Rogue (video game)

A procedurally generated dungeon in the 1980 version

In the original text-based versions, all aspects of the game, including the dungeon, the player character, and monsters, are represented by letters and symbols within the ASCII character set. Monsters are represented by capital letters (such as Z, for zombie), and accordingly there are twenty-six varieties. This type of display makes it appropriate for a non-graphical terminal. Later ports of Rogue apply extended character sets to the text user interface or replace it with graphical tiles.

The basic movement keys (h, left; j, down; k, up; and l, right) are the same as the cursor control keys in the vi editor. Other game actions also use single keystrokes—q to quaff a potion, w to wield a weapon, e to eat some food, etc. In the DOS version, the cursor keys specify movement, and the fast-move keys (H, J, K, and L) are supplanted by use of the scroll lock key.

Each dungeon level consists of a grid of three rooms by three rooms (potentially); dead-end hallways sometimes appear where rooms would be expected. Lower levels can include a maze in place of a room. Unlike most adventure games of the time of the original design, the dungeon layout and the placement of objects within are randomly generated.

The concept of Rogue originated with Michael Toy and Glenn Wichman.[2] Toy grew up in Livermore, California, where his father was a nuclear scientist. Once a year, his father's workplace allowed employees' families to visit, which included allowing them to use the facility's mainframe system to play games. Toy took interest in the text-based Star Trek game (1971), which represented space combat through characters on screen, and required players to make strategic decisions each turn. Toy took to learn programming and recreate this game on other computer systems that he could access, including the Processor Technology Sol-20 and the Atari 400.[3]

Toy subsequently enrolled in computer science at the University of California, Santa Cruz (UCSC) in the late 1970s. Working first on UCSC's PDP-11 and then its VAX-11, Toy began exploring what games were available over ARPANET, the predecessor of the current Internet. One game that intrigued him was Colossal Cave Adventure (also known as Adventure) (1976) by William Crowther and Don Woods. Adventure, considered the first text-based adventure game, challenged the player to explore a cave system through descriptions given by the computer and commands issued by the player. Toy was impressed by the game and started writing his own.[3]

Toy met Wichman, another student at UCSC who was also writing his own adventure game. Wichman had created his own variations on traditional role-playing games such as Dungeons & Dragons while growing up. Wichman chose UCSC specifically to study game design to become a board-game developer, and this led him into the computer sciences to get the opportunity to play and develop games. The two became friends, shared an apartment, and challenged each other with their own adventure game creations. Of the two, Toy was more proficient at coding, while Wichman had a better sense of the design of these games. Toy and Wichman soon found that most adventure games suffered from a lack of replayability, in that the game did not change on separate playthroughs.[3][4]

Around this time, ca. 1980, BSD Unix had started to gain a foothold as the operating system for many of the University of California's campuses. One element of the BSD distribution at this point included the curses programming library by Ken Arnold. curses enabled a programmer to place characters at any point on a terminal, effectively allowing for "graphical" interfaces.[3] When Toy saw this library, he and Wichman quickly realized the potential for it. After crafting a few games using curses to learn the library, they came up with the idea of an adventure game in the flavor of Dungeons & Dragons, but, to address their concerns with the static nature of adventure games, wanted to include elements that would change every time the game was played. The two came up with a narrative, that of an adventurer setting out to explore and find treasures in the Dungeons of Doom, specifically the Amulet of Yendor (a renowned wizard in the game whose name is derived from "Rodney" spelled backwards).[3] Wichman came up with the name Rogue, based on the idea that unlike the party-based systems of Dungeons & Dragons, the player's character was going at this alone. They also wanted to make sure the name was short to make it simple to type on command lines.[3][4]

As Toy was more proficient at programming, he led the development of the game in the C language, which generally produced fast, effective code. Wichman learned the language from Toy as they went along while providing significant input on the design of game.[3][4] The first two major aspects of the game developed were the method of displaying the dungeon on-screen to the player, and how to generate the dungeon in a random manner. Limited by choices of what a terminal could display, they stuck to ASCII characters, such as . for empty floor space, + for doors, and | and - for walls of the dungeon. They also used the "at" symbol (@) to represent the player, considering this showed the player "where they're at".[3] For the dungeon, they found initial attempts at purely random generation to be weak, in some cases having a stairway ending up in a room inaccessible to players. They found a solution through procedural generation, where each level would start on the idea of a 3x3 tic-tac-toe grid, with each room of various size occupying one space in this grid, and then creating the hallways to connect the rooms.[3] Once they could have their character move about these randomly created dungeons, they added equipment, magic items, and monsters. With magic items, they wanted the effects of these items to be a mystery on each run-through, and thus would initially present the items to the player only by a descriptor such as color, and only later in the game give the true name of the item once the player experimented or used another means to identify the item.[3] For monsters, they wanted to have more advanced intelligence routines as the player got deeper in the dungeons, but had started running into memory limits on the VAX-11, and simply made the monsters stronger with more health to pose more of a challenge.[3][1]

The two started testing the game with other students at UCSC, finding that despite the limited graphics, players were filling the gaps with their own imagination. Playtester feedback helped them to improve the procedural generation routines to balance the game's challenge.[3][1] One element that fell out from playtesting was the use of permadeath. Toy wanted to move away from the notion of simply learning the right sequence of steps to complete within adventure games, and instead make the player focus on finding the right moves to avoid the character's death at that moment;[3] Wichman later called this idea "consequence persistence".[5] Initially, a Rogue game had to be completed in one sitting, but by demand of playtesters, Toy and Wichman added the ability to save the state of the game, so that players could continue a game across sessions. They soon found players were "save scumming", reloading the game from the save file, an approach counter to their design goals. They changed this so that the save file was erased upon reloading the game, thus making a character's death effectively permanent.[3] They subsequently added a scoreboard feature that let players rank their progress with others, rewarding players with more points for surviving as deep as possible into the dungeons and making the Amulet of Yendor a lucrative goal.[3]

Around 1982, Toy's attention to Rogue and computer games caused him to suffer poor academic performance, and he was kicked out of the school, shortly finding employment at University of California, Berkeley (UCB) in their computer lab. Toy took the Rogue code with him to continue its development.[3] Wichman, still enrolled at UCSC, continued to help develop Rogue for a time, such as adding armor elements, but the logistics of working over the distance made it difficult for him to keep up, and he let Toy fully take over development.[3][4]

Prior to Toy's arrival at UCB, Ken Arnold had gotten to play Rogue, which had been distributed as an executable across many of the UC campuses. Though impressed with the game, he expressed frustration at the inefficient means the game updated the screen via his curses library over a modem line.[6] He had ideas for how to fix it, but at this point Toy and Wichman had opted not to release the code.[6]

When Toy arrived at UCB in 1982, he sought out Arnold to get insight into the nature of how the curses library worked. After the two got to know each other, Toy allowed him access to Rogue's source code.[6] In addition to helping to improve the interface and rendering of the game, Arnold helped to improve the procedural generation aspects of the game.[6]

With its popularity on the UCB servers, Rogue was selected as one of the game titles included in the 1983 distribution of 4.2 BSD, which spread across ARPANET and quickly gained popularity among colleges and facilities with access to this hardware. Among its fans included UNIX's co-developer Ken Thompson working at Bell Labs; Dennis Ritchie had joked at the time that Rogue was "the biggest waste of CPU cycles in history".[6] Rogue's distribution in 4.2 BSD did not include its source code, so after Toy and Arnold separately left UCB, they took the code with them, making it difficult for anyone to build off it. Rogue's source was eventually added under a BSD software license within 4.3 BSD in 1986, putting it into the open source.[7]

The Atari ST version of Rogue using graphical tiles created by A.I. Design and Epyx.

Toy left UCB sometime before 1984 and took a consulting position with Olivetti, an Italian typewriter company that at the time were starting development of their own computer based on the IBM Personal Computer (IBM PC) operating system.[6] There, he met one of Olivetti's computer system administrators, Jon Lane. Lane had previously seen the popularity of Rogue among the United States location he managed and had played the game himself along with Ritchie's observations on Rogue.[1] Upon meeting Toy, Lane proposed the idea of porting Rogue to the IBM PC as a commercial product, which Toy agreed. They founded the company A.I. Design to port and market the game. Though Toy's source code was necessary for the porting, Lane had to redevelop many of the routines for the game's interface. Lane took advantage of the more graphical Code page 437 character set on PC to expand the number of symbols to represent the dungeon, such as using a happy-face ☺ for the player-character. They also took steps to avoid potential copyright issues with TSR, the company that owned Dungeons & Dragons at that time, by changing the names of monsters like kobolds that were unique to that game.[6][1]

Toy and Lane initially funded the publishing, distribution, and promotion of the IBM PC version themselves, and though they continued to gain sales, they were only able to break even as they lacked the power of a larger distributor. Around 1984, Robert Borch, the vice president of publishing at Epyx discovered that Rogue had become popular by several of Epyx's employees and that they suggested that Epyx should help fund ports to other systems. Though Borch felt there was niche appeal to the game, he followed this advice and contracted A.I. Design to port the game to the Macintosh and Amiga upon which Epyx would take over distribution and marketing.[6]

Toy obtained a Macintosh and took the lead in porting the game to that system. Both Toy and Lane recognized that they could implement improved graphics with the Macintosh version, but neither had art skills to make the icons. Toy reached out to Wichman to help with these graphics. Wichman was initially cautious due to the fact that his credit for Rogue in the PC version had been cast as a "contribution" equal to the UCSC playtesters rather than as equal to Toy, Arnold, or Lane,[6] but ultimately agreed to help and joined A.I. Design. Much of the Macintosh version was developed in concert by Toy, Wichman, and Lane in a cabin at the Squaw Valley Ski Resort.[6] Following this, Epyx requested that Wichman lead the development of the Atari ST version, with the company providing Wichman a system to work on. This work occurred alongside Toy's work on the Amiga version. Wichman enlisted help from an Epyx in-house artist, Michael Kosaka, to create the art on the Atari ST version.[6] Epyx would also fund A.I. Design to port the game to other systems including the TRS-80 Color Computer.[4]

Borch recognized the difficulty in marketing Rogue through traditional methods compared to other games on the market at that time, and opted to push the title through software catalogs rather than retail channels. Though it sold well initially, Rogue's sales quickly declined, and it was considered a commercial flop. Besides the competition from more graphically interesting games, Wichman attributed the failure to the fact that the commercial version of Rogue was essentially the same game previously offered for free via BSD and did not pose a new challenge.[6] Epyx eventually went bankrupt in 1989, and A.I. Design disbanded. None of Toy, Wichman, Arnold, or Lane profited greatly from Rogue, though they became renowned in the industry for their participation on the game.[6]

In 1988, the budget software publisher Mastertronic released a commercial port of Rogue for the Amstrad CPC, Commodore 64, Atari 8-bit,[8] and ZX Spectrum computers.[9]

Numerous clones exist for modern operating systems such as Microsoft Windows,[10] Mac OS X,[11] Palm OS,[12] Linux,[13] BSD OSs,[13] and iOS.[14] It is even included in the base distribution of NetBSD and DragonFly BSD.

• 
  4.3 BSD from the University of Wisconsin, circa 1987. Rogue is shown in "/usr/games"
• 
  4.3 BSD displaying the man page for Rogue