My Daily Devotion: The NYT Daily Mini Crossword

23-Across: Three years and 997 crosswords later

Austin Steinhart
8 min readMay 16, 2021
My solve time for every cross I have completed the last three years by day of the week.

Thanks to my dear friend Zach, I have had the pleasure of diving into the world of crosswords over the last four years or so. While I originally imagined crosswords as a solitary exercise in being old and white (which is mostly true), I have come to enjoy them as a relaxing pastime that is thoroughly enhanced when done in the company of others.

In order to satisfy my obsession, improve my crossword skills, and share the joy of crosswords with others, I started solving the New York Times Daily Mini Crossword and racing against my friends to solve it the quickest back in 2018.

As a fun exercise to both explore a hobby deeper and to practice my data storytelling, I collected my solve times and frequency for the NYT Daily Mini Crossword from the first day I completed the puzzle, Monday, February 5th, 2018 (03m:26s) up until Thursday, May 6th, 2021 (0m:54s). Now to take a look at the data!

The Plan

  1. Share a high-level overview of the data.
  2. Explore my consistency in completing the Daily Mini.
  3. Explore streaks of completing or not completing the Daily Mini.
  4. Explore my solve time for the Daily Mini.

The first of a series of housekeeping notes: There are some days that I did not complete the crossword. The free version of NYT Crosswords only lets you complete the Daily Mini for the current day so if you miss it, you can no longer solve it. If you pay for a NYT Crossword subscription (here is a link for half off), you get access to the archive and are able to go back and complete the Daily Minis from previous days. I intentionally decided to not complete the crosswords I did not originally complete on the day they were published. This decision came with a tradeoff; more completed crosswords gives me richer solve time data but also erases a lot of good insight (the absence of data can be just as powerful as the presence of data). By noting the days that I did not complete the Daily Mini, I am able to explore how my completion rate and solve times have changed over time.

Overview

Let’s get started with some general observations. From February 2018 to May 2021, I solved the Daily Mini 997 times out of the 1188 days, giving me an 84% solve rate. I spent a total of exactly 26 hours, 59 minutes solving the Daily Mini during this time period.

The darker the blue, the slower my solve time. I did not attempt the Daily Mini on dates that are white.

My average solve time was 01m:36s (more nuance on this time later). My fastest solve time was 0m:10s on February 6th, 2019 and my slowest was 11m:20s on August 25th, 2018. My seven longest solve times all happened in my first year doing the Daily Minis so I think I just had to work out some kinks.

Consistency

More housekeeping: My use of the word consistency is intentional. My increase in completion rate is not because my skill increased (i.e. I couldn’t solve certain puzzles in the past) but rather because I am playing the Daily Mini more frequently.

Over the four years that I have been tracking my solve rate, my consistency has improved pretty dramatically. In 2018, I completed 72% of the possible puzzles and in 2021 so far, I have solved 90%.

Outside of the first month I started, I have completed at least half of the Daily Minis every month and since April 2019, I have not dipped below a 75% solve rate. I completed every Daily Mini in a total of five non-consecutive months, all near the end of 2019 which is also around the time that my solve rate peaked.

.

Each day had a relatively consistent solve percent except for Saturday and Sunday. I am not sure why there is this dip for these two days. Saturdays are 7x7 grids instead of 5x5 but I am not sure why that would impact my solve rate (more on the impact on solve time later). Getting lazy on the weekend!

Streaks

In addition to how often I completed the Daily Mini, I wanted to explore my streaks, meaning how many days in a row did I complete or not complete the Daily Mini. I’ll call the former an “active streak” and the latter an “inactive streak”.

Each bar represents the length of the current streak as of the date. Positive numbers represent active streaks and negative numbers represent inactive streaks.

My longest active streak was 105 consecutive days from October 26th, 2019 to February 7th, 2020. My second-longest active streak was 93 days from June 18th, 2019 to September 18th, 2019. (In the 36 days between these two streaks, I only missed 5 days. I was so close to a mega streak!) Despite having a few long streaks, it's pretty common for me to miss a day here and there. Most of my streaks are under 10 days (118 of 141) and my most common active streak is one day (31 times), followed by three days (29 times). Does one day count as a streak? Either way, the point still stands as my next most frequent streak was two days (16 times). (Does two days count as a streak?)

My longest inactive streak was seven days from February 10th, 2018 to February 16th, 2018; however, this inactive streak was a few days after my very first Daily Mini. My next longest inactive streak was four days from February 28th, 2018 to March 3rd, 2018, which was also during the first month that I started solving the Daily Mini. Let’s just say I was still getting into it the first month and let these slide. Outside of these 2 inactive streaks, I had 7 inactive streaks that were three days long, 27 that were only two days long, and 107 that were only one day long. I’d say I’ve been pretty consistent overall!

Solve Time

Now onto the good stuff. On the NYT Crossword app, there is a daily leaderboard in which you can add other people to see how quickly they solved the day’s Daily Mini. Since good ol’ competition is easily the most consistent way to get me invested in a game or activity, I was interested in exploring some different facets of my solve time.

Some more housekeeping:

  • First, I’ll note an inconsistency in the format of the Daily Mini: Sunday through Friday are 5x5 grids and Saturday is a 7x7 grid which creates a difference of at least 12 letters and at most 26. This has an effect on my solve time so I will be separating Saturday solve time from the rest of the week. I will use the NYT’s language and refer to the Sunday through Friday 5x5 grids as a “Mini” and the Saturday 7x7 grid as a “Midi.”
  • Second, there are four Daily Minis in which I took an abhorrent length of time to complete. Since all these times are over two minutes longer than my next slowest time, I decided to remove them and assume that these were times in which I walked away from my phone or got distracted (or let's be honest, maybe stuck). These four Daily Minis were all in 2018 and were all Saturday 7x7 grids.
An example of a Midi 7x7 grid on the left and a Mini 5x5 grid on the right.

My average solve time for the Minis is 01m:26s and for the Midis is 02m:43s. As we can see from the trend line on both graphs, my time has improved since I first began solving the Daily Mini and appears to be leveling out. Using a rolling average, in March 2018 my average time was roughly 02m:10s and in April 2021 it was roughly 01m:22s. I have been able to cut off a whole minute on my average time!

Looking at this data, this feels correct. I felt improvement as I continued to get more comfortable doing crosswords and believe that my solve time is more dependent on the specific puzzle rather than my skill at this point.

Using solve time as a proxy for difficulty, there doesn't seem to be a noticeable difference in difficulty for each day of the Mini as there is with the full 15x15 NYT crosswords. (The crossword gets progressively harder with Monday as the easiest and Saturday as the hardest. Sunday is a midweek difficulty puzzle that is just bigger.) My solve times for each day of the Mini are within seven seconds of each other (01m:23s to 01m:30s).

In Conclusion

If you haven’t figured it out by now, I’m pretty into crosswords, big or small. Hidden within the clues, I think there are some lessons I have learned from this hobby.

  • I always use a pencil because I try to not be afraid to guess a word even if it may be wrong. The best strategy can sometimes be to write down a guess to give you a hint for another word, meaning that sometimes you have to take a risk to make some progress.
  • I really only do crosswords with other people. The amount of information and random bits of wisdom that two people know will always be more than what one person knows.
  • Oftentimes, the best thing to do when you are stuck and can’t figure out any more words is to just walk away and pick it back up the next day. It always amazes me the new perspective I gain when I sit back down after a break. A four-letter word for “its subject to inflation” at first glance is “gold” but later becomes “egos” with a fresh set of eyes.
  • Lastly, despite being the most competitive person I know, I have learned that it is perfectly okay to look up the answer when I can’t figure it out and admit defeat.

--

--

Austin Steinhart

Human-centered data nerd learning how to tell real stories with real data