Choice is an important concept in game design. I would go so far to say that choice is the essential difference between movies and games; choice puts the audience in control.
There is a lot I could say philosophically about choice, its impact on gameplay, and how developers both use and abuse it. However, today I'm going to zero my focus on a very specific and limited aspect of choice, options.
Options are generally overlooked as an aspect of game design. Who cares about the player's ability to tweak graphical, audio, and controller settings when there's a game to be made? Yet, options define the player's experience almost as much as any other aspect of the game.
For example, a bad mapping of actions to the buttons/analog sticks on a controller can make a game unplayable. Despite this, with surprising frequency many games only implement the bare minimum for controller settings. In such cases you might find one or two preset button layouts with a single toggle for inverted look on the analog sticks. If a player does not find these preset layouts comfortable, intuitive, or enjoyable the game is ruined.
It is therefore critical to allow players to customize these types of user interface issues to their specifications. Just as a movie must be in focus, and a book must be printed legibly, a game needs to have a user interface that does not separate the audience from the medium. Options go a long way in attending to the huge diversity that exists in personal preferences.
So, let's examine a few examples of options whose presence or lack thereof added to or detracted from a game.
Case 1: Tetris
An interesting aspect of Tetris is the options it had available. At the time very few games had options, and those that did were fairly limited. Tetris, however, had options for changing music, adding a handicap, and even starting at a more difficult level. All of these helped the replayability of the game immensely. As spiffy as the music was, being able to turn it off helped preserve sanity after several hours of play. Once you got very good, being able to up the ante early on erased the slow, early part of the game. Tetris, and a few other pioneering games, redefined what it meant to have options.
Case 2: Smash Brothers
Smash Brothers is a game renowned for its customizability, with one exception. While all other items and settings have always been ridiculously thorough and deep to the point of insanity, each game in the series has overlooked adding an option to select which Pokemon will come out of the pokeballs.
This is a very, very small detail, and detracts only a little from the series, but has nonetheless remained a very obvious opportunity that remains overlooked. Two of the games even feature challenge modes where the developers restrict Pokemon themselves, tormenting players who with to have that power.
Case 3: Battle.net 2.0
Battle.net 2.0, and specifically RealID, is the catalyst for this entry. RealID is the ability to "friend" other players on a first-name (rather than anonymized codename) basis, allowing you to see them as online no matter what game they are playing. The ability to track good friends and relatives in this manner while you're playing games is a neat feature.
The problem with RealID is in its options. RealID is an all or nothing feature-set. You can't friend someone through RealID and not have them appear in your list of friends with their full name followed by their account name. You can't choose not to have your list of RealID friends visible to all your RealID friends. You either add friends through RealID and subject yourself to all the features you like and all the features you don't like, or you opt out entirely.
This failure to break RealID down into separate, optional features has been a major sticking point for many players. In fact, it has been such a problem during testing that the developers had to intentionally break other options in order to force players to test the RealID system. RealID simply wasn't something the testers wanted to use because they couldn't turn off the features they didn't like.
The thread which joins these cases together is a simple principle: options should allow a player to do something they want. Tetris players will often want to get to the exciting fast-paced portion of the game, and so giving them an option to skip to that portion makes sense. Battle.net 2.0 users may not want to identify their friends by their full names, or for their friends list to be readily visible, and so failing to provide an option to disable these features is a critical oversight. Players that can mold the game to their needs will always be happier than players who are forced into specific implementations.
So when deciding on options, be very careful when making assumptions about player needs. Considering everything individually, and whether it makes sense for certain things to be immutable or joined together. If a deadline is your reason to avoid implementing an option, so be it, but make certain that you have at least that as a reason rather than nothing at all.