Trying to Improve the New York Times’ Newest Brain Teaser

Why Digits?

I love games. My family has played every board game sold in the Boston area, but as we have all grown older, we have found less and less time to sit down for a three-hour game of Catan. In their place, we have taken up the New York Times' bite-sized brain teasers, which can be enjoyed at breakfast, on the train, or during a break from work. In particular, I enjoy the Mini (crossword) and the Wordle, as they get my brain thinking and provide satisfaction and fulfillment upon completion. 

The New York Times recently released a number game called Digits. Although it uses numbers instead of words or facts like Wordle and the Mini, I was intrigued. However, after playing for a few days, I did not feel the same sense of accomplishment and satisfaction that I did with the other games. I could have just dropped the game and gone back to my usual routine of playing two games a day, but as a designer, I saw this as a problem that I could solve.

Research and Discovery

It turns out I wasn’t the only one who felt this way about Digits. In this article explaining the creation, there are close to 200 comments, many of which echoed my concerns and criticisms.

Equipped with this information, I set out to collect as much data as possible from different user groups that I had personal access to: Students (from elementary school to college level), Parents, Teachers, and Math Tutors. To supplement this, I had the comments of a diverse range of NYT game-players.

Pain Points

Difficulty:

The most commented pain point that NYT gamers had was the level of difficulty. Comment after comment, (and the most upvoted ones at that) mentioned its ease.

“Why is this game being dumbed down every day??  When I started playing many weeks ago, I had to progressively work harder with each puzzle. Now, there is almost zero challenge from beginning to end. Please raise your standards.”

-Cheryl

“....But, there is no challenge to this game as it is far too simple. The player knows the answer and it is merely a matter of juggling the given numbers until you get there. I would suggest adding a timer. A chart showing where your time fits against other players would add a bit of adrenaline to the game.”

-Vic

One of the amazing things with NYT subscribers is their desire for more-whether it be knowledge, or improving a game. Already, suggestions were being provided on how to make Digits better. These suggestions provided inspiration for my Digits redesign to keep users engaged and excited to play. The question I continued to ask myself was: How could I make Digits a part of these users' morning routine?

General User Experience:

As Digits is still in beta, there are some general UI issues, but these seem to be much smaller and niche for users. However, this was a good reminder to check accessibility at the end of this project. 

“I’m not getting it either! The interface isn’t intuitive.

My submit button remains ghosted.”

-Berman

“As for the color palette of Digits? Back to the drawing board, I say.”

-BM

Lack of Division:

Both the math tutors and higher education students I talked to mentioned one seemingly glaring admission. The comments of the NYT concurred: 

“So far I've reached 3 stars on each of the rounds, but only used division once. And of the revealed solutions I've seen, only one has included division (it's possible I might have missed some, but not many.)”

-CRTH

Based on my personal experience, I have not used division once in over 16 plays. Having one of the four major operations be useless for users is an incredible issue. Fortunately, after some brainstorming (coffee!), and a lot of crumpled paper, I found a way to combine this issue with the lack of difficulty most people were finding.

The Design

Instead of working from screenshots, I emulated the entire Digits UI in Adobe XD, creating a panel for each major moment in the user journey. This allowed me to create a more accurate representation of the user experience and to test different design options more easily.

The Color

I decided to replicate the original color scheme and typography in XD, as most users did not express any dissatisfaction with them. Furthermore, in all of my in-person or Zoom interviews, no interviewees mentioned them.

As a designer, I am committed to creating products that are accessible to everyone. After completing the redesign of Digits, I carefully checked every panel for accessibility. I believe that design is only effective if it solves problems for people of all demographics. By making sure Digits remains accessible, we are ensuring that everyone has the opportunity to use and benefit from this product.

Improving Digits: Difficulty

Users found the problems to be too simple, as they were able to achieve a perfect score on all of them. In fact, many of the high school and college students did not even know what the stars were for, as they were able to achieve the full three stars every time.

When consulted, math tutors suggested that using larger numbers would make mental math slightly more challenging. However, the NYT should keep Digits accessible to a wide range of people, from elementary school students to retirees.

To do this, I created a scale that grew exponentially. Instead of having

Problem 1: 0-100

Problem 2: 100-200

Problem 3: 200-300

Problem 4: 300-400

Problem 5 <500

My research suggested that this scale might be better:

Problem 1: 0-100

Problem 2: 100-200

Problem 3: 200-400

Problem 4: 500-800

Problem 5: 800-1000

This approach would allow at least three, if not four problems, to remain within the same range as the current game. But it would also give people more difficulty with the larger numbers in problems 4 and 5. This approach would also be able to solve two problems at once--solving difficulty issues, as well as Division.

Improving Digits: Adding Division

Higher numbers provide more opportunities for division. When I conducted a small survey using larger numbers, participants often attempted, or even succeeded, in using division to solve their answers.

Improving Digits: Tracking Progress

For me, as well as many other NYT mini-game players, satisfaction can be derived from getting the Wordle in 2 guesses (due to skill…totally not luck), or completing the Mini in 30 seconds. Creating a way to track how well you did, both against yourself and potentially against others, was one of the main reasons I got into this project. Multiple users in the NYT comments agreed with me:

“Digits is fun! Can we get more/less points based on the number of calculations used to solve? That would make it more fun, imho.”

-Bryn

“I would like to see the scoring accumulate over time similar to Wordle. It would also be interesting to obtain a comparison between the ideal (least number of operations) solution and the one that that user provides.”

-Ron

I experimented with many different benchmarks, with varied success: time, number of operations per problem, locked levels, number of operations total, total stars.

Time:

The time it takes to calculate something is an interesting metric, but it can be skewed by distractions. However, the main reason I did not like it was because a graph of time would have an ever-growing X or Y axis. The "100%" bar could be 5, 10, or 100 minutes, which would make the data difficult to read as the longest time grew. The time-based benchmarking feature in Digits was an interesting experiment, but ultimately it provided an unfriendly user experience.

Operations

The statistic that I found most interesting was the number of operations required to solve each problem. This statistic was also mentioned by math tutors and in the New York Times comments. The number of operations required to solve a problem encourages users to find the simplest and most direct solution. Unlike the time it takes to solve a problem, the number of operations can never exceed 5, due to the fact that there are only 6 numbers given for each level of Digits. This makes the data easy to graph, and the final product looks and feels similar to the results of a Wordle puzzle.

The number of operations is not a fully accurate measure of how well you performed on a level, as it is possible to achieve one star with just one operation, while two operations would be required to get the exact correct answer (as a purist like myself would strive to do).

Stars

This issue led me to the idea of a dual-benchmark. One tab would track operations, and the other would track stars per problem. Users would be able to see how often they got the exact right answer for each problem. Just as operations reward quick precision, stars per problem rewards straight accuracy.