

Grab a new and different notebook — it should be separate from your other notebooks. Skip 3–5 pages at the beginning for a table of contents. Don’t overthink the setup/ruled vs dot vs blank — I like something big enough to take long-form notes. Optional step: name your notebook with a pithy name. Mine is the “I Don’t Knotebook”, or the “IDKbook” (I love word play, guilty as charged). You’ll also need an index card/small slip of paper that will stay with the notebook.

When a new concept appears while reading, studying, rereading lecture notes, in advising meetings, jot it down. Often times, you’ll run into things that you don’t have time to look up right now. Write the idea onto the index card that bookmarks the next blank page. This is a list of things you don’t know and can refer back to later.

Here’s an example of my index card where I add new ideas to the pile — sometimes I’ll get five new ideas in a row, other days, it’s waiting until something new comes from my reading.

An index card with a list of items in statistics to learn. Many items are crossed off.
This index card is both bookmark and memory device for the notebook.

Then, either at that same time or later, take about 15–20 minutes to look up the information and write down what you’ve learned, using whatever resources to teach yourself. The resources may be reading websites and online tutorials/explanations, textbooks in your office, watching YouTube or MOOC videos, reading posts in forums or on Quora, or summarizing a verbal explanation from someone else.

Use these resources until you have a basic understanding of the topic. Later, go back and log the entry into the table of contents along with a page number.

A notebook with three entries about reinforcement learning and Poisson regressions
Example pages from the notebook — in this section, I covered two topics in the same day.

Here’s an example entry for a few topics — I was working on selecting different kinds of dimensionality reduction, reinforcement learning, and clarifying my understanding of Poisson regressions.

On days when you not actively pulling content from your “live” research (like a paper or class), look to the list on the index card for things to learn — that’s the new topic for the day. This can be done with theory, statistics, experimental design, concept mapping — need to understand it? It can fit in the notebook.

I personally set a goal for adding something to the notebook 3 of 5 days each week. That way, I’m constantly engaging with new ideas without making it feel like a chore.

The Process

This general idea is powerfully simple: put a little structure around looking up unfamiliar information and make a good-faith attempt to understand it. By filling out a short notebook entry, it forces you to pinpoint your gaps in knowledge (which everyone has), and guide self-learning. Writing it down formally means you have to try to communicate that in a way that makes some sense.

Personally, I like filling out the notebook — each page is something new I had learned that day, which can be very motivating. On days where I’m feeling down, the notebook is one of my favorite ways to give tangible, visible progress of my learning and growth as a student. Entry by entry, the knowledge stacks up over time in the notebook — in a career where the gains can be difficult to see, it feels nice to point to my notebook and say “here are things I did and learned.”

It also forced me to be willing to revisit concepts I didn’t totally understand. For example, I wanted to develop an intuitive sense of precision/recall and false positives/negatives in machine learning. I revisited this topic several times in the notebook until I could intuitively and easily describe with examples. Don’t be afraid to revisit difficult topics and put it back on the index card — this is a judgment-free and private notebook to support your learning.

Finally, the system makes it trivial to “seed” new ideas that sound cool to learn, and guide self-learning. Just heard about attention in RNNs? Great! How about label theory from sociology? Onto the index card it goes, to learn about later.

I implemented this technique when I was studying for my qualifying exams my 2nd year and kicked it into high gear in my 3rd and 4th year. If done well, this can really step up your game in understanding concepts from your field and give you a tangible, visible indication of your growth through the PhD.

📝 Read this story later in Journal.

🍎 Wake up every Sunday morning to the week’s most noteworthy stories in Wellness waiting in your inbox. Read the Noteworthy in Wellness newsletter.

--

--

Stevie Chancellor

Professor at Minnesota CS, Georgia Tech PhD. Human-centered machine learning, work/life balance, and productivity. @snchancellor on Twitter