Wordsearch

Wordsearches are a type of word puzzle involving a grid of letters in which a series of words can be found going in various directions. Despite their simple appearance, wordsearches are a relatively recent invention, only showing up for the first time in 1968.

Background
The origin of the wordsearch is somewhat disputed, as while it's widely considered to have been first published in English by Norman E. Gibat in 1968, it is possible that the puzzle was published earlier in other languages, or patented earlier in English. Of particular note are Pedro Ocón de Oro, a Spanish puzzle maker who worked for multiple Spanish-language publications throughout the 50s, 60s, and 70s and published something called "Sopas de letras" in the 1970s; and James Patrick Carr, a faculty member at the Villa Grove High School in Illinois who may have attempted to patent the word search under the name "Slate R Straight", only to have the idea swiped by the patent company.

Regardless of where the true origin lies, Gibat's wordsearch was first published in the Selenby Digest, a small publication distributed in Norman, Oklahoma. According to Norman, the goal was to create a puzzle that could be solved in line at a supermarket, hence the choice of publication source. The original puzzles also were locally-themed, as the first two had word lists consisting of Oklahoman cities and Norman's street names, respectively. As the idea grew in popularity (particularly among teachers who saw it as a good resource), an unknown individual sent the idea in to a higher-level publisher for syndication.

Puzzle Application
While a traditional wordsearch consists of a square grid, a list of words to find, and the expectation that words are found in one of the eight orthogonal directions, these "rules" are often broken for the sake of a more interesting puzzle, particular when it comes to hunt puzzles.

Grid Shape
The original wordsearches (and many modern wordsearches published in newspapers) were arranged in squares or close-to-square rectangles. This is possibly due to space constraints, as publishing a puzzle among other articles or in a single-page collection of other puzzles requires careful planning. A single, evenly-sized puzzle can be resized to fill a particular space, and having as many puzzles as possible be rectangular allows for the most economical use of space. Knowing that a puzzle is bound to this particular space is also something that can be comforting to solvers, as a small grid tends to imply a quick solve.

This is not the case with hunt puzzles, which are often published on a webpage or otherwise less constrained by space. The result is that puzzle writers have the freedom to organize their letters into any form they please, which can produce some impressively large and/or specifically-shaped puzzles, evoking shapes, cars, or even animals. While these uniquely-shaped grids may not always have a purpose in the unique solve of a hunt wordsearch, they will certainly leave an impression on the solver.

Conversely, some hunt puzzles choose to stick with smaller, traditionally-shaped grids, but play with the boundaries of that space. A small, square grid may not have wide enough dimensions to fit particularly long entries in, forcing solvers to continue partial answers outside of bounds of the grid, or wrap around to another side of the grid to continue travelling.

Grid Fill
Most wordsearches will be contiguous, filled with letters from one end of the grid to the other with no interruptions (barring aforementioned odd grid shapes). Similarly, entries don't get interrupted; whenever an entry starts, it's expected that the same entry will end somewhere else in the grid without breaks.

Obviously, this is an easy rule to break for the sake of a twist, as it is actually quite common for wordsearch grids to contain gaps where single letters should be or where entire chunks should be. This forces solvers to work with partial information, filling in the missing spaces as they go.

While less common, it's also possible that certain spaces will be replaced not with blankness, but with other non-letters like emoji or Unicode symbols. Alternatively, one could replace certain letters with incorrect (caesar-shifted, cryptogrammed, or just flat-out wrong), but this risks preventing solvers who are used to close-but-not-quite red herrings showing up in the grid from recognizing the twist.

Word List
Word lists are usually used to show solvers what they're actually meant to find within the grid, and are almost always present in newspaper or collection wordsearches. It prevents setters from having to worry about accidentally letting other words show up in the grid and it prevents solvers from having to worry about the human instinct to find patterns even when none are present.

Alternatively, writers can choose to obfuscate their word lists, or remove them entirely, in order to create a puzzle that requires more thought. Most commonly, reduced and removed word lists appear in wordsearches that have a theme to their entries, so as to not feed into the aforementioned pattern-seeking behavior. For example, a puzzle having the seven colors of the rainbow as entries may choose to only show first letters, only show lengths (usually keeping the entries listed in alphabetical order to disambiguate identical-length words), or omit a list entirely and instead emphasize colors or sets of seven in the flavortext.

Word Direction
While traditional wordsearches are limited to the eight orthogonal directions (N, NE, E, SE, S, SW, W, NW) for entries to travel, a twist may allow them to go at sharper angles, or change direction entirely throughout their path. In many cases, these gimmicks will be either explicitly stated, or clued, as "word travel in straight lines" is one of the core conventions of a wordsearch, and changing it doesn't otherwise change the outward appearance of the puzzle. Making sure this change is either clear or satisfying to discover is key to making it work in hunt puzzles.

Snaking Puzzles
Allowing for directional changes within a wordsearch has become popular enough to spawn a subgenre: snaking puzzles. These types of puzzles only allow for movement in the four cardinal directions (N, S, E, W), but the path taken can turn 90 degrees any number of times from start to end. When in app form, snaking puzzles also tend not to include word lists, instead providing starting letters, forcing solvers to find the correct words from only initial positions. Usually, there's still a unique solution that uses every letter in the grid, so words to in fact have to be "correct", and not just coincidentally valid.

Hidden Messages
One thing that is much more common in hunt puzzles than any other type of word search is the inclusion of hidden messages only viewable after the puzzle is solved. The most well-known (and most well-used) way to hide these messages is by having the unused letters from the puzzle spell them in reading order. This also allows for the setter to provide hints to observant solvers towards the general area of missing entries, as spotting the message before the puzzle has been completely solved will inevitably show which letters still need to be used up before the message is the only thing remaining.

Other less-common-but-still-documented ways of presenting a message include having entries reading in a particular direction either spell out a message via the entire words/first letters, or having a message spelled out via letters at word intersections.

Strategy
As identification of a wordsearch is minimal-effort (aside from cases where a wordsearch has been massively obscured, or another puzzle has been altered to the point where it may be confused for a wordsearch), strategy here will be primarily focused on solving wordsearches.

Solving Traditional Wordsearches
Traditional wordsearches (read: containing a word list, no gimmicks related to word direction or grid fill, no hidden messages) have one primary challenge: finding the words. Depending on the size, this can be a simple or disturbingly difficult task. In general, initial focus should be on entries that contain rarer letters (i.e. high Scrabble point-value letters like J, K, Q, X, Z). Puzzle writers will tend to fill gaps in the grid based on relative frequency of other letters, meaning these rare letters will stay rare (unless the word list contains a lot of them anyway).

Once some entries have been found, focus on searching areas where very little has been found. While not a rule, it's common for writers to spread their entries out, and try to use as much space as possible. This means that large, unused spaces in the grid should be treated with suspicion.

Lastly, if you're missing a small selection of entries near the end, pick out a set of 2 or 3 letters from a given entry and search for that combination in the grid. Individual letters (especially common ones) are difficult to use as starting points, but combinations are much rarer in general, and can usually be used to narrow down possible search locations. Additionally, solving digitally (such as through PDFs) can have its advantages. Searching for/highlighting all instances of a given letter can make those specific entry searches go much faster.

Notable Examples

 * - Contains both a length-only word list, and a unique twist on entry direction: all of the words found in the grid can be paired with another word to form a common phrase. Those extra words also describe the shape or direction the word was found in, with entries like TIDAL WAVE (TIDAL in a zig-zag pattern) and SIERPINSKI TRIANGLE (SIERPINSKI arranged so that when all letter are highlighted, they draw a triangle)
 * - A system that can generate 100 million word searches (without word lists). Solvers need to recognize the theme of both prime ministers as entries and prime numbers being used in the generation process. They also need to find the hidden message between the entries telling them their next steps (one of which includes determining exactly how the grids are generated, mathematically).