Wrap-up: The Making Of Making Things Difficult
This is not a solution to the puzzle, but provides notes from its poser. This type of answer has been approved by the community.
Caution: This post contains lots of spoilers, right from the start.
I drafted a lot of ideas for various titles in this FTC, but I really wanted to make a small metapuzzle if I could. When I saw "Making Things Difficult" as a title on the list, I latched on to that one as my metapuzzle title, and with transadditions (anagrams plus a letter) as a mechanic that provided flexibility and a possible hook back to the title.
The Mohs scale was an early idea, because I wanted to make puns about how "hard" the meta was; one trip to Wikipedia later, I had decided quickly on what would end up being the final payoff of the puzzle set. (Some puns that did not make it into the set: "scratching the surface", "on a scale from 1 to 10", several 'rock' puns. It's probably a good thing that these never came to light.)
There was originally an attempt to use transadditions of other Mohs scale minerals, but ultimately I scrapped that idea and tried the mechanic seen in the puzzle. I think I prefer the method I ended up using; it doesn't require as many answers, and it was less restrictive. (Answers that were in the original list included CANTICLE (puzzle would have referenced A Canticle for Leibowitz) and the LEFTORIUM, Ned Flanders' southpaw store from The Simpsons.)
After changing the meta mechanic, I generated new answer lists. The individual puzzles came about from looking for satisfying pairs between my generated answer lists and the FTC titles.
The mechanic from Factoring The Time was one of the first ones added to my list of possible subpuzzle ideas. To me, clocks in puzzles almost always guarantees how the final answer is extracted; the comic gave a really nice way to go from clock to numbers to clock again, on the way to the answer. (There's a joke in my puzzle hunt team about how to use clocks in different ways, so that's why this mechanic was at the top of my list.)
Keyboards are Disgusting was the last puzzle to be made. The other puzzles were all mathematics-based in some way, and I wanted to include a word/cipher puzzle somehow. The keyboard cipher puzzle was on the draft list, and I found an answer I was happy with on the list.
Snacktime Rules is directly inspired the comic and the belief that the puzzle
set wouldn't be complete without a strictly mathematical puzzle.
I really like making grid deduction puzzles, and a nonogram puzzle gave a lot of flexibility for an answer. Originally, the griddle pun was just meant to play on grid and grill/grille from the comic; when I found out that one of the many alternate names for the nonogram puzzle type is 'Griddler', it was just icing on the cake.
Creative and Logistical steps
Creativity and logistics ended up somewhat intertwined, by the nature of the meta mechanic restrictions. Note that there are even more explicit spoilers in this section.
Making Things Difficult
After dismissing a mechanic where all the answers were transadditions of Mohs minerals, I landed on the idea of having four subpuzzles where the answers clued NINE and MOHS using both their initial letters and the transadditions. I wasn't sure at first how unique I would be able to get the transadditions to be, so I tried to keep as many restrictions as possible to keep things on the right track: for each of the answers, the transaddition used is either the only English word that is a transaddition of the answer, or is by far the most common. With that restriction and with the theming, I hoped that would be enough that there would never be a wrong transaddition suggested. (If anything, the transadditions ended up being so strongly clued that the fact that the initial letters of the answers spelled NINE was completely missed as solvers jumped directly to the Mohs scale and to the final answer. This NINE was supposed to hint towards corundum, and towards the 'appropriate length' involving adding a letter.)
I wrote a Python program to go through the UKACD word list and output words that, say, began with N and had a unique transaddition in the list, where the added letter was M. I filtered these later through https://nutrimatic.org and threw out any where there were other English transadditions that were more common. (The answer to Making Hash Browns has five other transadditions according to nutrimatic, but they are four capitalized names (one Greek, one Jewish, and two Indian) and one Latin scientific name.)
Factoring the Time
Logistically, I would have preferred if I didn't have to rotate any of the clocks, but unfortunately the prime positions that would spell U and O don't multiply to valid times. (e.g. 22:79) Since some of them had to be rotated, I decided to rotate a lot of them, so that it didn't seem weird/inconsistent that just a few were rotated. To distinguish between the 12-hour time and the 24-hour time, I was originally planning to use just a simple P.M. indicator, but I decided to go with the "Think fast" line from the comic instead, which was specifically used after the clock was switched from 12- to 24-hour.
Keyboards are Disgusting
xkcd has been making jokes about Dvorak for years (personal fave) There's lots of other keyboard layouts, but I tried to stick to non-QWERTY-based Latin-script keyboard layouts from the linked Wikipedia list.
By losing a letter each time, it meant that this cipher wasn't just something to dump in a cryptogram solver. I was also happy to realize how much the layout changes were going to mess with the punctuation, turning commas into letters and letters into semicolons, even as the spaces were conserved normally. The actual message wasn't written with a particular story goal in mind, but I wanted to make sure each line had the new missing letter removed from either an obvious word or from a word that is clued from the context of the previous line; each new layout starts with two words ending the sentence from the previous line. With that kept in mind, it was mostly free flow.
In an early draft, I thought about doing this with just QWERTY, Dvorak, and Colemak, and having the story involve the narrator relearning the layout as it changes each time; in this way, the cipher would be changing each time, starting with QWERTY->Dvorak, and then Dvorak->Colemak, etc. I decided that was probably more confusing than interesting, and changed the text to always use QWERTY as the basis with a greater variety of other layouts.
The numbers were a late addition; originally, I was just going to let the changes be sudden and leave the solver to figure them out. But I was worried that not having easily defined start points and end points would be discouraging for the later sentences, which still look like nonsense once properly deciphered. So I added in the A1Z26 codes for the first two letters of the name of the keyboard layout, hoping that with QWERTY and Dvorak (and possibly Colemak) as well-known, the others would be meaningfully distinguished. As it turned out, phenomist solved this section almost completely before realizing what the A1Z26 codes meant, so the first draft wouldn't have been too bad. Still think it was a good idea to add them, though.
My only rule for making this one is to have a mix of easily recognized traits, simple but probably new calculations, and one hard calculation. By 'easily recognized traits', I mean ones that would have sequences in OEIS, like this or this. I wanted the other calculations to be just weird enough that they probably didn't have OEIS entries, and I wanted most of them to be doable by hand. (For example, 'binary palindrome' is really easy to do by hand, once you realize it's faster to work backwards by simply generating the binary palindromes in order until one is more than 99.)
The binary-to-octal one almost definitely requires a computer; there's some shortcuts that can narrow it down by hand to at most a couple dozen candidates, but finding the factors of 0o1100001=294913 is not something that I imagine has an easy shortcut. (Though looking at the answers now, I realize that if one were to assume wrongly that they only had to look at the prime numbers, they'd still have gotten the right answer; the first composite number to give a prime binary-as-octal is 115.)
Making Hash Browns
I only briefly considered trying to clue the answer pictorially in the nonogram, but that would have made the grid very large, and I'm not sure how many people would recognize the land between two rivers from a pixellated attempt to draw the rivers. So I decided on 3x5 letter shapes. At first, the letters were in the black cells and the letters were aligned, but that made the nonogram too easy, since so many of the rows/columns had a unique solution. Then I tried staggering them in a couple ways, before choosing the version seen in the puzzle. At this point, I switched which cells were shaded and which cells weren't, to make the letter shapes less obvious while solving the grid. Finally, I added some symmetric salt-and-pepper in the unused corners, since once again, the puzzle was made far too easy by having the edge rows/columns as single blocks of cells. I was using http://a.teall.info/nonogram/ to quick-check solvability at each point, and then solving myself to check difficulty/strategies.
Most of my resources are mentioned above. Wikipedia for keyboard layouts, OEIS for parts of Snacktime Rules, my own Python code for another part of Snacktime Rules, and http://a.teall.info/nonogram/ for quick-checking nonograms. I wrote my own Python code to check the UKACD wordlists for possible answers to feed the meta, and used https://nutrimatic.org/ to check those answers.
This puzzle had a few typos when it first went up. In one case for Snacktime Rules, I forgot that $98<99$, and so I appended a condition to fix the answer counts. There's also some typographical errors in Keyboards are Disgusting, but luckily those didn't affect the solve majorly.
Your thoughts/mental process
Most of the time what was running through my mind is, "Gosh, I hope this is actually as fun to solve as I think it is." My great fear is that I'll make a puzzle that I think is interesting and fun, but everyone else finds mind-numbing and tedious. Luckily, the puzzles are all pretty small, and often have a way to write a program to help with parts that might otherwise be tedious.
The meta ended up being overclued, but I don't think it suffered from that in the same way that it would have really suffered from being frustratingly underclued. If I did it again, I wouldn't change this meta, but I'm definitely planning a harder one for the future.
This was my first meta posted to PSE, and I'm pretty happy with how this all went.