top of page
Post: Blog2 Post
  • Writer's pictureJacob Cohen

Puzzle Survey Results!

Real quick: If you just want to see the full chart of ratings, without all my commentary, click here. However, if you want to experience the rankings in a more dramatic, suspenseful way, that's what the rest of this post is for.


The Post

As y'all may know, I recently sent out a survey where you had the opportunity to rate nearly every Puzzles for Progress puzzle. This was mostly to inform which puzzles would appear on Puzzles for Progress in the future (and it will), but I thought y'all might find the results interesting as well.


A disclaimer to start off–I got eight responses. The survey contained 25 puzzles; the average respondent had an opinion on over 20. which is enough to form some conclusions, but not enough for this exact ranking to be super meaningful. Now, let's blow right past that and give the survey results maximum fanfare!


The Most Popular Puzzles

So, with that in mind, the most popular puzzle is, drumroll please (oh, who am I kidding, you can see it in bold text below)...


1. The King's Adventure. (Score: 83–I'll explain how these scores are calculated soon.)

(The most recent King's Adventure, from Issue #21.)


This was...not what I expected! The King's Adventure, invented by Israeli mathematician Gyora M. Benedek, is more commonly known as either Hidato or Number Snake, with the slight twist that 1 and 36 don't need to be next to each other. The puzzle is a PfP veteran, to be sure–it's appeared four times out of 23 puzzle pages so far. (Also, it's appeared in each puzzle page whose issue number is a multiple of 7. Will this trend continue? Subscribe to find out). I've always considered it to be a solid team player, sure, but not a superstar–which is why this result surprised me! But the data has spoken, and I shall listen.


I'm just speculating here, but I think part of the King's Adventure's success may be how its difficulty slowly ramps up over the course of the solve. It's usually pretty easy to find a piece of logic to start off with (the 17 and 18 in the above puzzle), but by the end, more holistic thinking is required. Anyway, idk–I didn't ask about why people liked the puzzles they did.


Methodology

Before I share any additional results, a little bit about how the scores were actually calculated. I asked about 22 puzzle types, 2 puzzle categories (linguistics puzzles, like Numbers in Toki Pona; and basic word-guessing puzzles, like Silence), and the "general format of one larger puzzle (instead of three smaller puzzles)," which I did back in Issues #4 and #11. So 25 puzzles overall, grouped into five sections of five. I did not ask about Word Search Word Search, Arangams, Traffic, Quadruplets, Sifting, and Persephones, mainly because they're bad.*


I showed an image of each puzzle for reference before asking about it. The questions were all worded essentially as below:


How much do you like [PUZZLE NAME]?

- One of my absolute favorite puzzles!

- I definitely like it

- It's nice

- Meh

- It's just not my kind of puzzle...


My goal was to word the options so they would each be selected approximately the same amount, thus giving me maximum information. It ended up being fairly top-heavy anyway–people were hesitant to select the bottom two choices. (That's very nice of y'all, but not as helpful!)

In any case, I assigned words to numbers using the following scheme:


100 = One of my absolute favorite puzzles!

75 = I definitely like it

50 = It's nice

25 = Meh

0 = It's just not my kind of puzzle!

No opinion responses were ignored.


I then averaged all responses to get the scores for each puzzle!**


Now that I've given a quick runthrough of my procedure, I'll resume telling you the results very slowly and with a lot of commentary.


Me Resuming Telling You the Results Very Slowly and with a Lot of Commentary

The puzzle that came in second was the one that I expected to win from the outset:


2. Capsules. (Score: 78.)

(The second-most recent Capsules, from Issue #17.)


Capsules was invented by Naoki Inaba, and is more often called Suguru. However, I first encountered it in the New York Times magazine, where it was called Capsules and created by Wei-Hwa Huang. I fell in love immediately, and I included it in the first edition of Puzzles for Progress. I even made a 9x9 Capsules puzzle with no given squares at all, published in Issue #8! (For context, that's very cool.)


With six appearances so far, Capsules has been the most frequent puzzle on Puzzles for Progress, and as the puzzle's blurb says, I did write a whole blog post on Capsules awhile ago (before Puzzles for Progress existed). It's at tinyurl.com/confluxwix/post/conquering-capsules. As I write this, the blurb on the homepage of the Puzzles for Progress website refers to "fan-favorite Capsules." (It makes no mention of the King's Adventure. Perhaps I'll revise it a bit.) It did also receive the highest score from the ancient survey I sent out in July. Everything was set up for Capsules to succeed in this survey–and it did!


Now, for number three:


3. Enclosures. (Score: 75.)

(The most recent Enclosures, also from Issue #17.)


Enclosures is definitely an interesting puzzle–a sort of reverse Capsules, and one of only three puzzles in the survey to be a Jacob Cohen original! (That is, I invented this puzzle type, directly from my brain.) It has appeared in Puzzles for Progress on three occasions.


I'm frankly a bit surprised Enclosures managed to score so high, especially based on anecdotal evidence; but the numbers deserve to be trusted over my own intuitions. Anyway, it's quite nice in my opinion that Capsules and Enclosures are right next to each other–I'm not sure if they're friends or enemies, but definitely appropriate for them to be so close to one another.


Coming in fourth, my personal favorite puzzle:


4. Spiral. (Score: 72.)

(The most recent Spiral, from Issue #22.)


I. Love. Making. Spirals. I just do. I've found myself coming up with Spiral segments while trying to fall asleep, showering, or otherwise "off-duty." Four Spirals have appeared so far on PfP (which is a lot) but even that's selling it short, as it got a late start (while it technically debuted in Issue #6, I only really got the formatting down by Issue #13.) I think the #4 rank is about what I expected, though it's definitely #1 in my book.


Now, let's wrap up our sojourn through the most popular puzzles with number five:


5. Crossword. (Score: 70.)

(The most recent crossword, from Issue #21.)


Crosswords aren't as cool as Spirals, but they're still pretty cool. Along with Capsules, it also appeared in Issue #1, and has appeared in a total of five puzzle pages overall. Y'all know how crosswords work. I don't have strong feelings about them one way or another, but it's a sign of their


The Least Popular Puzzles

Before I reveal the chart, I would also like to skip to the bottom to talk about...those puzzles. Third from the bottom, 22nd overall:


22. Hidden Wisdom. (Score: 50.)

(The most recent Hidden Wisdom, from Issue #22.)


I first found Hidden Wisdom in The Colossal Book of Wordplay, by Martin Gardner–also where I learned about Word Ladders, and where I got the idea for Odd Jobs. It was just a one-off in that book, a page at most. But I expanded it out into four Hidden Wisdom puzzles far, partially because they're one of the easier puzzles to make. But I get it: it's kinda hard and kinda confusing.


I will note, though, that Hidden Wisdom wasn't very polarizing: every response but two was the median "It's nice." (In other words, 50.) I don't know if I'd be less inclined or more inclined to have this puzzle return if the scores were split between 100s and 0s, but I hope this means it won't be controversial that Hidden Wisdom will be taking a break for awhile.*** Now, second from the bottom, the puzzle I expected would come in last:


23. Overwrite. (Score: 44.)

(The most recent Overwrite, from Issue #9.)


I like Overwrite, but I've known I was in the minority here for awhile. And so I understand how it gives some people headaches, and just doesn't feel that rewarding overall. I published three Overwrite puzzles in the early days, but I've since retired the puzzle. That said, Overwrite is a bit more on the polarizing side–the people that do like Overwrite tend to be a bit more enthusiastic. And, coming in last place, the wooden spoon of puzzles:


24. Nurikabe. (Score: 41.)

(The most recent and only Nurikabe, from Issue #19.)


This one's a bit of a shame, honestly, because I think Nurikabe is a pretty fun puzzle. But it is on the complex side in terms of rules–there's a lot going on. Ultimately, I think its complexity, coupled with my poor explanation, probably dethroned it in the end. Not every puzzle can be a superstar.


Now that I've gone through the top 5 and bottom 3, I'll now show the full chart:


The Chart

What else stands out to me, apart from the mediocre image quality? (Sorry.)


Well, the number of appearances is not as correlated with score as I might have hoped. Crossword is fifth from the top, yet Word Squares is fourth from the bottom. Among puzzles that have only appeared once, the Acrostic, Word Ladder, Numbergrid, linguistics puzzles, Sudoku, and (narrowly) Killer Sudoku do above average–Acrostic leads among that group by under half a point. I saw that Word Ladder was doing very well early on, but that proved to be a mirage.


Anyway, I'm probably a lot more invested in these survey results than all y'all, but I still wrote them up anyway because why not. That's it from me!

–Jacob

Cross-posted with Chromatic Conflux


*Well, the first three were bad. Quadruplets probably would have done poorly, but it is also unrepeatable by its nature (the puzzle is based specifically on the case of a four-dimensional cube). Among these six, it's probably the one I'm most inclined to repeat. Sifting was objectively unrepeatable. And with Persephones, there just aren't enough persephones to really do another...maybe if I stretched it?


**For math nerds, I actually did something a bit more complex: I included two bonus values of 63 (the average overall) in each average. This is similar to something called an uninformative prior in statistics. To illustrate why it's useful, imagine this situation:

Puzzle A has one 100 rating and ninety-nine "No opinion" ratings.

Puzzle B has ninety-nine 100 ratings and one 75 rating.


Which puzzle are you more confident in? The simple average would tell you Puzzle A, but just throwing in two copies of 63 will pull down Puzzle A's average a a lot while having basically no effect on Puzzle B's average. Generally, this has the effect of pulling puzzles without as much data towards the center. It's worth noting that since puzzles had a maximum of 8 ratings, the theoretical maximum was 93 (not 100), and the theoretical minimum was 14 (not 0). Of course, if a thousand people did the survey instead of 8, ratings of 100 and 0 would be possible.


However, since puzzles had approximately the same number of data points as one another, a simple average would produce approximately the same results. In other words, if you don't understand what I did, ignore it.


***By the way, if I have twenty-four puzzle types, and I'm doing three a week, that would mean each puzzle appears once every two months on average. But accounting for some puzzles being better than others, and also my desire to include newer varieties, I don't think it's at all radical to think that Hidden Wisdom won't return for three or four months at least–even if it's not hated.

43 views0 comments

Recent Posts

See All

Comments


Thanks for subscribing!

bottom of page