B/MysteryHuntGuide

From LevWiki

Jump to: navigation, search

So, you're new. As you'll soon see, solving puzzles at the MIT Mystery Hunt is a bit different than solving the sudoku or jumble in the paper. The puzzles are longer, harder, and generally without explicit instructions.

Typically, puzzles come in batches of 8 to 10 called waves. Each puzzle yields a unique solution, which is a word or phrase. Puzzles are also arranged into metapuzzle groups, which (depending on the Hunt) may be the same as waves, or may have to be identified by some other method. The solutions of each metapuzzle group form a metapuzzle with its own unique meta solution. Generally solving a metapuzzle means finding some common theme among the solutions in that round, possibly in conjunction with auxiliary information such as the puzzles' titles. The metapuzzle may also be directly given to solvers, in the form of a rubric that requires the individual puzzle solutions as input before the final answer can be deduced, or it may just consist of the puzzle answers with no further information.

Solving metapuzzles typically makes subsequent waves available; normally each wave is also released at a specific time to any teams that haven't yet solved the meta to unlock it. There are generally 8-10 waves, and their 8-10 meta solutions form the meta-meta puzzle. Solving the meta-meta sends you to the run-around, the final race-like phase of the Hunt in which the leading teams follow a sequence of directions to find a single coin hidden on campus. When the coin is discovered, the Hunt is complete. It is unlikely our team will get to the run around, but don't feel bad - the prize for winning is having to create next year's Hunt.

#Puzzles ! How puzzles work

Most of the ~100 puzzles that we'll attempt as a team will take between 0.5 and 12 man hours to solve! They can be significantly hard. In most cases, the challenge of a puzzle comes from trying to figure out how it needs to be solved. By comparison, you already know the rules for solving a newspaper sudoku or crossword; solving it is a matter of implementing those rules. Mystery Hunt puzzles generally obscure these rules - the challenge is inferring them. Sometimes, though, puzzles can be labor-intensive.

Often there are multiple steps in a Mystery Hunt puzzle. For example, you may have a crossword grid and a list of clues, so obviously the first thing to do is solve the crossword; but then you may have to figure out what to do from there in order to obtain the final answer. Make notes of anything unusual you detect. For example (again), if you have a crossword, but you have to remove one letter from each clue to make it make sense, that's a pretty good sign you should write down the removed letters separately and see if they spell something.

Many puzzles have flavortext at the top - seemingly unrelated comments that actually may conceal hints about how to solve the puzzle. The title is also sometimes (but not always) a hint.

Some puzzles are action-based, not pencil-and-paper. For example, it's common to have a scavenger hunt, or to require teams to submit an art project, or to have a "party" to which each team must send one member (who should be on the lookout for any strange sights, sounds, or smells there).

#Mechanics ! Mechanics

Waves of puzzles are released to us online; each team has its own password to the Hunt website, which enables members to reach the puzzles that are accessible to that team. Our wiki main page will have a link to our Hunt site. In practice, we generally print out puzzles because most are easier to do on paper.

The team organizing the Hunt determines the process for reporting answers. The process used last year (which we expect will be repeated this year) is that if you want to submit an answer, you press the "submit answer" button on the puzzle's webpage and provide a phone number at which you can be contacted. Hunt HQ will then call you and you can report your answer. They may ask you for spelling; this is not necessarily a sign that you are right, wrong, or close. Answers are called in by phone, instead of by internet, so that solvers can't try out long lists of guesses. If you think you have an answer but aren't absolutely sure, please check with a teammate before submitting. HQ gets annoyed if they have to deal with lots of wrong answers.

Please remember to update this wiki frequently to report puzzles you are working on, are stuck on, think you have solved, or have verified with HQ. Also use the puzzle-by-puzzle chat rooms; even if nobody else is around working on the puzzle, you can report partial results, or things you've tried that didn't work. We want to share our knowledge and avoid redundant work. The Remote Control (a rotating job held by someone at our SF headquarters) will help make sure the Boston group isn't duplicating the work of the Bay Area group.

#Principles ! Things to keep in mind

There are several principles that are worth thinking about when trying to infer how to solve a puzzle:

  1. The Mystery Hunt is a marathon, not a sprint. Don't drive yourself nuts working on something for too long. Pass it off. Take a break. Seek inspiration elsewhere. Work in teams. Get enough sleep. Only bother if you are having fun. Trite but true...
  2. The solution must be unique. If your plan to solve a puzzle could yield more than one answer, you probably need a different strategy.
  3. When trying to figure out how a puzzle works, think about how constrained the puzzle-writer must have been. For example, if you have a crossword puzzle and your hypothesis about the puzzle's structure involves forming a paragraph by reading all the Across answers in order, that's probably not right (it would be too hard for the puzzle maker to make a crossword grid where the Down answers were words).
  4. Puzzle makers try to be clever, so be on the look-out for puns and nerd-humor. Puzzle titles and the context of their wave will often relate to their solutions, or the rules for attaining their solution. For example, in last year's Hunt, waves were named for characters one of which was "the Swedish Chef." A competing team cleverly guessed the meta-solution without solving any puzzles in that wave - "Chocolate Moose." As another example, we solved a puzzle titled "up down and all around" which featured photos of pipes, and had the solution "endoscopic."
  5. An ideal puzzle is elegant - it would be bad form if a puzzle demanded that you measure the location of letters on a page down to micron acuracy. Asinine rules are pretty uncommon. In particular, there are often hints as to what's important. For example, if there's a list of clues each giving a word, and the clues appear in alphabetical order, that's telling you that the order you get them in doesn't matter (but you may need to put them in another order). For another example, if the puzzle is a story with lots of awkward words or strangely structured sentences, then that's probably what you should be paying attention to.
  6. Part of elegance is that puzzles are compact - they usually don't give you superfluous information, so if you're stuck, see if there's anything in the puzzle you haven't used.
  7. Most puzzles are self-contained. While a rule may require you to get information off the web, or from physical places on the MIT campus, puzzles at the same meta-level tend be quite independent from one another.
  8. A puzzle maker aims for solvable puzzles - and will be sad if no teams are able to complete his or her puzzle. If a puzzle rule seems mind-bogglingly difficult, it probably isn't the road to a solution. That said, sometimes puzzle makers will underestimate the difficulty of their creations, and hints are occasionally distributed.
  9. It is easy to over-think puzzles. For example, last year I saw a page of various objects and thought the solution would lie in characterizing their symmetries (cubic, cylindrical etc.) and then decoding the mathematical representations of these symmetries. Turns out they were all objects from an Ikea catalogue, and their product names were what mattered. D'oh.
  10. Per above, often an image or element of a puzzle refers specifically to that object ("Ikea Pilbo Coffee Table" vs. "a table") rather than a generic representation of a table.

#Toolbox ! The puzzle-solver's standard toolbox

The following are commonly appearing puzzle types and tricks. See also our list of solving resources.

  1. Many common kinds of puzzles are likely to appear without instructions, with the expectation that the solver knows what to do. Some of these are well-known, others may be familiar only to puzzle nerds.
    1. Crosswords (ordinary and cryptic)
    2. Acrostics
    3. Cryptograms
    4. Anagrams / jumble
    5. Sudoku
    6. Kakuro (cross sums)
    7. Find-the-differences
    8. Connect the dots
    9. Mazes (2D or 3D... or more)
    10. Jigsaw puzzles (again, can be 3D)
    11. Paint by numbers
    12. Tableau logic puzzles
    13. Battleships
  2. Often you'll have a list of words, or of collections of letters, or even of sentences, that require some manipulations to turn into something reasonable. Here are some standard kinds of wordplay...
    1. Anagrams
    2. Homophones
    3. Synonyms
    4. Antonyms
    5. Pairings (words that go with another word to make a common phrase)
    6. Inserting or deleting letters
    7. Spoonerisms
    8. Hidden words (inside another word)
    9. Charades (concatening words to make another word)
    10. Reversals
    11. Puns
    12. Rearranging spaces or punctuation
  3. You will often have to order parts of a puzzle. For example, a puzzle may be a list of verbal clues or pictures, each yielding a letter, which must be placed in the correct order to lead to the answer. Commonly used orderings:
    1. Anything numerical: years; seasons of a TV show; MIT building numbers; lengths of things. If a puzzle yields (say) a list of words and they're all different lengths, that's a sign length may be the way to order them.
    2. Spectral ordering - ROYGBIV often orders puzzle components.
    3. Alphabetic
    4. Chronological - days or years ...
  4. Also, note some common ways of extracting letters from words. Often a puzzle will generate a list of words, and the final solution is obtained by extracting one letter from each of them:
    1. First letter of each word
    2. Last letter
    3. Middle letter (if odd length)
    4. Diagonals - first letter of the first word, second letter of the second word, etc.
    5. Indexing - take the nth letter of each word, where each word has its own n and this value either is given to you or must be inferred from some other clue
    6. Other special letters - for example, if each word has exactly one doubled letter, try taking that letter
    7. If these methods don't seem to give you anything, keep in mind that you may also have to read the result backwards
  5. Lastly, most of the time, a puzzle will use at least one or more encoding. Luckily for you, the majority of encodings used in the Hunt will come from one of the following - so look for patterns in your puzzle that might fit one of these codes:
    1. Numeric encodings
      1. numeric substitutions - A=1, B=2, C=3 is just the start of it.
      2. 5-digit binary - a popular encoding in which sequences of 5 dichotomous cues indicate (up or down, one or zero, light or dark) a number 0-31, which is typically converted by a numeric substitution into letters
      3. ternary - similar, with a sequence of 3 cues, each of which can assume 3 different values, encoding a number 0-26, just enough for the alphabet and a space
      4. hexadecimal - numerals 0-9 plus letters a-f can encode 16-ary letters or numbers
      5. ASCII numbers range from 32-255 and encode computer characters. ASCII values 65-90 are the uppercase alphabet, and 97-122 the lowercase. Be on the look out for numbers in these ranges. They may also be represented in hexadecimal.
    2. Pattern encodings
      1. Morse code - be on the look out for anything that could be considered dots or dashes
      2. Sign Language - look for hands or more abstract renderings of hands
      3. Semaphore - This international flag language can be encoded as angled lines within a circle (The hippy peace sign is stacked semaphore characters N and D for "Nuclear Disarmament"). These can be put subtley into images, look for hands of clocks, bullet holes in targets, etc.
      4. Braille appears as 2 x 3 arrays of binary elements (ups v downs, darks v lights), and can be sneakily embedded in images. Any 6-digit binary encoding is worth trying as braille, but especially if things come in 2 x 3 arrays.
      5. Nautical flags - correspond to letters
    3. Word encodings may be used to hide secret meanings or constrain solutions
      1. Alphabets such as Alpha-Bravo-Charlie
      2. sequential word patterns such as subsequent words starting with later and later letters of the alphabet, or alternating between vowels and consonants, or starting with plosive consonants only, etc. etc.
      3. word trivia - words that can be typed only with one hand, or words containing all vowels frequently show up as constraints.
    4. Picture encodings
      1. As mentioned above if you can identify the specific object shown, that may be the most relevant info.
      2. Indexing - If an image contains repeated identifiable elements, such as 5 cheshire cats, you can count 5 letters into the name of the identifiable element: 5 cheshire cats = H.
      3. Stenography - clues or data can be embedded in the image via manipulation. For example, the mathematical difference between a doctored image and the original may reveal a new image.
      4. meta-data - computer files of images can contain comments and file-type information that doesn't display in image viewing software. Try opening the image in a text editor. (Actually a similar comment applies to HTML files; if the text seems too straightforward, read the HTML source)
    5. Audio encodings
      1. backwards playback
      2. repeated elements in audio can be used to make other encodings such as 5-digit binary
      3. spectral analysis