Another MIT Mystery Hunt is in the books, and in many ways it was a wonder. Teammate, the constructing team, packed the weekend with innovation after innovation. Early puzzles were preceded by a loading screen that was itself a puzzle — and if you failed to notice its puzzly nature, the loading screen stuck around for longer and longer, as if the site was having a breakdown, when really it was the constructors saying “You might want to pay a little more attention to this.” Damned clever. Grasping the meaning of the loading screen allowed you to step behind the curtain of the Mystery Hunt, and explore a Puzzle Factory where various AI puzzle bots needed help completing the very Mystery Hunt you were at that moment solving. Widgets on the factory floor actually changed elements of certain puzzles, requiring serious coordination between solvers to set things properly so that everything could be solved. This was all really quite special and ingenious, and exactly why the Mystery Hunt holds a special place in the hearts of so many hardcore puzzle people.

The 2023 Mystery Hunt was also, as has been observed by many, far, far too difficult overall. There was hardly a softball to be found — every puzzle seemed to take hours and hours. Many answers could only be gleaned after a grueling amount of work, or a pile of challenging ahas, any one of which could throw the solvers permanently off the track. The constructing team themselves saw the trouble early on, and began to give away free answers — a necessary but demoralizing thing to have to do. Even so, the Hunt extended into Monday morning, many hours later than is ideal.

Gauging the difficulty level — and thus the length — of your puzzle event is one of the most daunting challenges faced by a constructor or a constructing team, and this is by no means restricted to the Mystery Hunt: I run several small puzzle events each year, some for children and some for adults, and even though these events might be only an hour or two in length, it takes a lot of effort and forethought to get the timing right. Still, there is no question that the Mystery Hunt is on a whole other level of challenge in terms of calibrating the difficulty — it’s so sprawling and complex, and often it is put together by teams without a lot of experience in large-scale events, who are learning as they go along.

And so I would like to offer a word of advice to future Hunt-writing teams. I have theories about how you can accurately estimate your Hunt’s endpoint; I have thoughts about how many puzzles a Mystery Hunt needs to have. But neither of those topics is nearly as important as the following nine words:

YOUR PUZZLES DO NOT NEED TO BE INCREDIBLY HARD.

Keep this maxim at the forefront of your minds as you plan your event, and success is all but assured.

To be clear, the first time I captained my team, Palindrome, as we ran a Hunt — back in 2008 — we did NOT understand this, and I in particular did not understand this. Not only were our puzzles too hard overall, but in several places I personally required solvers to all but read my mind in order to understand how a final answer was meant to be extracted. It was not a particularly good Hunt, and Palindrome spent the bulk of fifteen years trying to reclaim victory so we could show what we could do now that we had some experience under our belts.

For 2022, I (once again the captain) was laser focused on making sure the Hunt ended at a reasonable time. Not including the “Star Rats” prologue set of puzzles (which were released a month early), solvers faced approximately 167 puzzles across Hunt weekend, including metas. Of these, 47 were solved by a group of testers in about an hour, and 83 puzzles required two hours. We expected the remaining 37 puzzles to take three or four hours.

So fully 75% of our Hunt’s puzzles were solvable in under two hours: Does that mean our puzzles were easy? I don’t think so, no. They were in many ways typically bananas Mystery Hunt puzzles, just the sort of thing someone who attends the weekend expects to see: There were many puzzles about esoteric subjects, and rarely were there any instructions. And when our puzzles tested at two hours or less, that was because of an effort by a team of four to eight people working together — a single solver would have taken far longer. “Easy” is not the first word to leap to mind as I page through the archive of the 2022 Hunt, nor do I recall anybody complaining about the simplicity of our puzzles in their post-event write-ups.

With the data from our testsolves, and with the knowledge that teams would be working on different puzzles in parallel, we felt comfortable that a team would find the coin by Sunday at noon. In the event, the winning team crossed the finish line just before 10:00 a.m. Had the proportions of our puzzles been reversed — if 75% of them had been three- and four-hour solves, and just a quarter in the one-to-two-hour range — we could well have needed to resort to desperate measures to make sure someone won at a reasonable time.

Am I suggesting that Palindrome ran a perfect Hunt in 2022? Not hardly. There was a midpoint meta that proved an unexpected bottleneck for many teams, and it was pointed out in several places that we could have done a better job integrating the Hunt’s storyline. But for the most part I think we absorbed the lessons we learned back in 2008, the most important of which is to resist what I call Mystery Hunt Fever: The uncontrollable desire to amp up the difficulty of your puzzles again and again, draining away much of the puzzle’s fun in the process.

It is possible that this important lesson can only be learned with direct experience — Lord knows that was my main teacher. But I lay it out here in case it does someone a little good. The Mystery Hunt may have a reputation for extreme difficulty, but that reputation obscures the truth of the matter: Your Puzzles Do Not Need To Be Incredibly Hard.

The MIT Mystery Hunt is just a few days away, which means it’s time to sharpen your puzzle-solving skills on something that is in no way actually a puzzle. A puzzle by its nature is something purposefully constructed, often with painstaking effort so as to have the right difficulty for its intended audience. When I was part of the Mystery Hunt constructing team last year, many of the puzzles took weeks or longer to put together.

As opposed to a round of Spaghetti, which I construct by choosing five words at random from a pocket dictionary.

You’re saying to yourself, “Wait, how can a handful of random words be a puzzle?” To which I respond: It isn’t. But in the game of Spaghetti, we pretend these words are a puzzle, and then we try to solve it.

If that sounds ridiculous and impossible, you are both right and wrong. It is ridiculous. But we have seen over the years that it is not impossible — every time we run this bizarre experiment, participants point out complex patterns in the given words. Sometimes those patterns are convoluted and absurd, and there is a certain amount of fun in that. But other times what the solver discovers is so surprising and elegant it makes you wonder if the words really are chosen at random (they are).

Here are the five words for this round of Spaghetti:

VERIFY
WHITE
EMULSION
JOCULAR
SQUAT

To help you arrive at your solution, you may add a sixth word of your choosing to this list. Perhaps obviously, don’t just say “The answer is [whatever]” — you need to explain how you arrived at that answer.

Even if you can’t come up with a solution to this puzzle (and why should you, seeing as it is NOT A PUZZLE), check back on this post throughout the day, and read the submissions from other solvers. Give a thumbs-up to the solutions that you like. You can vote for as few or as many comments as you wish. The player who submits the solution that attracts the most likes will be the winner of the game.

(Pictured above: What DALL-E returns on the prompt “spaghetti puzzle.”)

It’s a new puzzle hunt! Eleven fun and varied puzzles, including a metapuzzle. Originally commissioned by the Mohonk Mountain House and run during its “Wonderful World of Words” weekend. Now you can solve it at home — do it solo or grab a friend or two!

Come meet Eureka K. Jones and his amazing inventions right here.

At this year’s convention of the National Puzzlers’ League, the traditional Saturday-night extravaganza was constructed by over a dozen people, each contributing a single Lord of the Rings-themed puzzle. Here is my contribution. I watched one team solve it exactly as I intended (stare stare stare, Oh!, solve solve solve) and another team get totally stuck (stare stare stare stare stare stare….) Hopefully more of you will fall into the former category, should you try it.

It’s a new puzzle hunt for kids! Summer Camp Color War features puzzles based on Tug O’ War, Capture The Flag, Scavenger Hunt, and many others — nine interlinking puzzles for ages 12 and up. You can grab the set for your brainy, puzzle-solving kids over at Puzzlesnacks.

On the most recent episode of Dave & Ethan’s 2000″ Weird Al Podcast, I’m interviewed (along with my friend Joe Bohanon) about the work I’ve done with “Weird Al” Yankovic — the New York Times crossword he and I co-authored back in 2018, and the cameo appearance he made during this year’s Mystery Hunt.

A few scattered thoughts on putting together a Mystery Hunt, for the 2023 constructing team and anybody else who aspires to do this.

Put on the best Hunt you can and don’t worry about expectations

It is often debated, particularly in the weeks immediately following the event, whether or not the MIT Mystery Hunt has grown too large. It absolutely has. I don’t mean for the solvers; I mean for the constructors. Each team that takes on the responsibility feels an urge to match or exceed what came before. There is only so far this can go before it is simply not possible to put together the Hunt in one year’s time, and I think we might be nudging that limit right now. For a number of people on Palindrome, the 2022 Hunt became a full-time job on top of their actual full-time jobs. I do not recommend this. A constructing team should carefully judge what they can accomplish in the time allowed, given their resources and keeping in mind that a year is barely sufficient to put together a Hunt of the size people have grown to expect. If you need to pull back to a mere 125 or so puzzles, no one in the Hunt community will say boo.

Life finds a way…

…of messing with your plans, and you would do well to keep that in mind from the start. When Palindrome put together the structure of its Hunt near the start of 2021, our team mailing list was 117 members strong. But as aspirations turned into responsibilities, some people drifted away. Others intended to stay but were sidelined by the usual travails of life — medical emergencies, legal emergencies, family emergencies, work emergencies. What we planned to accomplish with 100+ people was in fact accomplished with 50-60 people. If we’d had 50-60 people at the start, our plans might have been a little different.

There are going to be small teams and there are going to be large teams

I like to think Palindrome did a few things right this year, but nothing in my mind was righter than having different tiers of accomplishment — only nine teams reached the Hunt’s grand finale, but sixty teams successfully thwarted the Voracious Bookwyrm, earning the midpoint token. Another 48 teams on top of that solved the opening meta. There is a lot to be said for allowing every team, large or small, to succeed at something significant.

Communicate, communicate, communicate

Sixty teams conquered the Voracious Bookwyrm, but it would have been an even larger number if Palindrome had been clearer that there was a midpoint reward to be earned. We wanted it to be a nice surprise, which may have seemed reasonable, but it left some teams out in the cold — they moved on to the Pen Station rounds without completing The Ministry, and so they missed out on the midpoint token. The Hunt has room for surprises, but what the teams are trying to accomplish should perhaps not be one of them.

Similarly, we were unclear when communicating with teams about when our HQ would shut down, and what exactly that meant. So you had some teams thinking the door to the Hunt would be slammed shut entirely at 6:00 p.m. on Sunday evening, when in fact we thought teams would keep solving — it was only that hints and interactions with Palindrome would end at that time. This caused some confusion here and there. I fully shoulder the blame for that. Sorry.

And that’s only talking about outward communication, toward the solving teams. I could go on at length about internal communication, with the fellow members of your construction team over the course of the year. As team captain, I sent out weekly updates about how we were faring as our deadlines zoomed toward us — but there were elements of our Hunt, including the unlocking structure, that were not as well communicated, and that caused avoidable problems late in the year.

Do as I say, not as we did

Our set of prologue puzzles, the “Star Rats” round, was a big hit. I’m glad we were able to do it. But future constructing teams should feel NO OBLIGATION WHATSOEVER to make a prologue set part of their responsibilities. The Hunt is hard enough to put together without layering on even more, optional puzzles. Seriously, I hope a prologue round DOESN’T catch on.

Believe the testsolvers

I almost certainly do not need to tell you the importance of testsolving. Nearly every puzzle constructor has learned that how they think solvers will interact with a given puzzle, and how solvers will actually interact with it, are frequently two different things. Testsolving reveals all: Oh look, your clever joke in the flavortext sends solvers down a completely unforeseen rabbit hole. Gee, the aha moment you thought might be too blatant is in fact utterly invisible. Golly, solvers are failing to recognize which of the many possible solving paths is the correct one.

On Palindrome, the testsolve sessions were almost always set up and attended by the author, and in my opinion that is 100% the way to go. Handing off a puzzle to be tested out of the author’s sight is not nearly as valuable — that might do the trick for something like a newspaper crossword, but not for a many-layered experience like a Mystery Hunt puzzle.

Testsolving only works if you believe the testers’ experience will be typical, and if you believe what the testers say in their feedback. (Have a way for testers to give feedback!) It will be tempting, now and again, to say “Well, the testers went down this particular false road, but that was a freak occurrence that real solvers will be unlikely to duplicate.” No. Your puzzle has a flaw, and the testsolvers were good enough to reveal it to you. Figure out how to fix it.

We tried very hard to allow a puzzle into the Hunt only after two clean testsolves. One is not sufficient. I had a puzzle in the 2022 Hunt called The Messy Room. It was a variant on a standard type called the dropquote, in which letters above a grid drop down into that grid to spell a quote reading left to right row by row. In my version of the puzzle, letters could drop down into one grid or rise up into another — and also the words in the quote were presented out of order. This was hard enough, but in the original version, words in the puzzle’s seven quotes were intermixed — some words from quote A could be found in quote B, and words from quote B fell into quote C, and so on. The first group of testsolvers to tackle this… knocked it clear out of the park. They weren’t stopped for a moment. They solved the puzzle in a little over 75 minutes. It was perfect. I organized a second testsolve largely to set a good example — surely it wasn’t necessary. And you have already jumped ahead to the punchline: The second test was a disaster. My testers couldn’t make sense of any damn thing. I cut our misery short after 45 minutes and sent the puzzle back to the drawing board.

There exist puzzlemakers who will give extra weight to the good test over the bad. Don’t be one of them. I try not to lean toward pessimism in general, but when it comes puzzle testing, it’s definitely the way to go. Assume your puzzle’s worst test is what every group of solvers will experience, if you don’t do something about it.

Post-production: Oh, right, this is important

Palindrome was so focused on getting its 200 or so puzzles over the conveyor belt, we didn’t start preparing the puzzles for the website until late November. This caused quite a crunch throughout December. Get your website going earlier and you will be a happier team for it. Plus this will let you get in one more testsolve after the puzzle is theoretically absolutely final — errors can and will creep in at this stage.

But wait, there’s more

Scriptwriting. Costumes and props. Logistics. Creating and editing videos. Making sure everything is as accessible as possible. Getting permission from MIT for this, that, or something else. Dealing with the forms MIT insists on having signed by every solving team. It’s easy to overlook or underemphasize these things when there are so many puzzles to create. The non-puzzle elements will demand to be dealt with, so best to have a plan from the start.

Maybe it is time to retire the scavenger hunt

I don’t speak for my team on this point, but I believe it nonetheless. Particularly if the Hunt continues to be virtual, which inflates the number of participating teams, give serious consideration to whether or not to include a scavenger hunt. Administering the results has become a brutal time sink. This year we received submissions from 168 teams, and each resulted in a process that took approximately one hour to complete. You do the math.

Yes, the scavenger hunt can be fun, and yes, it’s a good way of making sure you’re in contact at least once with every team. But there are other fun ideas out there, and other ways of making sure every team gets contacted.

Don’t underestimate the personal toll

Winning the Mystery Hunt is exciting! Getting started on the next year is also exciting! Anything and everything is possible. But it won’t be long before reality seeps in. Decisions will get made that not everybody agrees with. Some people will find it difficult to find a place for themselves in the construction process. Pressure will mount, and that pressure will be handled by different people in different ways. Personal relationships could become strained. Following in ✈✈✈ Galactic Trendsetters ✈✈✈’s footsteps (that’s where the apostrophe S goes, right?), Palindrome had an ombudsperson who specifically fielded concerns from the construction team and relayed them to the organizers, and several course corrections were made over the year as a result of this.

The most important part of constructing the Mystery Hunt, in my opinion, is not pleasing the Hunt community with the result (that’s a close second), but coming out of the experience healthy and with all the same friendships (if not more) that you had at the start. Take a breather if you have to. Sit down with the person you’re butting heads with, assume you are both working in good faith, and talk out the problem. The Mystery Hunt is work, yes, but it should also be fun. Keeping it fun, however, may take a little effort. Time spent on that effort will be time well spent.

It remains to be seen whether I will write a more extensive post about what it’s like to help make a Mystery Hunt — I thought I would, but I’m finding the task a little overwhelming at the moment. But it’s easy enough to at least talk about the puzzles I contributed. Here they are, with commentary:

Algernon: A labyrinth variety crossword seemed the obvious approach for a puzzle about a lab rat. I make small Labyrinths all the time for Puzzlesnacks, but this was one of my few large-scale versions, and with an added twist to boot. The first attempts were on 13×13 grids with only two words per row, but six failed attempts later the light finally dawned that maybe I needed to stretch this out to 15x.

Alice’s Adventure in Wonderland: A variation on Fraternal Twins, a form I invented a couple of years back. I’m pleased with how this came out, all the more because there was an extra hidden constraint that wouldn’t be revealed until much later — after you fill in the grid, see if you can extract a second answer using this strange glyph. (Credit to Mark Halpin for coming up with the second-level idea.)

Battery Pack (with Mark Halpin, Dave Shukan, and Foggy Brume): A puzzle you can’t technically solve without first solving all of the other user-manual pages, like Narnia Beeswax below. But I include it here because, well, it was one of my puzzles, and that’s what this is a list of.

Communicating With The Aliens (with Justin Ladia): A meta-puzzle that only a handful of teams got to see — it was in the last region solvers explored before heading into the finale. If you’d like to try this without solving every other puzzle in the round, you’ll find the answers you need here.

Danni Dewey: A mini-meta, one of five needed for this part of the Hunt. Not exactly solvable without doing a lot of other stuff, but again, I’ve included it for the sake of completeness.

The Enchanted Garden: A variety type I’d never attempted before, and I’m pleased with how it came out. This is one of the few puzzles I actually got to watch a team solve over the weekend — thanks to Illegal, Immoral & Fattening for letting me peek in at their spreadsheet.

Hell’s Kitchen: We meant to have a number of “scrum” puzzles, technically whizzy video-game-like experiences that teams could play at the same time across multiple computers. The one scrum puzzle that reached the finish line died during our final run-throughs and had to be replaced with a back-up puzzle. Hell’s Kitchen was meant to be another such puzzle, but as fall headed into winter I could see that our tech team had far too much on their plates. I grabbed the answer and turned it into this.

Ice Cream Roll: I created this type a couple of years ago with the help of Mark Halpin — you might have seen it now and again in the New York Times. This puzzle is usually called Jelly Roll, and it usually has two layers instead of three, and it usually isn’t quite this hard, to construct or to solve.

Just a Dream (with Mike Nothnagel): A simple enough idea but a mondo pain to research, so my friend Mike’s jumping on as co-author was much appreciated.

Kid Start-Up (with Lea Berlin): One of my favorite things about working on the Hunt this year was the participation of my daughter, Lea, who played a major role on the art team, most notably with her cover to the Star Rats prologue. (Zappy forever!) She and I co-constructed this fairly straightforward puzzle for the Hunt’s second act.

The Mad Scientist’s Assistant: Another puzzle type I created for other purposes, here given massive doses of steroids so that it could serve as a Mystery Hunt puzzle. When I started this, it didn’t immediately dawn on me that, duh, all of the stacks of boxes would need to be the same size. That made it a pretty challenging construction, but the result was well-received by my team.

The Messy Room: Whoops, I skipped right past this one in my first version of this post. Maybe that’s psychologically revealing: In retrospect, while it fits the theme of its answer to a T, this was perhaps a bit too much puzzle to throw at people so early in the Hunt, when we wanted to start off with lighter fare. And the solving process involved a lot of brainless pattern searching as opposed to actual thinking. This puzzle earned its place on the “most requested free answers” list.

Narnia Beeswax: Completing a round of the Bookspace Hunt earned the team a “user manual page” for the Plot Device component they have just acquired — each one a puzzle, of course. Note that this puzzle has six separate answers.

Rack ’em Up: I won’t spoil the answer, but I’ll say that when I hit upon pairing this answer with this particular metaphor, this puzzle just about wrote itself.

Scream (with Gavin Edwards): Gavin and I first met in 1990, sharing a tiny office as the newest entry-level proofreaders for PC Magazine. He left before I did, moving on to take the next steps en route to his career as a successful celebrity journalist. (Buy his latest book!) And so we were out of touch until the age of Facebook. That’s where he read my annual write-ups of the Mystery Hunt, which ultimately led to a 30-years-later reunion when he joined Palindrome. With his encyclopedic music knowledge, it was a natural that I would turn to him for help on this puzzle. In the end, he did almost all of the legwork, including the pain-in-the-ass overdubbing of the Jaws theme, which we needed to add for reasons I won’t spoil here.

Spy Game (with Katie Hamill, Kevin Wald and Wil Zambole): It’s not for nothing that it took four of us to get this together: This is perhaps the meatiest and most Mystery Hunt-y of my contributions this year. This one took a lot of tweaking and adjustment to get it to its final state, and we were hashing over specific word choices for ages.

Swingin’ (with Gavin Edwards): One of the first puzzles created for the Hunt. My daughter did (most of) the art and I think you will agree that it is adorable.

You Took The Fifth (with Katie Hamill): “You should do an Ace Attorney puzzle,” Lea said one day. I explored the editor and agreed that the possibilities were endless. I roped in my friend Katie to help (she’s great at wordplay and she’s a lawyer, so who better?), and together we came up with this rather surreal series of dialogues.

It’s National Spaghetti Day! As I’m sure the originators of this annual celebration intended, let’s celebrate by playing a nutty word game!

In many puzzle events, solvers face a wide assortment of puzzles, with each puzzle leading to a final answer word or phrase. These answers are then re-used in what we like to call the metapuzzle, or just plain meta. The answer to the metapuzzle is often a satisfying punch line that puts a bow on the entire event. That meta-answer can frequently be found by noticing an important pattern in your collection of answers. It might be that all of the answers are words in the titles of James Bond movies, or can all be turned into new words if you change a letter to X.

People who love puzzles and puzzle hunts have gotten insanely good at spotting patterns in a series of answer words — so good, in fact, that years ago I invented Spaghetti, a game in which I challenge solvers to find the answer to a meta puzzle even though no answer exists.

You heard me. I am about to give you a puzzle that has NO ACTUAL ANSWER. The words have NO PATTERN WHATSOEVER — at least not on purpose. I chose them entirely at random. It should be completely impossible to transform these words into a satisfying solution.

And yet, it can be done. I know that because people have found “answers” for these random collections of words every time we’ve done this. And frequently the explanation for how the puzzler arrived at their answer has such a degree of elegance that you’ll swear it must have been the intended answer all along.

So here we go with Spaghetti 2022! Here are five words. In the comments, tell us what your answer is, and why. You may add an optional sixth word to this list — whatever word you wish. Or you can stick with these five.

CAUSE

NECESSARY

WOOZY

CONSTITUTE

QUAINT

Can’t figure out an answer? No worries. Come back often and read the comments, and vote for your favorite responses. You can vote for as many as you like. The solver who gains the most votes will be declared the winner.

One last note: Some of you know that I am helping to put together the 2022 MIT Mystery Hunt, one of the year’s most anticipated puzzle events. As such, you might be wondering if this post is a Mystery Hunt puzzle in disguise. It is not.

One month from now, the 2022 edition of the MIT Mystery Hunt will begin. I’ve been working on it all year with my friends on Team Palindrome, and we’re very excited to show you all the puzzles we’ve been making. So excited, in fact, that we simply couldn’t wait: If you go to starrats.org, you can start the Mystery Hunt right away! (And if you’d rather wait until January 14th, that’s fine, too — the Star Rats puzzles are an entirely optional appetizer.)