My Ultimate Learning Tool: A Mysterious Wooden Box?

A simple possession that has propelled my long-term education.

Riley Howsden
8 min readMar 2, 2022

I don’t keep many things. As much as I would love this to be a statement about minimalism, it is more a result of numerous relocations. After moving to California, I would share with others that this was the eighth state I had lived in in the last ten years. If I did a full sweep of my residence, I do not doubt that I could tally the number of items I have owned for over a decade on a single hand. That list would include my original Nintendo (NES), an acoustic guitar, and this strange wooden box.

Why have I kept this box?

The box itself is not a small item, coming in slightly under two feet long and weighing approximately five pounds. Despite the size, I have lugged this box around to many locations, including workplaces and most vacations. At first glance, one might wager that the box is a sentimental item, a rather ugly one, from a distant or deceased relative. However, it was constructed circa 2009 by my brother-in-law and me after discussing a few prototypes. Since its creation, it has been worth its weight in gold.

What makes this box so valuable?

On the backside of the box, there are five drilled holes. A thin lip is present on the bottom of the front side, which one can flip open. The top of the box has a much larger panel that opens, revealing five separate compartments. The wood itself looks rough, worn, and appears to have been painted green in some areas and stained in others. It would be hard to determine how I use the box from this description alone; I’m leaving out the most crucial detail — what the container holds.

What am I storing in this box?

When others would see me with the box, some would venture to ask what it was. In these instances, I would frequently comment that it was my “spider box” and that I collected a bunch of eight-legged creatures. When asked if I was joking, I would insist that the person select one of the circular openings on the back, stick their finger in, and pet one. I would add some colorful commentary such as “Don’t worry, they don’t bite often, and as far as I know, only one of them is poisonous, so your chances of survival are decent.” After this comment, I often get crazed looks, but I eventually reveal to the inquirer what the box contains; note cards.

A Notecard System for Long-Term Learning

The concept behind this box came together as I started to prepare for the Physics GRE, a standardized test commonly used to assess one’s potential as a graduate student in the field of, you guessed it, Physics. I had taken the exam before and ended up with an average score and knew that I would need to perform at least in the top 10–20% for my application to gain traction at the best universities. I’ll be the first to admit that my study habits at the time weren’t ideal; for most of the tests I had taken in college, the material was limited and could be stored in memory long enough to pass. Of course, the Physics GRE was a whole different beast, containing an extensive set of concepts from numerous classes. It just was not possible for me to store all of that information in short-term memory — I was going to need a less greedy way of comprehension, one that could retain knowledge for years instead of weeks.

Notecard Learning Theories

At this point, I had started to consider note cards as the best tool for sustaining long-term memory. I was pretty skeptical, though — I had seen people with notecards on a single keyring and wondered if the order of the cards, instead of the content behind them, was the core of the memorization. The best comparison might be recalling song lyrics, which the brain strings together in order; if we asked the question, “does this word appear in this song?” that might be a more challenging question than singing the entire song altogether! My first theory on notecards was that a shuffling mechanism was needed to avoid ordered memorization — the keyring was out.

Of course, not everyone was using a keyring to study notecards. Another option was a stand-alone box, where the cards were free-flowing and did not have a specified ordering. After going through a set of note cards, one could shuffle them and start from the top, bypassing the issue with ordered memorization. While this was an improvement, something still bothered me about the structure. If some of the note cards were trivial and others were challenging, one would spend an equal amount on both when they needed to focus on the problematic cards. One solution might be to discard notecards as one grows confident in retaining the embedded knowledge. Still, this manual mechanism seemed harder to manage and, at scale, appeared to be a nightmare. My second theory was born; an optimal notecard system needed an automatic filtering process.

Armed with these two theories, “The Box” was born. The concept behind how it works is relatively simple — there are five separate areas in the box, each with the ability to store approximately 300 notecards. When I create a new notecard, it gets placed in the left-most area, and as I answer notecards correctly, they move one space to the right. An opposite effect also occurs; if I get a notecard wrong, it moves to the left. Once I gain comprehension of a notecard, it will eventually hit the end of the box, where it is “retired,” at least temporarily, to a static storage spot, sorted by category. In the future, when I feel the need to revisit a topic, I will often reinject cards back into the system to ensure that I have retained the information.

Spaced Repetition

Although I did not know it at the time, the system I had created was already in existence, an evidence-based learning technique known as spaced repetition. Experts have proven that this method would increase the learning rate of information. I had discovered this only a few years ago when discussing my process with a coworker. They referred me to a Wikipedia article on the Leitner system, which looked surprisingly familiar.

While there are a few common structures for spaced repetition, mine more closely followed the expanding structure; since the newly added and the most difficult notecards remained on the far left, this was the stack I was referencing and studying the most. The time interval between drawing a notecard stack from another section grows when moving to the right. Since I try to review one stack of notecards each day, the rough distribution across a month spent on each section, from left to right: 16 days (52%), 8 days (26%), 4 days (13%), 2 days (6%), 1 day (3%). With each study session, I would use the holes on the back of the box to push out approximately 40–50 cards from the bottom of each stack through the open front lip. Depending on my answers, the cards would return to the top of a specific pile. Using this structure, I can review around 1400 notecards each month.

However, this doesn’t directly translate into the number of notecards I can retire each month. In a best-case scenario, answering notecards correctly every time, I would be retiring about one-fifth of 1400 (since I need to answer each card correctly five times in a row). In reality, that number is halved, and somewhere around 125 cards are retired each month (~1500 each year). Since the box holds 1500 cards, it is not unusual for a single notecard to live in the box for over a year after its creation.

Inventory Organization

When I first began building up my notecard collection, I quickly noticed that certain long-lasting notecards were making inventory management difficult. One might assume that notecards added simultaneously on a single topic would exit the box somewhat in sync. I could then bundle up those cards and store them in a similar location. However, this was far from the truth, and I needed a better way to categorize and track cards as they went through the system. Soon after, I began to label each of my notecards with more information. For notecards that originated from books, I started using the notation below:

The letters are a simple way to identify the book the information originated from, and the numbers reference the page on which I could find the content if needed. If the source was something other than a book, such as Wikipedia, a broader categorization such as “Wiki/Statistics” ensures that I can adequately group topics as they exit. Although less formal, another axis I use to categorize cards is color; those derived from the same source will have the same color. While this isn’t necessarily useful for storing the cards, it is valuable while actively studying them. For example, different concepts across fields often use the same terminology, such as “consistency.” However, if the card is purple instead of blue, I can identify that it comes from a specific computer science reference instead of statistics.

Physical vs. Digital

I should make one final remark; why do I use a physical system for learning when digital products exist, such as the impressive Anki software? The irrational component is that I used this physical system for years before discovering Anki. Perhaps a sunk cost, but I am biased towards the investment I have placed in the notecards. The second and more rational component is the unseen magical power of physically writing something down. With the ease of Anki, I would find myself creating more notecards than I needed on a given topic. While I could spit out notecards left and right, I had diluted my overall learning. With each physical notecard I write, I carefully consider how crucial capturing that information for the long-term is; the barrier to entry is much higher, and therefore, the quality of the average notecard increases. There are also findings that handwriting is associated with more robust neural encoding and memory retrieval.

Final Thoughts

Across my career, I have used this system for many things — from preparing for the Physics GRE to passing actuarial exams to dealing with the explosion of knowledge within the statistical and machine learning spaces. If you are seriously interested in developing better memory retrieval on what you learn, I highly recommend investing in a system like this. Anki is an excellent program to get started, and it is also free, and who knows, maybe a few years from now, you’ll be the crazy person with a “spider box.”



Riley Howsden

“Half-Stack” machine learning propagandist in the gaming industry — a critic of all data presented deceitfully, unless it contains a meme, then it must be true.