BAPHL 4 post-mortem
BAPHL 4 was, of course, the largest puzzle hunt I’ve been involved in creating, and the first where I’ve ended up in charge of making the event happen. It seems to have gone very well.
The hunt would not have been nearly as good without our hordes of testsolvers, especially the three whole teams that Charles rounded up. They caught lots of mistakes, and lots of puzzles that were unreasonably hard. We spent much of the last week before the hunt cutting puzzles down to size wherever we could, and yet the three teams took 8 to 11 hours to solve the Normal Mode (admittedly with distractions), prompting us to cut even more.
There were a few cases where we did this by replacing a Hard Mode puzzle with a thoroughly-testsolved Normal Mode puzzle, which means that for much of the hunt, Normal Mode teams were solving the same puzzles as Hard Mode. This would presumably be why only one of the five Normal Mode teams finished. I apologize for that part. Normal Mode is a nice goal of BAPHL, but it’s hard to do. It means you have to calibrate the difficulty of each puzzle twice and testsolve two versions of it, and already you can never have enough testsolvers.
On the plus side, the difficulty ended up being a reasonable challenge for the 24 Hard Mode teams, with the winners finishing in 3 hours, and with at least 14 of them finishing within the total 7 hours.
The puzzles, in the form that we distributed to people, can be found on the BAPHL 4 page.
Here are my behind-the-scenes comments on the individual puzzles. (These have substantial spoilers, if you were planning to try the puzzles.)
E-Z Does It (Adam Hesterberg, substantially modified by me)
This puzzle started as a much longer puzzle named “Easy Enigma”, where the answers to a bunch of clues would turn out to start with two letter sounds. I shortened it, especially because I wanted a puzzle there that was truly easy so teams could unlock round 2.
For a while, its title was “Easy Does It”, but many testsolvers were still not finding it easy, so finally we changed the title to the one that hits you over the head with the mechanism.
Beginnings and Ends (Bryce Herdt)
We were worried about this puzzle’s difficulty for a while. We had testsolved several versions of it that seemed to be just a bit too hard for BAPHL, having clues that were unordered within each line, as well as harder clues such as “Silly person” cluing GOOFBALL. But we got the puzzle into a reasonably finished state several days before the hunt.
Bit by Bit (Amos Eshel)
We had a very hard time getting the difficulty right on this one. We must have gone through at least 10 versions of the puzzle, oscillating between far too hard and far too easy. We were still worried about the hard version on Thursday night, and getting it testsolved (by a non-native speaker of English, who found it very hard and time-consuming) was only slightly reassuring. And yet in the hunt it worked out fine.
The Chase (Catherine Havasi and me)
This was our least-testsolved puzzle, because it required being on site. The major glitch was that the clue to look for “the lightning man” in the Harvard Bookstore, which was supposed to lead you to a biography of Samuel Morse, simply didn’t work. Some teams didn’t encounter the problem because they had already solved the puzzle by that point — but those are the stronger teams. So this puzzle had an unfortunate flaw that only slowed down the teams who were already behind.
To make matters worse, once teams had the clue to use Morse code, they would likely look at the Morse code chart in the intro packet, where I had mistyped the code for the letter U.
We had tried to actually order a copy of The Lightning Man and get the Harvard Bookstore to put it in their window, but they brushed us off. I tried and failed to find another nearby way to clue Morse. Finally, I just had the instruction point people to the catalog computer in the Harvard Bookstore. I checked that the computer would give you the book if you searched for “lightning man”. I checked again on their website, which I thought used the same catalog, on Friday. In the 15 minutes I had to spare on Saturday morning, I checked most of the route again, but I skipped the end, knowing that most of its clues were literally carved in stone. But that meant I also skipped the bookstore.
And that’s why, if you used the Harvard Bookstore’s catalog in person on Saturday, you turned out to get some unrelated book. It’s a law of the Universe: the one part of your puzzle you don’t check will be dead wrong.
Totem (Jason Alonso and me)
Being the final metapuzzle, this puzzle was written and testsolved approximately five months ago. We made one change to it since then, which was to add the L to the middle hex of the puzzle.
Originally, we intended for the radii of the hexagon to give you six-letter combinations that give unique eight-letter words when you add letters to the front and back, all of which happened to end in L so that they could share the middle hex. Testsolvers didn’t actually have much trouble finding these. But later, testsolvers who tried the puzzle with incomplete information tended to try to use the L in their answer (it’s clued six times, so it’s probably important!). So we put it in the puzzle, making it easier. Not that that’s a bad thing.
I don’t know why I clued “ESOTERIC” by the fact that it has an obscure anagram (COTERIES). A bonus for Scrabble players? Pure sadism? But people seemed to be able to deduce which word could be anagrammed anyway, which was my hope.
Normal-mode solvers got all the word lengths marked on their hex grid, which for hard-mode solvers would have been a dangerous amount of information. It enables many, many opportunities for backsolving — and in fact, one of our normal-mode testsolving teams managed to backsolve the entire middle meta from Totem! But I actually think that’s a good thing to give to less experienced teams, so they can experience the joy of backsolving.
Keeping Tabs (Sam Trabucco)
Sam sprung this puzzle on me when I thought he was going to give me a different one, and that’s great, because I consider it to be the most original puzzle idea in our whole hunt. It’s elegant in an unconventional way despite being chock full of red herrings.
I didn’t really get to find out how teams reacted to it during BAPHL, but I remember having people test-solve it in my living room while I went off to edit more puzzles, and suddenly hearing a chorus of “OHHHHHH”. That’s a good sign.
Something Isn’t Right (Charles Steinhardt)
A familiar puzzle type, but one that has lots of room for variation, and which tends to be fun both to write and to solve. Making the puzzle was Charles’s idea, and we did it in the last week when we found ourselves a puzzle short, despite that we then needed to find a whole lot of testsolvers to find false positives in it.
We swapped its answer with another puzzle when we found that VIEW is really hard to clue with five words, and its current answer can be made of five words, as long as one of them is allowed to be extremely obsolete. (“Do me hangen by the hals.” —Chaucer)
We Have To Go Deeper (Glenn Willen)
This puzzle originally came with almost none of the helpful markings that show you your options for moving from one level to another - those were only in the normal version. But testsolvers really liked the puzzle once it had them, so the hard version got them, too. The extraction also used to be considerably more complicated than it is now.
The puzzle still took many teams a bit of time, even though Glenn thought we were turning it into the easiest puzzle in the hunt.
The Futoshiki Kick (Charles)
Should I admit this? I’ll admit this. This meta was written two days before the hunt. In fact, it was written during the complete testsolving run.
Amos had written an interesting metapuzzle that plugged every answer into a different logic puzzle, so that you had to solve most of seven logic puzzles, with a hidden twist as well, to get the answer. Once I saw how long we were running, though, I decided quite unilaterally that the puzzle containing seven other puzzles had to go, even though it was a metapuzzle. Sorry about that, Amos.
So then we needed a metapuzzle, written really fast, that used seven answers that already existed. I asked Charles, and he delivered. Fortunately, the constraint on the answers — that they were all different lengths after you perform puzzle-ception — was still helpful.
That’s why the mechanism for inputting answers into the metapuzzle had to be that messy double-crostic pyramid thing combined with a decoding table. But despite that part, I think the puzzle worked well. And it’s interesting that it was a metapuzzle that tells you almost exactly what to do, and yet preserves the essential feature that you’re more likely to solve it as you get more answers.
Incidentally, because you read the answer from the 1s through 7s in the seven rows (the one instruction the puzzle doesn’t give you), that makes the puzzle possibly solvable as a drop quote when you have none of the answers. An ugly 7x7 drop quote without spaces, but it probably helps that it starts with “THE ANSWER”. I’m glad nobody solved it this way (that I know of). I think one team might have been trying to do it, though, because they called in “DISCOVER” followed by “SUPER DISCOVER”.
Dilation (Derek Kisman)
I dropped the ball on this puzzle for quite a while, letting it sit in my inbox un-testsolved for months. A couple of weeks before the hunt, I found out that the puzzle in my inbox was damn near impossible. But Derek managed to adjust the difficulty until we had reasonable versions of it for both hard mode and normal mode.
Apparently, many solvers knew or could infer what a GTO is. That’s good. I didn’t understand that clue at all.
At some point in making the final copies, I got confused and thought “exalt” made the obscure word UPRAISE from UPBRAIDS, instead of PRAISE. The normal version ended up with six blanks and the hard version with seven. A silly way for the hard version to become harder, but it doesn’t seem to have tripped anyone up.
Paint by Letters (Ben Aisen)
Some people liked this puzzle, and some people really, really hated it. No knock to Ben, who got us a very flexible puzzle on short notice that I for one enjoyed. I happen to like convoluted liar puzzles, but I can see why it would cause frustration: there are many opportunities to get an answer that’s very subtly wrong, and when you get one of those, you go to the effort of filling in the grid and you get nothing for your effort.
I had to basically argue with a team over the phone about this puzzle at one point: they were claiming that the puzzle was broken because they had a valid solution that didn’t give them an answer, I was trying to point out the contradiction in their solution, and they were having trouble understanding what I was saying because I had to convey something like “But then ‘N and C are false’ is false“, and you can’t hear quotation marks over the phone.
The advantage of this puzzle was that it could be adapted to basically any answer. (Not necessarily nicely, though. When the answer was FEDERAL AIR MARSHALS, we had it make a phone number to call for the answer instead, but even that ultimately required too large of a grid.)
As an editor, I should not have let statement A stay in the puzzle. It’s an instruction. If it’s a claim that can be true or false, then you can suppose it’s false, and then you can’t really say you’ve found a contradiction because the classical logic you would use to find a contradiction has gone poof.
I assume Ben’s reasoning was that A must be true because it is implicitly true of every logic puzzle. But if it’s false, it’s a non-logic puzzle. This all gets very philosophical, but in a way the puzzle has two solutions, and one of them is to not solve it. You don’t normally solve a puzzle by not solving it, but this puzzle effectively mentions the possibility right up front.
You might even suppose that there would be a reason for the puzzle to have no solution — perhaps that is part of the “puzzle-ception” you have to perform, or perhaps the instructions are a big red herring, hiding another puzzle within them. The flavortext I added about paradoxes (because Arthur says “paradox” repeatedly in the movie) didn’t give people confidence that logic would apply to the puzzle, either.
To change the topic, I had fun choosing the supposedly changeable letters here. If you try to perform puzzle-ception, you get statements including “The letters in ETOSP are HERE”, which I would call true, and “This puzzle contains as many DISAPPOINTments as a statue in Harvard Yard”, which would hopefully clue you in that you’ve got the wrong puzzle.
Zero Gravity (me)
This was my favorite puzzle to write. I hope solvers enjoyed it too. I think I heard from some who don’t like cryptarithms and therefore found it infuriating; I had it testsolved by someone who likes cryptarithms and thought it was great.
I had a team call about the fact that they couldn’t distinguish the Ns and Zs in sections 7 and 10. I replied with “Yes, those are both Ns and Zs. I’m sorry.” Then I realized I could have been misinterpreted and called back:
“I’m not actually sorry. It’s part of the puzzle and I’m quite happy about it, in fact.”
“We know what you meant.”
There actually is a unique solution to the Ns and Zs, up to an arbitrary swap of two 4s and two 2s which doesn’t change the answer.
Messing with Geometry (me)
The fact that this puzzle had to be the keystone of the entire hunt was quite terrifying, for a while.
I started out with the constraint that I had to make a puzzle with two versions, whose answers are similar-looking but very different. I also knew I wanted to have a puzzle that referred to the Inception scene where Ariadne folds up city streets into a cube.
That led to this puzzle idea, which seemed like it might be too constrained to create, but I was elated to find that the overlaps between three answer words and geometric words — in very particular places on the cube — would lead to the answer changes I was looking for.
And then began the unexpectedly harder task of writing the rest of the puzzle in a way that was solvable. The original version of this puzzle did not even tell people they were looking for geometric shapes; it only hinted at it in the title. I had testsolvers who managed that, remarkably, but I also had testsolvers who never picked up on that and just found all kinds of crazy dictionary words. It also did not say that the words had to be non-overlapping; I had assumed that would make itself clear when you found the right words, of course raising the question of what happens when you find the wrong ones.
Originally, I also had separate “hard” and “normal” versions, where only the “normal” version had the board surrounded by a bank of possible shapes to use. I felt at first like that weakened the puzzle — it was essential that you wouldn’t use all the shapes, because there was another version of the puzzle with different shapes, but this seemed to make it a bit too obvious that this was the puzzle with another version.
It took me far too long to realize that the hard version of the puzzle was broken. The correct process of solving could get you to a state that didn’t spell an answer if you were unlucky.
One problem, pointed out to me long before by Amos, was that the cubes in MAGNETOSPHERE (in the modified version of the puzzle) do not just spell the word SPHERE; they also spell HEPTAGON or SEPTAGON, whichever one you think is the name of the seven-sided shape. There was no way to rearrange MAGNETOSPHERE, within the constraints of the puzzle, to prevent this. I hand-waved this away by saying that no reasonable solver would prefer to take HEPTAGON over SPHERE, and even if they did they should eventually find that PENTAGON/SPHERE uses more cubes than HEPTAGON, and that makes it more “answery”. In some wibbly-wobbly way.
Much later, testsolvers found that letters in ELLIPSE, CURVE, and the A from the answer could spell SPIRAL. This was actually found by testsolvers in Normal Mode who had the shape bank, because the shape I had originally chosen as a “curve” could in fact be seen as a segment as a spiral.
But up until the day before BAPHL, I was still just trying to rearrange and re-rearrange the board until I could make a “hard” version with no shape bank and no red herrings. Then I had an emergency last-minute testsolver try it, and one of the first things he said was “Seriously? Septagon?” The next thing he found was SPIRAL cropping up somewhere different. So that was it for that version of the puzzle.
I hastily replaced it with the more-testsolved “normal” version. If it made it too clear that it was the modifiable puzzle, that was not something to be concerned about anymore, as by then we were much more worried that too few teams would be able to finish in 7 hours.
As it turned out, the normal version was the level of difficulty we should have been aiming for in Hard Mode all along. But there wasn’t time to write an easier Normal Mode version. So the end result was that both normal and hard solvers got a hard puzzle, instead of normal solvers getting a hard puzzle and hard solvers getting a broken puzzle. At least the disaster was averted.
There were only about 7 of us running BAPHL on site, and I’m extremely glad each person was there, because I think something would have broken if we had even one fewer.
30-ish teams is really an incredible turnout, and I hope BAPHL can get as many people involved when it happens again in the spring. Our biggest advice to Team Plugh, then, is to make sure they gather enough people who will be able to run the hunt.