I appreciate great puzzles in Games and other puzzle events, though I have a great deal to learn about writing such puzzles myself. Here's some of what I've learned:
- Perhaps most importantly, your goal as a puzzle-writer is not to write hard clues. It is instead to write fun clues. Thus layering complexity, making things insanely hard, is not necessary or even desirable. Keep your clues simple.
- ACME, three-time Mystery Hunt authors, have the rule that once a puzzle is constructed, it is never made harder. I think this is a good rule of thumb. The puzzles we wrote for the 2004 Mystery Hunt certainly broke this rule and were poorer as a result. One of my teammates on that effort later characterized it as our collective feeling that we had lucked into winning the hunt and that the other teams were smarter and had better puzzle skills, so we had better write really hard puzzles. We should have kept it simpler.
- One rule we had in Genome that was vital in running a smooth hunt was that every puzzle had to be tested in its final form. This was really time consuming, but very necessary. We were not permitted to test solve, then "just make a little change" and create the puzzle. We had to solve it again in final form. This is the single biggest reason that puzzles are broken on the road, and we didn't have any inconsistencies in our puzzles as a result of this policy.
Personally, I found puzzle writing to be by far the hardest part of running a Game. It was not the most time consuming part, but it was the hardest part, because you can't budget time to "think up a great puzzle". Sometimes they come to you when you sleep; sometimes you don't have to think of them at all and they just appear in your head; sometimes they take days of concerted thought. I have a great deal of respect to those who can repeatedly crank out great puzzles.
I have written puzzles for two events:
- Chasing Pauling. There are actually 5 bridges over 101 near the zone of interest. The first was too far south, so we planned to use the 4 northernmost ones. The southernmost of those, however, was completely blocked off (and COVERED in trash). Thus we compressed to two letters each on two bridges. We were pretty concerned about disrupting traffic (given that a playtest team completely stopped a lane of traffic), but it worked out OK in the main event.
- Tiles. I thought of this one while swimming laps and staring at the tile cross on the side of the pool. Gordon Dow helped with making the first draft a bit easier and wrote the flavortext that we ended up using.
- Minigames (on Angel Island). "Cars" was taken from the interactive-fiction game "The Magic Toyshop" where it was known as "Dodgem"; "Words" was a tic-tac-toe analogue (I believe the wordlist was found here; "Matchsticks" was the well-known game of Nim; and "Dots" was also taken from "The Magic Toyshop" (and adapted to a larger grid by Moe Badi).
- Wordlists, inspired by Roger Barkan's outstanding "Poor Constitution". Not surprisingly, no automatic procedure exists (that I know of) to automate the process of writing these. The first draft tried to get all the colors of the rainbow, but I think black/white worked out quite well. Gordon Dow wrote two of the lines ("TAMIL LIEGE ..." and "SICKS TINA ...").
- Ducks and Fish. Many teams, as we expected, only found one half of the puzzle. XX Rated short-circuited this puzzle by assuming it was a historical knight's tour, which it was (this particular knight's tour was constructed by al-Adli ar-Rumi, "who flourished around 840 and is known to have written a book on 'shatranj' [chess]".) The notation was taken from the Knight's Tour Crypt used by the NPL.
- Balls - so disappointing that this clue did not work out. This was the first clue I wrote.
- Voting Booth, inspired by System's Twilight. I think we hit the right level of difficulty on this one. In retrospect teams pointed out (as we expected) that they could have reverse-solved this one just by immediately noting which squares would not be filled in, but no one figured that out at the time. Heh. Lakeside School was my high school buddy Kjell's elementary school; it was a good landmark on a road that otherwise had none.
- Ribbons. This was the last clue we wrote; as most teams reached it in the middle of the night and after a tough clue we wanted something that was a little simpler and a quick solve.
- K. We were pretty concerned that the right radio stations would be available to solvers, but that didn't turn out to be a problem.
- 2 CDs. This was a pretty tough puzzle in the middle of the night. Only 5 teams saw it ...
- Stocks ... and then this puzzle immediately after was a big blow. This was one of our hardest puzzles. I think many felt it was a little unfair, especially with the twist at the end, for which I apologize. Elegant (I think), but tricky.
- A Cappella CD. I redesigned this 5 times to accommodate a changing route. Team Advil had horrible problems with this clue and we felt it would be quite difficult (but it was a daytime clue so that bystanders could help), but our testsolve teams and the teams that saw it had little difficulty, and I think it came out A-OK. We ended up picking the easiest tracks we possibly could (definitely many were trickier than the ones we used). This was the right choice.
- Decoding IG1 (metapuzzle). This puzzle didn't work out too well. One of the huge problems was that it was unclear which way the characters were oriented. While the two biologists on Team Advil were confident that we'd chosen an orientation that made sense, some of the players vehemently disagreed, and we should have made it much clearer what the correct orientation was. This soured the puzzle for teams that tried it. I am still fond of the puzzle - having solved it myself (since I did not write the final text) I thought it was a lot of fun. But at the end of a Game with no sleep, it might have been a little too hard. I generally like metapuzzles very much; the tricky part is to ensure that it can't be solved early while still letting teams make progress on a possible solution. We accomplished this by not revealing the identities of each character until they reached the final location; we felt it was possible to solve the puzzle knowing the identities but not possible (24 times harder!) to solve it without them. Mystic Fish could have short-circuited the entire game (look near the bottom) if they'd only listened to Larry Hosken.
John's Game Home Page
› NIT 2000
› Wonka Tour
› 420 Game
› 2001 SINS (GameNY)
› Blau Foundation
› Shelby Logan's Run
› The Genome Game
› Justice Unlimited
› Griffiths Collection
› Mooncurser's Handbook
› World Henchmen Association
Games (< 24h):
› Apprentice: Zorg
› Smaller Events
› MIT Mystery Hunt
› Microsoft PuzzleHunt
Other Game Sites
... not quite The Game
› now in Slovakian!)