Diagnostic puzzles. The challenge is to collect enough information about a problem to identify an item or type.
Research puzzles. The player needs to look up information in various sources in order to assemble a complete picture and learn how to proceed.
NPC manipulation. The player has to get an NPC into a particular mood or state of mind in order to accomplish the desired outcome. Here are some games that make use of persuasion; conversation games in general are also often of this type, though they may be more story- than puzzle-oriented. Make It Good is the grandmother of all NPC manipulation games, and ferociously difficult.
Resource management. Game requires you to assign the use of a limited resource (time, space, money, etc) carefully in order to win. Scheduling puzzles where you have to pick where to spend your time are a subset (though they’re seen more often in visual novels than in classic IF).
Geographical puzzles. Puzzles involving management and manipulation of space. A common example of this puzzle type is the maze: simple mazes are often deprecated in modern IF criticism because they tend to be easy but boring to solve. However, some games do offer interesting variations in which the solution.
Timing and sequencing puzzles. The player may know roughly what needs to be done, but the challenge is to get actions into the right order or optimize performance so that they occur quickly enough.
Classic logic puzzles. Sometimes these are reuses of old chestnuts like the fox-goose-corn puzzle or the puzzle of finding four cups of water with only a 3-cup and a 5-cup measure. Occasionally they get a bit more sophisticated.
Riddles. The player is asked to think of the solution to a riddle in order to move forward. These can be arbitrary and difficult to solve and many are now deprecated.
Wordplay. Puzzles turn on manipulating the way objects are described or some other aspect of the text in the environment. Anagramming, letter removal or addition, puns, and classic sayings have all appeared as the basis of wordplay puzzles.
Cryptographic Puzzles. The player must solve a cryptography challenge, either by himself outside of the game or by using some in-game cryptographical mechanism to resolve it.
Invented Language. The player must learn an invented language and either translate or speak in it.
Approaches to Puzzle Design and Hinting
Multiple solutions. Many or even most puzzles in the game offer more than one viable solution. This feature often (but not always) goes hand-in-hand with a simulationist approach to world modeling, since rigorous simulationism makes it possible to offer the player solutions which haven’t even occurred to the author.
Adjustable puzzle difficulty. Games may allow the player to choose a harder or easier game mode at the start of play. These games approach the adjustable puzzles in different ways: in some cases, choosing an easier version means some puzzles are already done for you; in others, the same puzzles occur but are simplified in the easier settings.
Built-in hints and tutorials. Games sometimes provide built-in hint systems, tutorials, or even built-in walkthroughs; hints may be static or may adjust themselves depending on the current game state. Another popular approach is to provide per-room hints that address only problems the player can see at the moment.
Single-puzzle games. Game turns on resolving one major puzzle. There may be a number of subcomponents to this puzzle.
Achievements and easter eggs. The game has optional features (either explicitly called out as achievements, or concealed as easter eggs) to reward the player for special invention or cleverness.
Graded Success. Instead of making the player solve all the game puzzles to win, these pieces assign a range of endings (from abject failure to best-possible-outcome) depending on how well the player did and which problems he solved. Sometimes this is expressed as a numerical score, sometimes not.
Play once. The player is discouraged from ever UNDOing a move or restoring a saved game (other than as a way to take a break from play); instead, he is supposed to try to win on the first play-through. Or, in a more story-based game, the player is encouraged not to restart or replay after finishing.