I am obsessed with writing notes. I document a lot of my personal and professional life through notes. If I come up with a new research idea, I write a note about it. If I come across a relevant paper, I write about it in my notes. If I find a recipe I like and use frequently, then I add it to my notes so I don’t have to search the web to find it again. Every time I read a new book, I write a note about it, which includes any quotes I liked and my review of the overall book. For example, I just finished reading Digital Minimalism by Cal Newport; here’s a quote from my notes: “A foundational theme in digital minimalism is that new technology when used with care and intention creates a better life than either Luddism or mindless adoption.”  I love to play disc golf, and I track all of my past plays using my favorite note taking app, Evernote.
Since I am an avid Evernote user, I share all of my examples exclusively in Evernote. If you are looking for a new note taking app, I share some helpful suggestions on how to pick the best note taking app for you in my previous post.
The Key Idea: Write Everything Down
The main theme behind this post is that for anything meaningful that we encounter in our work, we should be writing it down. For example, with an open-ended discipline like computer science research, we never know which notes will lead to new scientific breakthroughs. If we don’t write ideas down as we have them, then we risk losing them and in turn potential innovations and breakthroughs. For these reasons, I recommend writing everything down as research notes.
Although I focus on computer science research in this post, these ideas can apply to any part of our lives that we deem meaningful. For example, I write notes for all of my hobbies in Evernote. In fact, I often learn new note taking skills through the process of writing personal notes, which I then apply to my research notes. Writing good notes is a transferrable skill that I can use in every part of my life.
An Example Note
Here’s an example of a note that I use every single quarter, and it is just 5 lines of content in Evernote:
This note explains how to create a Python3 virtual environment. I have written a templated example on the first line, followed by a concrete example that I frequently use, and finally a link to the original source of the information, in case I want to review the material. In this case, the source is a page from the official Python documentation. Technically, I could easily look this up online each time I want to create a new Python3 virtual environment rather than writing a note in Evernote, but this way if I don’t have internet access I can still look up the answer in my own notes. You might be wondering “don’t you eventually just memorize how to do this?” Sort of, but I never remember the exact details. Rather than guessing how to do it, or relying on my (often faulty) memory, it’s easier to write it down once and look it up later.
The above note is an example of documenting one of the processes I perform as I conduct my research. By knowing how I intend to use a note, I can structure it accordingly to maximize my ability to reuse it in the future. In this case, by documenting a step by step process for creating a Python 3 virtual environment, I make it easier for myself to create virtual environments later on. Also, if a student asks me how to create Python3 virtual environments, rather than explaining or demonstrating the steps for every student individually, I can just share this one note in Evernote by turning on public sharing for it and sending each student the sharing link.
Writing Notes for Ideas, Processes, References, and Results
However, not every note should be written as a step by step process, depending on what the note represents. I group research notes into four different categories:
- Ideas: new research questions or ideas I might want to explore in the future.
- Processes: a detailed, step by step guide to completing a specific procedure that I performed in the past.
- References: Bits of information or pointers to information sources that may be useful in the future.
- Results: A description of my findings after completing a specific research process, ranging from 1 paragraph to multiple pages of content.
I think of Ideas posts as being “before” posts, or notes recording my thoughts before I conduct the research. Process notes are what I record while conducting the research, and results notes are what I document after conducting the research. Before, during, after. In this way, I can cover all aspects of the research process. Reference notes I think of as just-in-time notes, for specific situations that come up frequently but not with obvious regularity. To get you started on writing great ideas notes, I’ve created a template anyone can use, available here. I’ve also made an infographic summarizing the four note types, for easy reference!
Writing Process Notes: Another Example
In the past, I wrote abbreviated versions of process notes, assuming I could “fill in the blanks” with my memory and past experiences. However, I have found that I often revisit process notes months or even years after I originally wrote them. By then, I have completely forgotten the original context in which I wrote the note, making the abbreviated version useless. My most vivid memories of these kinds of debacles involved installing the database management system SciDB. I had to re-install SciDB four times throughout my PhD, basically once per year. These were in the very early days of the software, when the installation process was difficult and largely undocumented. The first time, I didn’t bother taking any notes. Installing SciDB took several hours! Unfortunately for me, when I had to install it again months later to upgrade the software, I knew I was in for another multi-hour installation session, but this time I actually took notes. They weren’t perfect, but they had a lot of useful information. Each time after that, I refined my notes. Near the end of my PhD, when I installed SciDB on the MIT servers for the last time, I was able to complete the installation in under half an hour. Here’s a snippet of my SciDB note, to give you another example of a process note:
It’s pretty hard to remember exactly how to do something you only do once a year, especially if it’s super complicated. For example, it’s not like people just stop using TurboTax because they’ve been doing their taxes for years.
Writing Great Process Notes: Some Tips
The key to writing great process notes is this: write down every step such that someone who is not you could follow them and successfully complete the procedure you are writing about. Even when you write notes only for yourself, you do not stay the same person year after year. An additional tip is to write process notes as if you are writing them for someone who has only baseline background knowledge, such as someone who is completely new to your research area. In this way, you ensure your notes are usable by anyone who might join your research project in the future.
When you use someone else’s tutorial to complete the steps, then make sure you add a link to the tutorial in your notes. That way, if you revisit the note and don’t remember a step, you can always return to the original source. I link back to the original sources in both of my process example above. However, a risk is that the tutorial or information page will eventually disappear, which has happened to me before. I recommend recording all of the important steps directly in your note in addition to the link. Sometimes I will paste steps directly from a tutorial page into my notes, such as for cooking recipes (not research, I know, but still useful!).
This goes without saying hopefully but write the steps in the order in which they should be performed, ideally using a bulleted list, or at least headers saying “step 1”, “step 2”, and so on. You see in my SciDB installation note that I use numbered lists to track steps. If order doesn’t matter, then mention this in the note.
Another tip is to practice searching your own notes before searching the web. The next time you go to look something up using your favorite search engine, just see what happens when you execute that search in your note taking app. For procedures you regularly perform, if you can’t find the answers in your own notes, then you should probably write a new one or update existing notes. Or if the answer changes and your notes go out of date, then you need to update them so they stay relevant. Developing a regular practice of using and reusing your own notes should help with note maintenance.
Writing Ideas Notes: an Example
Unlike my process notes, my ideas notes are pretty unstructured. Sometimes the ideas notes are just one line, like in this example:
But these days I try to write ideas notes as self-contained thoughts, where I include the motivation, context, and if relevant the original inspiration for the idea, for example if a certain conversation or paper led to the idea. I prefer to write one note per idea, but this is not a strict rule. Here’s another example, which I originally wrote back in 2016:
I was at a conference presenting some work I did during a research internship the previous summer, and I was having lunch with my intern mentor. I was lamenting how I thought I couldn’t write a paper on my thesis work that the HCI or visualization communities would care about. My mentor pushed me to brainstorm ideas that were related to my thesis work and was kind enough to share some of his own ideas during the lunch, and I wrote down the best idea that came from our discussion. As it turns out, the idea recorded here eventually became a published paper three years later. The published project would have never happened without that initial discussion, but likewise the resulting project idea would have never come to fruition if it wasn’t remembered afterward (by me or Evernote).
Writing Great Ideas Notes: Some Tips
With ideas notes, the most important thing is to write down the idea as soon as possible so you don’t forget it. Add as many details as you have time for, but make sure you write every idea down, even the bad ones. If you are brainstorming and come up with multiple ideas, then feel free to list them in the same note. Here’s an example list from when I was brainstorming domain names for my blog site:
The list was so long that I struggled to get the entire note to fit into a single image! The name Bits in Bites doesn’t appear until near the bottom of the list, which is unfortunately off-screen in the image. As you can see, I came up with a lot of bad ideas 🙂 But what is key in these examples is that I didn’t limit myself when creating my ideas notes. Write as much as you can while the context is still fresh. Give your inner editor a break so your inner creative can shine through and generate lots of ideas and lots of details for those ideas, regardless of quality.
Writing Reference Notes: an Example
Reference notes are probably the most commonly written note in computer science research. You find an interesting bit of information and you want to save it for later, so you write a short note about it. I generally keep them fairly simple. I write a descriptive title about what the resource is, and then add a link to the original source. Here are some examples of reference notes I’ve written in the past:
I also keep reference notes for logistical things, such as my National Science Foundation (NSF) ID for submitting grant proposals, a link to the brand and logo information for my university, or a short description of my dietary restrictions for food orders for easy access.
Writing Great Reference Notes: Some Tips
Honestly, I don’t write a lot of reference notes. I am selective in what reference information I include in my notes. Why? Because reference notes are generally not connected to any specific research idea, process, or project, so they tend to be under-connected and therefore underutilized within my own information system. My general rule of thumb is to only add a reference note for content you have already had to look up multiple times. By the second time, you know that you have needed this information twice, and thus may need it again in the future. Here are some examples:
- Professors in computer science periodically submit grant proposals to the NSF to raise funding for their research labs, myself included, so keeping a reference note on hand about one’s NSF submission ID makes sense.
- My lab runs a lot of user studies, which we use to collect data to answer specific questions about human analysis behavior. We often run multiple statistical tests on the collected study data to determine the statistical significance of our findings. When you run multiple statistical tests over a dataset and measure significance using p-values, it is good practice to adjust the p-values based on how many times you reused the same dataset to conduct the tests, hence the reference above to Bonferroni correction.
- I regularly present my work to other research labs and research groups, who generally ask for information about my presentation so they can advertise it within their organization. I often give the same presentation many times, so I keep reference information on hand that I can paste into an email or google doc for sharing purposes.
- When attending a conference, daylong retreat, networking event, and so on, food may ordered for attendees. I have strict dietary restrictions so it is nice to have an existing explanation on hand that I can just copy and paste into a registration form or email.
- When I read books, I like to save quotes about data, science, and knowledge. The last note in the gallery above is a snapshot of some of my favorite quotes so far.
However, there are some exceptions to this rule, since in academic research you never know when information will actually prove useful for a project. My typical exceptions are notes that I write when I learn from other researchers, such as when attending colloquium talks at my university, attending talks or chatting with fellow researchers at a conference, or writing notes on a paper I recently read. The third example above is a reference note for a paper I read on using GOMS models, a certain type of data-driven psychological model , to analyze website usability for people with visual impairments.
But again, these notes rarely connect to any particular idea or project in my information system, and as a result just end up floating around among my other notes with no established connections. For this reason, I try to only write reference notes for content I am likely to reference again in the future.
Writing Results Notes: an Example
Results notes are the most important note type in my information system. Basically any activity that stimulates my thinking on a particular research project or research idea I write down as a results note. For example, I use results notes to capture what I have learned from writing code, analyzing experiment results, and reading related papers for a given research project. Here are some examples:
As you can see, my results notes tend to be text heavy. They generally include a synthesis of my analysis findings, followed by my reasoning for why these findings are significant or interesting from a research perspective. The analysis findings themselves may be in the form of a visualization, such as in the first example, or even copy-pasted command line output, as in the fourth example. However, research is not relegated solely to running experiments and writing them up. In fact, a significant fraction of my research involves simply writing down my ideas and finding related research to bolster them. In the second results note example, I mention a paper I read at the time, and how the paper’s perspective on AI is interesting for visualization research as well. The third example is a snippet of an old draft of my doctoral dissertation. I often draft initial paper text in Evernote, then paste that text into a separate LaTeX editor to produce a shareable LaTeX project. This way, my paper text can stay close to my notes while I am still playing with nascent ideas, reducing context switching between applications. Later, when the ideas have matured more, then I may edit directly in a separate LaTeX editor, such as Overleaf or TexShop.
Writing Great Results Notes: Some Tips
I break results notes into two types: results from experiments, and results from thinking about my or others’ ideas. Both types are with respect to my current research project. I’ve written out separate steps that I recommend for each type below.
- Explain the motivations and goals behind conducting the experiment at the top.
- Describe your experiment as if someone will need to replicate it using your notes.
- If your experiments involve programming code, then add lots of comments to your code and include a detailed README.md file explaining how to run it.
- If possible, use virtual environments to setup the code, that way you can avoid most installation snafus in scenarios where you have to run the experiment on multiple or different computers.
- Write a summary of the analysis findings, so you don’t have to interpret the raw data directly in your notes.
- Include a reference to the raw data so you know where to find it. If it’s small enough, paste the raw data directly in a separate note and add an internal link to the results note.
- If performing this analysis for a paper, summarize your ideas into a concise paragraph or two that can be pasted into your paper.
For developing research ideas using existing sources:
- Explain the motivations and goals behind the analysis at the top.
- Describe the key sources that inform the current ideas and arguments. Include full citations, and even include quotes to ensure the source ideas and attributions are clear.
- Include links to PDFs of the key sources, for easy access later on.
- Write your own ideas and arguments in more detail, citing the key sources as needed.
- If performing this analysis for a paper, summarize your ideas into a concise paragraph or two that can be pasted into your paper.
Getting Started: A Template
To help everyone get started, I created a structured template for writing great ideas notes. The template is available here. A handy summary of the different note types is also available in the infographic I shared above (also available on pinterest).
My philosophy is that my brain will work better if I don’t have to memorize as much of my work. My brain is like a room, I can make the most of its space only if I keep it uncluttered. To this end, I offload the memory burden from my brain by writing a lot of notes as part of my research and even my daily life. However, it is not enough simply to dump my thoughts into random documents. Imagine what the library would be like if books were haphazardly placed on shelves, and if even the content within a book was random from one page to the next. No one would find the library particularly useful and would probably avoid going there, except perhaps to amuse themselves. The same goes for my notes. I try to keep my notes organized and detailed, so I know exactly what to look for and where to look when I need it.
In this post, I share four kinds of notes that I write as part of my daily work: ideas, processes, references, and results. I provide examples of each type of note from my own database of notes in Evernote, and practical tips for making the most of each type of note. In my next post, I’ll explain how I organize these notes into a powerful information network to boost my research productivity.
 Card, S.K., Moran, T.P. and Newell, A., 2018. The psychology of human-computer interaction. Crc Press.
 Newport, C., 2019. Digital minimalism: Choosing a focused life in a noisy world. Penguin.