[Following three previous articles - for '20 Unusual Control Schemes', '20 Open World Games', and '20 Difficult Games', respectively, Gamasutra's 'Game Design Essentials' series looks at the design lessons from titles in which 'the player must solve mysteries' - from finding secrets to wrestling with algorithmically generated content.]

Not really all that very long ago, a game was released called The Legend of Zelda. For the time it was ground-breaking, but as often happens with games, many of the really interesting things about it were not copied in the inevitable onslaught of games it inspired in the following years.

One of those features was Zelda's huge variety of secret passages. While other games, including later Zeldas, will sometimes throw in a poorly-hidden secret chamber once in a while, the first Zelda did it all the time. Nearly half of its overworld screens, in fact, contain a special area of some type, offering rewards ranging from handfuls of rupees to entire dungeons that must be found to win the game.

The existence of so many things hidden in the game that don't have to be found lends the game a certain quality, one best described as verisimilitude. Verisimilitude is a useful word to use in describing video games. It means the quality of being like life, but the connotations are more profound than petty "realism," which has been redefined in the game review sphere as the quality of the graphics. Properly used, the word means that there seems like there is a world outside the borders of the screen, happening regardless of what the player does. It implies the existence of a fully-fleshed world, one that's more than a mere collection of polygons or tiles that might as well be sealed in Plexiglas. It allows a game to better enable the player to forget that it is, really, just a game.

Of course in games this is usually faked, although with varying degrees of success. Grand Theft Auto and Dead Rising, where one can tell the story to take a running jump and go off and explore on one's own, fake it pretty well. Games that force a strong, linear narrative on the player, not so well.

Not all games require that the player be immersed, and pretend to actually be a participant in the game world. Indeed, it's probable that too many games desire this state, and fail badly when they try. But there are other reasons for wanting to instill mysterious elements into a game. But I should define my terms. By saying "mysterious," I mean games that hide some aspect of their play or workings from the player. These are games in which the unraveling of hidden knowledge is essential to the game. These are games in which the player must solve mysteries -- thus the word "mysterious."

There are two major ways to do this. The most common is to include a lot of pre-made content the player must discover. This is, by far, the most common method, and nearly all games do this in one way or another. First-person shooters only reveal the portion of their territory which is visible to the player's camera. Even simple games like Pac-Man could be termed mysterious, in that the behavior of the monsters is algorithmic, but difficult to figure out during a game. All games are mysterious if the category is defined too broadly, so we will concern ourselves with games that go to unusual lengths, games that take the matter deeper than just the unveiling of new content:

The standard disclaimer: This is not intended to be a "top 20" list, and these games are not presented in a ranked order. Some of the games may seem to only be peripherally associated with the topic. That's because, to truly understand some concept, it's useful to look at both obvious and non-obvious cases. Most of these games are older, but this is as intended. Older is not the same as worse! The purpose of the article is to show how other games have done it over the years, and hopefully to inspire you in working on your own projects. You could even play them to see the principle in action.

Games That Emphasize Finding Secrets

1. Super Mario Bros. 3

The white blocks

Developed by Nintendo

Designed by Shigeru Miyamoto and Takashi Tezuka

Reason for inclusion:

In addition to the usual secrets that come just from being a Super Mario game, the third game in the series has some amazingly obscure things to find. If it weren't for Nintendo of America's own player's guides from the time we would probably have never found the Magic Note Blocks, treasure ships or white mushroom houses, let alone the special properties of white background blocks and how they could be utilized to find a Warp Whistle.

The game:

The Super Mario Bros. games make searching for secret power-ups and passages into a play mechanic itself, and SMB3 had probably the most deviously hidden ones of them all. Looking for invisible blocks has always been part of the Super Mario Bros. games, and there's an unavoidable aspect of trial and error to doing it. But for obscure secret areas, Super Mario Bros. 3 takes the proverbial cake.

It's well-known to fans of the game that some of the rectangular, decorative blocks, which can be stood upon as platforms, have special properties. They come in several colors, but only the white ones are in any way unusual. If the player stands on a white block for several seconds holding the control pad down, he will "fall behind" the block. The NES then uses a sprite priority feature to make Mario appear to be behind the background. The player will stay back there maybe eight seconds, invisible if he's standing by any of the background hills or such. There is no clue in the game that this will happen. It is pure mischievousness on the part of the designer.

This is pretty random by itself. While behind the scenery, the player passes by enemies, making him effectively invulnerable -- but since in all the levels of Super Mario Bros. 3 there only a handful of white blocks to be found, this function rarely enters into strategy. And yet, this secret function is required to obtain one of the game's three Warp Whistles.

Because doing it allows the player to warp ahead levels, its function is analogous to running at the top of the screen in the original Super Mario Bros. (ably discussed by Jeremy Penner in Gamer's Quarter #7.) But its significance is greater. By taking a nearly universally-ignored aspect of platform gaming, the color of a type of generic background element, and attaching significance to it, the designers imbued the whole game world with far more mystique than it would otherwise have. There is only one important white block in the game, found in the very first world, and no other random background element carries special significance; all the others are designed to be clearly read by the player, from pipes to note blocks to donut lifts.

Yet the white block trick implies that there is a lot more to the world than what is visible. It demonstrates mysterious properties that only an expert could know. It was knowledge of prime importance on early '90's school yards. It carries with it the whiff of magic. Thus, this one trick lends weight to the whole game. The fact that everyone knew the trick within months of the game's release doesn't prove it to be ineffective -- rather, it proves it worked. As Dr. Strangelove remarked, the whole point is lost if you keep it a secret.

Design lesson:

In a way, it's the ultimate dumb trick. There is really no way to know ducking for several seconds on a certain spot in 1-3 will lead to one of the game's most valuable treasures. But the hint is there, and the trick is known. And although it's only one trick in a game with many, it's obscure enough that it makes the player wonder about the entire rest of the game.

2. Bubble Bobble

Item generation mechanism, vast assortment of items, Bubble Alphabet, game modes and codes, and ending conditions

Published & developed by Taito

Designed by Fukio Mitsuji

Reason for inclusion:

How could this one not be included? Possibly the most mysterious game ever produced for arcades, with the only likely exception being its sequel, Rainbow Islands. Lots of stuff about this game was only fully understood when someone finally disassembled the code a few years back.

The game:

Bubble Bobble's multiple layers of secret knowledge make it a game for wizards. A huge array of arcane trivia must be learned to exploit Bubble Bobble's many systems. Some of it is only now becoming commonly known.

First thing to know is that the arcade game has secret codes. That was almost unheard of at the time -- yet Bubble Bobble not only has them, it will tell the player about them in the right circumstances, and using one is required to see the game's real ending. The circumstances in question, however, are nearly obscene: a game must be played to level 20, 30 or 50 without dying in order to see one, and death is rather common in Bubble Bobble after the first few levels. Once on the proper level, the bonus item normally generated in the level will instead be a door to a secret room, containing huge score bonuses and the code printed on the background... but the code, itself, is written in code.

At first it just looks like background decoration of runic glyphs, but it's really a secret message from the developers. The secret is spilled on the ending screen, where the player is informed "But it was not a true end!" and the specialness of the glyphs is revealed. That is, if he won in two-player mode. Winning with only a single player in the game sends him back to a random level. D'oh!

One of the secret codes makes the doors that lead to the bonus levels appear regardless of lives lost, which is nice to know once the player's already figured out the code, I guess. Another code, when entered from the title screen in the machine's attract mode, puts the game into "Super" mode, which mixes up the enemies in the levels and is somewhat more difficult. But it is only in Super mode that the game can truly be won, by working through all 100 levels again and finishing up in two-player mode. (And without dying, incidentally, but fortunately most people who win actually play as one player through most of the game, only putting player two in after the last boss has been beaten, and having just joined he won't have ever died.)

This is far from the only mysterious thing about the game. Probably the most mysterious is the formula that determines when special items appear. Most players believe it to be random, but actually they're all directly determined by the player's actions. The game maintains counts of a large array of trivia, like bubbles popped, times jumped, water bubbles burst, and times wrapped around the screen. When a counter exceeds a threshold value, a flag is set scheduling the item to be generated in an upcoming level.

The counts and flags are not reset when a game is finished, which is why new players often get treated to advanced items in the very first levels. But it also means if someone knows that, say, popping water bubbles makes level-skipping umbrellas appear more often, he can take advantage of that fact to skip ahead more often. Some of the tricks that make items appear cannot be done on every level, so it also tends to cause certain items to "clump," generating more often on particular boards.

It's nearly a textbook example of chaotic behavior in game design. The player's many varied actions lead, through unknown processes, to results that you'd assume are random, but seem to have some mysterious consistency to them. It is an interesting way to bring players into a game: to pin success on objects that appear almost on command, but without telling exactly what that command is.

Design lesson:

Pseudo-random numbers are used by many games, but Bubble Bobble, while apparently random, in fact contains few of them. Once their principles are understood they can be gamed, but they are not explained to the player. The result is that different play styles produce consistent variation in the game's response to them, which can provoke responses from players so complex that they border on superstition.

Links:

The Bubble Bobble mechanics discovered through source code examination are explicated at the Bubble Bobble Info Pages.

3. Rainbow Islands

Bubble Bobble with even more secrets

Developed by Taito

Designed by Fukio Mitsuji

Reason for inclusion:

The item generation scheme for this game isn't quite as complex, but it's still possible to make the items you need with some planning. Then there's the means players must undergo to obtain jewels, which must be collected in order to get the true ending. And then there's the super goals obtainable by getting the jewels in color order, and special permanent powerups for doing so! And then there's the secret levels! And then there's the secret ending for not dying! And so forth! Exclamation point!

The game:

While definitely a successor to the play of the first game, Rainbow Islands discards much of Bubble Bobble's item generation criteria. Instead, just killing a lot of enemies without dying will eventually produce the more useful power-ups. But it has even more secret areas and endings to find. It is probably the king of the Bubble Bobble play style because of it.

The basic mechanic of the game is to create rainbows, this game's analogue for the bubbles in the first game, and trap enemies beneath them. Then the player can jump on the rainbow, sending it crashing down and killing all the enemies beneath it -- sending them flying around in an arc before coming to rest on a platform. When multiple enemies are killed this way at once, they produce colored gemstones.

Levels can be finished three ways. The main way is to just go through and beat the boss, but if even one level is cleared that way, the player cannot get the best ending. It's really a trap, for although the enemies are the cause of lives lost, and spending a long time in a level also kills the player, the player can't just make progress normally and really win the game. He has to worry about other things along the way.

To really beat a level, one gem of each color must be obtained before the boss. Gems are generated by killing multiple enemies at once, but their colors seem random at first. Perhaps predictably, there is a trick to getting needed colors. When an enemy flies around after dying, the X-position of the place it comes to rest determines the color of gem that will appear there. The left edge of the screen is red, the next seventh of the screen over is orange, the next yellow, and so on through the spectrum, ending with violet at the right edge.

Each level has four stages, and the gems carry over from stage to stage. So to "super-clear" the whole level, the players must have gotten one gem from each vertical stripe of the screen in the stages leading up to the boss. Doing this causes the boss to reward him with a giant gem upon its death, and finishing the normal last level with all seven giant gems makes the real last levels available. (Cleverly, they're all based on other Taito properties like Arkanoid and Darius). Those levels have mirrors instead of gems, but they're obtained the same way. Getting all of them allows the player to win for really, really real... if he can beat the last boss, that is.

But there's more. If the players collected the gems in color order, from red to violet, then when the boss is reached, a special door will appear. Entering it skips the boss, rewards the player with his hard-earned diamond, and grants him a permanent power-up that lasts the rest of the game. This is all on top of the usual find-the-hidden-object gameplay shenanigans.

Design lesson:

Riddles and mysteries abound in this game. Secrets cannot make a game all by themselves, and Rainbow Islands' core play mechanic, while good, isn't quite as strong as Bubble Bobble's. Yet while both are long games even when played normally, no one who completed either game in the arcade and seen the Bad End could mistake that there was more left to do.

Links:

Perhaps the most complete resource on the game is the Rainbow Islands Info Pages.

4. Marble Madness (home computer versions)

The Water Maze

Developed by various developers working for Electronic Arts (original version by Atari Games)

Originally designed by Mark Cerny

Reason for inclusion:

It's not too well understood even now, but the computer versions of Marble Madness contain a secret level called the Water Maze. It's obscenely difficult and obscure to get through, but it's not obvious how to get to it, either.

The game:

It's not well-known (despite being featured in magazine ads), but the home computer editions of Marble Madness published by Electronic Arts contain a secret level of sorts. In the first level, if the player jumps off the pointed sides of the trench leading to the goal and lands on the floor on the left side of the screen, there's a secret passage to be found. It can't be located just by bumping around; he must wait on a specific spot until the timer reads 13 seconds remaining. Done correctly, an elevator will appear leading down, and soon after he'll be in a new area with 99 seconds on the clock. This area is the Water Maze, which appears only in EA's ports of the game.

The new area is a bit more fidgety than the real game. Instead of a series of action movement challenges, it is more a series of puzzles where the player must figure out what to do to move on. Many of the puzzles require two marbles to pull off, so while a single player will be let into the level, he won't really have a chance of completing it. Many of the puzzles involve elevators or rivers, objects which don't appear in the main game. The game rules are also different there, for if the clock runs out the level resets, but if the player dies the game ends immediately.

It greatly changes what it means to play Marble Madness, and it's maddeningly difficult besides, so it's fitting that it's relegated to an obscure easter egg in a dark corner of the game.

Design lesson:

The Water Maze plays differently than the main game, is fairly small, and finishing doesn't cause anything special to happen, so it's good that it's relegated to being a weird little secret. But that moment when a player happens upon it for the first time is really something.

5. Ogre Battle

Derandomized combat, obfuscation through complexity, character acquisition, and the Chaos Frame ending spoiling mechanism

Published and developed by Quest

Designed by Hiroshi Minagawa

Reason for inclusion:

This is a game that nearly everyone has to resort to a FAQ to play at some point. While the player is progressing through the game, the game is also keeping score in an obscure way. After finishing the last level, the score directly indicates which ending the player receives. But it's hard to keep the score up, and it's never explained exactly how it changes...

The game:

Ogre Battle is a game that's interesting for many reasons. It's a Japanese RPG that still takes a fairly simulationist approach to the play, and that's reason enough for examination all by itself. It's a decent challenge and has a strong strategic (as opposed to tactical) component, moves along in real time, and even has a strong economic component. But when a fight begins the player is largely left out of it. His characters decide for themselves what to do.

About those characters... They each have a stat called Alignment that maps generally to the D&D usage of the term: high alignment numbers, going up to 100, mean Lawful, and low alignment, down to 0, means Chaotic. The player will typically have many parties on the field at once, and each one has its own alignment value. Defeating darker enemies like vampires raises alignment, and lighter enemies like clerics lowers it. There are other things that modify alignment too, but once it gets close to its limits on either end of the scale it becomes more difficult to change it back. Character advancement often depends on alignment.

Now, there are other things to note here. When characters that are low in alignment take over one of the game's town bases, the game says the town is captured. If the same town is taken over by a high-alignment unit, the message says the town is liberated. These are very important messages, for the game also keeps track of another stat, this one describing the player. It is Reputation, and while character Alignment is an easy-to-read number, Reputation is never reported directly except through a very small bar. Here is a primary rule governing Reputation: capturing towns lowers it, while liberating them raises it!

A number of other things also affect Reputation, and tellingly, they do it in a way that makes the game more difficult if the player is trying to keep it high. Beating enemies with characters of much greater level than them (more than two levels difference) lowers reputation, while doing so with characters of less difference, or even of lower level, raises it. As game time passes on a given map, the player earns income from the towns he's taken over. Money carries over from map to map, but income resets. This might inspire some players to dally on a map and build funds, but hanging along too long lowers reputation.

So why is reputation important? Ah, that makes the game even more mysterious. In addition to characters joining up or not depending on player reputation (generally undocumented opportunities), reputation directly affects the ending the player receives after winning the game. A really low reputation will give the player a bad ending; in one, the player's character becomes a new tyrant who gets assassinated by one of the characters in his group.

Of course, none of this is explained to the player except in the manual, and there but vaguely.

Design lesson:

The player's actions throughout the game are put into an algorithmic blender, and at the end that value (among other things) selects the final outcome of the story. The result is that Ogre Battle is actually scored, just as much as Pac-Man or Rogue, but the player isn't told that score with any degree of fidelity. The N64 sequel takes this even further: the player doesn't even get that bar telling what his reputation score is.

Links:

GameFAQs has a nice mechanics FAQ on Ogre Battle that provides a more fleshed-out description of the Chaos Frame system.

6. Magic Sword

Many items, Bubble Bobble-like item functions, and surprisingly deep strategy

Published & developed by Capcom

Designed by Y. Ohnishi, T. Sadamoto and Y. Okamoto.

Reason for inclusion:

Sorta like Bubble Bobble with barbarians, a hack-and-slash platformer with a lot more personality and depth than it has any right having.

The game:

Magic Sword is an inexplicably awesome little arcade game. It's a fantasy hack-and-slash that doesn't take itself too seriously, and feels more like a relative of stuff like Pang and Rainbow Islands than contemporary Capcom fantasy games King of Dragons and Knights of the Round.

Those two games are actually brawlers, more like Final Fight than D&D. Magic Sword manages to feel more genuinely adventurous because, mostly, of the wide variety of secret things to find. The game's "helper" character system plays a role here. Helpers are assistant characters who follow the player and generally copy his movements. They work a bit like options in Gradius, but the player can only have one at a time, and they can take damage. As the player makes it through the tower's 50 levels, he encounters various doors with different kinds of locks on them. Opening a door uses up a key but makes available the contents -- usually a helper and a item. There's actually not a key shortage per se, and the early levels give the player an essentially infinite supply of each type (but finishing a level with extra keys is worth more points, and more bonus health).

The keys add an element of resource management to the search for treasure. Most of the doors are plainly visible, but opening all of them costs the player a key if he has one to spare. Inside there are a variety of items and helpers, but the player doesn't know what unless he's memorized them on previous playthroughs. (Some even contain enemies.) The helpers have different strengths that make them better suited for some areas than others: some have powerful melee attacks, some fast distance attacks, some magic that's strong against undead enemies, and one, the Thief, locates hidden objects for the player. A couple helpers are themselves secret, appearing only in a few places or from special means, and are fairly high-caliber discoverables for this kind of game.

The thing that makes the secrets in Magic Sword particularly impressive is the amount of coding that went into them. The helpers each have their own subroutines andeffects upon the game for finding them. The big tradeoff for including truly secret secrets in games is: why waste manpower including content that will only be seen by a small percentage of players?

One reason is that it randomizes the game. If players can't always find a useful object because he doesn't know what makes it show up, or doesn't even know it exists, it increases the diversity of experience, and ultimately makes the game more replayable. Different players end up facing different challenges. Beating a boss with a Lizardman is easier than with a Thief, but the Thief will show hidden chests that may provide their own advantages, which could be just a useful in the boss fight.

Another reason is that it makes the game feel more mysterious. Most games these days are laid out on a linear track without opportunity for meaningful deviation. Deviation could come from providing alternate territory to explore, like in Mario games, or it could come from varying the resources available to the player.

Design lesson:

Although the enemy and item positions are generally hard-coded, there's enough randomness in Magic Sword, enough changes to the game situation made possible by whatever helper or item he has, that in practice it's really quite replayable. This way, the implications of the pre-made levels vary according to the player's helper and carried item, where the player can only have one of either of those things at a time. It's a style of game design that's relatively uncommon anymore, yet can greatly extend playability if done well.

7. Athena

Block-breaking, many items and functions and lots of secrets.

Developed by SNK

Reason for inclusion:

The game plays like a successor to Super Mario Bros. because of the huge number of breakable blocks that hide treasures inside them. Shattering them is key to success, but there are lots of weapons to choose from, and some of the best for killing enemies are bad for breaking blocks.

The game:

Lots of people are down on Athena on the strength of its horrible NES port, which is loaded with bugs, spotty controls and a brutal damage model that can kill a full-health player in a moment. The arcade game is rather more polished, although still very difficult.

Athena came out not too long after Super Mario Bros., and much of it feels like it was directly inspired by that game. Where Mario's blocks only occasionally contained important items, Athena's worlds are composed mostly of breakable stone, hiding a much larger percentage of stuff to find. At the start of a life the player has no means to break them; killing enemies provides the initial tools needed to do that. There are several different kinds of weapons available, and one of their prime distinctions is that each allows the player to destroy blocks in a different way: up close, within a limited range, from a distance, in horizontal lines, directly above or by destroying a large swath. Some weapons are better for breaking blocks than killing monsters.

Once the player can get blocks open, it's revealed that there are dozens of possible things to find, including different levels of armor, helmets, shields, weapons, and miscellaneous stuff, and they're all over the place. Unusually, among the good stuff, many blocks contain bad items. Super Mario Bros.: The Lost Levels occasionally threw in a poison mushroom or booby-trapped Starman, but Athena blocks constantly provide armor and weapon downgraders, or poison, or time-downs, or inventory destroyers, or even obnoxious floating head enemies. Castlevania's dagger candles only wish they were this annoying.

Yet the presence of those items is what adds texture to the game. A danger with many kinds of video games is the "so what?" factor. What is it that distinguishes this level from all the others in the game? Mario does it with enemies, gaps and walls to overcome in different arrangements, hidden passages scattered around, and the availability of power-ups. The first Super Mario Bros. doesn't have that many different game elements, but the way they're arranged approaches art. Athena does it by limiting the blocks that can be broken depending on the power-up obtained: if you just have a yellow sword, which takes two swings to destroy a block, and the item you want is buried beneath five layers while an endless stream of horsemen attack from behind, it's probably not worth it to go for. There are enough different classes for these items that sometimes a very weak item is more useful than a very strong one, simply because it's better at breaking blocks.

The game also, by the way, has as convoluted a win condition as either Solomon's Key or Mighty Bomb Jack, except instead of just downgrading the ending, the last boss is actually invincible if conditions are not met. Which, you know, it would have been nice of the game to tell the player before actually fighting the last boss. (Hey, I didn't say the game did everything well.)

Design lesson:

The core of Athena lies in the way the player can break blocks, but he can't always smash the ones he wants. Sometimes a bad item helps to obtain a very good one, even while an average one would not.

8. Mighty Bomb Jack

Mapping the pyramid and special ending conditions.

Developed by Tecmo

Reason for inclusion:

Like Bubble Bobble and Solomon's Key, Mighty Bomb Jack is absolutely loaded with secret items and areas, and forces a player to discover lots of well-hidden things in order to see its best ending.

The game:

Although maligned by many reviewers, it's a fairly clever little game. It's a little spacey in its implementation, for sometimes enemies spawn right over the player or inside walls, but at least it's less buggy than NES Athena. The sequel to the semi-obscure arcade game Bomb Jack, the primary method of play is the same as that game. The player controls a guy with an extremely high jump as nearly his only skill. He can control his arc by pressing the jump button again in the middle of a leap. When it's pressed, it zeros out his vertical velocity, ending jumps early and slowing descent. It does nothing to change horizontal velocity, so by pressing the button quickly players can glide across long distances, provided there's space to do so and an enemy doesn't generate in the air right in his way. All told, what this means is that the player can effectively "glide" by rapidly pressing the button, and pull off some amazing escapes by sliding horizontally between approaching enemies.

While it's not exactly a "Metroidvania," one of the neater things about the game is that, although it's composed as a sequence of branching levels, it still maps out coherently. The manual provides an illustration of an Egyptian pyramid with the outlines of the first areas mapped out on it. As successive levels are explored, their locations can also be placed on the map. If the player does this, some interesting things about the game become evident.

In a couple of places the corridors extend outside the pyramid. When that happens, the graphics change to sky and clouds. If the game is mapped out, it becomes clear that there are some places where there is a significant void, a place where it looks like there should be something, and this is an important clue to hunting down the secret rooms. The principle at work here is that the areas aren't laid out arbitrarily but according to a deeper structure, and this aids exploration by giving observant players a subtle hint as to passage locations.

Two of these secret areas have special import. While the player can work through the game normally and get a plain ending, to get the best ending the player has to find special items buried in the depths. Finding them is particularly difficult because often the player must jump on unmarked breakable blocks on the level that disappear when this is done, and use that to drill down and find hidden treasure chests, some of which contain the items that make the secret door appear. Some of these chests are invisible at first, and are hidden in the middle of large empty rooms. When the player is being relentlessly chased by killer mummies and parrots that appear out of thin air, jumping on every piddling spot on the screen can be something of a hassle, but it definitely makes it challenging to find them.

Design lesson:

Like Solomon's Key (which was also developed by Tecmo), one can win the game without getting the best ending. Unlike many multiple-ending games these days, the player doesn't feel gypped by a less-than-optimal ending. The day is saved regardless of the ending obtained. Hunting up those crystal balls out of their maddening secret rooms just makes a happy ending happier.

Some developers put in bad endings as a way to punish the player for playing in a way they don't like (Ogre Battle for instance), or for not figuring out some obscure trick (adventure-style Castlevanias). In Castlevania II: Simon's Quest, Simon Belmont's fate gets worse the faster the game is completed.

Now, understand -- I am averse to posing many iron-clad "Don't do it!" rules for game designers to slavishly follow. There is a good counter-example to nearly every piece of such advice I've ever heard. But I would suggest that, if you're going to put real multiple endings in your game (and not just trick ends like in the last case of the second Phoenix Wright), that you consider allowing the player to feel good about his accomplishment regardless... especially if, to get anything else, the player must jump through the kinds of flaming hoops that Mighty Bomb Jack presents.

Links:

Most of the resources on this game are in Japanese. The best site I know of on the game can be found at Mighty Bomb Jack Walkthrough Page (Google Translation). Be sure to check the map of the pyramid. The full details of the game's endings and how to get them can be found in a GameFAQs article that summarizes the old Nintendo Game Counselor's handbook entry on the game
.

Games That Rely on Algorithmically-Generated Content

9. Starflight

Fractal-like world generation.

Developed by Binary Systems

Reason for inclusion:

Starflight fit on two floppy disks, yet still managed to contain thousands of star systems, most with several planets each with their own distinctive map and minerals to harvest. It was all done by making canny uses of hashes and fractal map generation schemes. Another game that uses techniques similar to this is Elite, for its map.

The game:

Starflight provides many lessons for developers concerning not just generating huge game worlds, but in making those worlds interesting. It uses what I call the "mining" approach to algorithmic content, which is used by roguelikes too: the idea that the randomized areas the player explores are exploited for random wealth, and can be depleted, forcing the player to expand his horizons in order to obtain more.

Here, each individual world tends to be interesting mostly for the minerals it contains. The basic advancement mechanic is to find planets, land on their surface with a scout ship, scan to find stuff to mine, then travel back to base to sell it for better equipment. Some planets have adverse weather effects, some have dangerous wildlife (which can be collected, stored and sold) and some have dangerous conditions on the surface, but there are enough planets with normal atmospheres and no life that players typically don't have to bother with messing around with those if they don't want to.

Advancement loops exist in many different kinds of games, and often don't have a lot of thought put into them. For example, the classic advancement loop used in arcade games is that the player gets better solely through practice. He has no game-sponsored additional powers on level 50 than he had on level 1. Another common loop, which originated in RPGs but has since spread far beyond them, players get better through gaining experience points from defeating monsters and collecting gold from them in order to buy better equipment.

Starflight's loop is financial, for obtaining money is the means of bettering the player's ship. And money is gained primarily through exploring all those thousands of algorithmically-generated planets.

Design lesson:

Randomized games (or games that use obscure algorithms to generate their content) frequently use a mining metaphor for their advancement loops. The world is generated automatically by the computer, without much rhyme or reason given to placement of dangers or rewards, and to improve his state the player goes into it to take advantage of what he can. As rewards in the safer areas are depleted, he's forced to travel to more dangerous areas to continue to advance.

Of course, even static exploration games with pre-made areas are like this. The difference is that algorithmically generated game can have much more area, and thus exploration can be a greater portion of the difficulty barrier towards earning a reward than combat or puzzle-solving. Other games that use this general play style, not coincidentally made by some of the same guys, are Starflight II and Star Control II. Star Control II is particularly interesting, for it has a strict time limit that means the player must explore efficiently, and it provides clever clues as to which planets have better resources.

Links:

There is a fan-made sequel that's been in the works for some time, Starflight III: Mysteries of the Universe, that has the support of the game's creators.

Star Control II has a fan-made recreation, titled The Ur-Quan Masters.

10. NetHack

Item discovery, random generation, random items.

Published, developed and designed by the NetHack Dev Team

Reason for inclusion:

NetHack is a game that seems eternally capable of surprise. I myself have played it for well over a decade, and even now I still occasionally happen upon some aspect I wasn't previously aware of.

The game:

It takes the massive difficulty of Rogue and, instead of lowering it, moves it. Moves it away from the Random Bad Luck column and into the Stuff To Learn column. Many complain about its difficulty, but because knowledge is more reliable than chance, the most experienced hackers eventually find the game too easy. This is partly because it's been the subject of some of the most determined game FAQ writers in the world, and partly because -- and this must be said -- its source code is open for everyone to see.

NetHack contains random dungeons, so there is still a mining-like aspect to the game, where new dungeon levels must be sought out, not just because it gets the player closer to his goal, but because once a level has been completely explored there is no treasure left to find. Its predecessor Rogue took this to extremes, in that random items were the only stuff to be found, and some of those items were food rations which were required for survival. The pressure is taken off of food a bit here (rations tend to be common, dead monsters can be eaten, and the player can pray for help when weak from hunger), but it remains that finding treasure on the dungeon floor is very important.

In addition, and unintuitively, monster generation isn't strictly according to dungeon level. In fact, after the first couple of levels are gained and the player's hit points are out of the danger zone, finding a great item will nearly always be the better advantage for a player than gaining another experience level. Stuff like highly-enchanted armor, a pair of speed boots, or a cloak of magic resistance provide more benefit than the extra hit points, to-hit chances and minor other advantages levels provide. Further, and unintuitively, the maximum difficulty of generated monsters is the average between dungeon level and player level, so the act of gaining levels actually makes the monsters a little tougher in response.

All random dungeon games come down to mining to some extent. In games like Angband, that forget dungeon levels once the player leaves them, the practice of sticking around the same levels and regenerating them over and over to build up loot is so prevalent that there's a name for it: scumming. The name comes from the idea that, by generating lots of uninteresting levels, eventually one with a nice item will rise to the top. Not only is scumming a successful strategy in Angband, the game requires it. Requires it, and has even come to be designed around it: successful players usually hang out on the levels on which are generated mushrooms award permanent bonuses to stats until they max out their ability scores.

The most mysterious aspect of NetHack is the item identification system, which obscures the identities of most magic items at the start. A player might find speed boots lying around, but if he doesn't know what they are he might not put them on. Or if he does put on unknown boots, it could turn out that they're actually of a bad type, like fumble boots, and cursed so they cannot be easily removed. This is similar to the system used in Rogue, but there is a major difference between the two. In Rogue, the function of most items can be figured out only through reading a scroll of identify (which itself starts out as a random item) or by use, which is often wasteful, or even dangerous. NetHack provides far more ways of figuring out what objects do, and a player who has learned them all finds the game much easier as a consequence.

Design lesson:

Two lessons here. The first is that random games tend to be helped by item identification systems, but they have to be designed around them. If the objects aren't selected randomly then the player can remember what their types are from prior plays. It does make the game more difficult, however. If you decide to include random unknown objects, here are three techniques, corresponding to Rogue, NetHack and Diablo --

Rogue: Unknown items are useable, and difficult to figure out without great risk or resource use. No bad item will directly end a game if used, but many of them can lead to indirect death from monster attacks.

NetHack: Easier to figure out items if the player has discovered the special techniques, but still falls back on Rogue behavior if he doesn't.

Diablo: Unknown items can be used, but their special features are not active until identified. This tends to just be a way to make the player return to town from time to time.

The second lesson is a bit more obscure. To make an identification game work, there must be bad items as well as good ones, to provide a risk for using without identifying. Every class of random item must have at least one bad item in it. But once the player knows what all the bad items are, the incentive to not try things on randomly is much diminished. Every bad item discovered out of all the types in the class makes the rest much easier to figure out. This means that in NetHack, which is a very long game, the item identification system tends to only be important in the early phases, and partly because of this, the game becomes much easier once most of the stuff has been discovered.

Links:

The NetHack Wiki is a great resource in learning about the game. So is the Usenet group rec.games.roguelike.NetHack.

11. ToeJam & Earl

Random objects, identification risk.

Developed by Johnson-Voorsanger Productions (ToeJam & Earl Productions)

Designed by Greg Johnson

Reason for inclusion:

Roguelike in its item identification game and random levels, it is a strange experience to play it now simply because players aren't used to figuring things out for themselves as much anymore.

The game:

Rogue and NetHack were created by, and for, college students. ToeJam & Earl on the other hand was made for typical console gamers, then mostly kids. It's got randomly generated levels, objects that must be identified (including some very bad items), no continues, and rather high difficulty. Now just guess how well it did in the marketplace.

Actually... eventually... it did pretty well! Bill Kunkel (writing as "The Game Doctor") called it a favorite of his around that time. It did so well, but was so different than everything else in stores before or since, that when it came time to make the sequel Sega second-guessed the creators, telling them to make a game completely different from the original. The result was by no means a bad game, but now most everyone agrees is inferior to the first, despite having much better graphics and an actual storyline.

ToeJam & Earl's mysterious aspects come from the level layouts and the presents laying around. Levels are not just randomly arranged (which is a fairly shallow way of randomizing a game) but have environmental obstacles and aids that sometimes require expending resources to get through. The presents work like the scrolls in Rogue: using one causes it to take effect immediately.

There's some additional nuance in ToeJam & Earl. It can be played as a two-player game, and if the players are far from each other the game goes split-screen, letting them explore where they want without being tethered to each other. However, if the players are both on the same screen when a present is opened, it affects them both, whether it's a good present or bad. Opening one type of present results in the immediate loss of a life, and its effect will hit both players if near each other, just like the others, as will the Extra Life present.

One of the game's best design choices is that it ameliorates one of the flaws of the roguelike design with the inclusion of its Randomizer present. Opening that one rescrambles all the other presents in the game! It is a tremendous setback for the players, not the least because it randomizes itself in the process. However, it does fix the big problem with relying on an identification game in the design, for it's not true that, once an item is known, it is known forever and need never be identified again. Until the Randomizer is identified, the player must be more careful the more presents he IDs, not less.

Design lesson:

Identification games are interesting only so long as there are still things to be learned. Rogue solves this problem by having a relatively small dungeon, so it's very unlikely that every item will appear even in a victorious game. ToeJam & Earl presents the chance that items could become re-randomized. But it's not fair to just reset known items "just because," the scrambling must always be the result of some mistake made by the player. In the later ToeJam & Earl III for Xbox, for example, the attacks of one of the enemies can scramble items.

Links: An interview with designer Greg Johnson.

12. Dungeon Hack

Random dungeons & item identification, with a backing in D&D official rules

Developed by DreamForge Intertainment

Reason for inclusion:

It's another random dungeon game, this one pretty obscure, but interesting because it uses the D&D license -- which imposes interesting limitations on the game.

The game:

Dungeon Hack is a random dungeon game, and it has randomized items, but it is not traditionally regarded as roguelike. The game is set in a generated Wizardry-style maze, although it helpfully includes an automap. Also, its dungeons are a lot less free-form than those in Rogue. There are locked doors that can only be opened with matching keys, making the experience a lot more linear. (The maze generation algorithm, thankfully, ensures each level is solvable.)

Dungeon Hack marked a return of Dungeons & Dragons computer games to the quick-play, random determination tables in the back of the 1st edition AD&D Dungeon Master's Guide. Since then, the games have gradually trended toward meticulously planned layouts with specific treasures intended to provide specific advantages (or disadvantages) for a party against the foes in that region. The Dungeons & Dragons gold box games follow this pattern.

Dungeon Hack, despite its differences and attempts to look like an Eye of the Beholder-style game, does seem to take some ideas from NetHack. There are various devices along the walls in each maze whose function must be discovered through play, just like NetHack's dungeon features. While items must be discovered through play and their appearances match up with function, most of the items match up exactly with the loot in the 2nd Edition Advanced Dungeons & Dragons guides. The roguelike ideal is not to include just any old stuff but to provide items that provide suitable risks for identifying things through use, without being too risky.In this area, Dungeon Hack falls short.

Yet there is one aspect of the game that was reused later in a surprising place. Dungeon Hack contains a number of artifact items. They don't act like "true" D&D artifacts, which tend to be things more like The One Ring than a Sword +3 vs. Reptiles, but are nevertheless unique items with special powers. Each of these items, however, is part of a set, which if matched up with all the other items and worn at the same time provide considerable bonuses. One can't help but think the creators of Diablo II were taking notes.

Also different from most random dungeon games, there are no "wandering" monsters added after initial generation. Each level has its starting population and that's it. One could think of each level has containing only a set number of experience points, in fact. And it uses D&D-style resting mechanics, where the player must clear out an area of monsters in order to regain hit points and spells. And food is consumed while resting, so the limited food system limits not exploration time but healing and spells.

Design lesson:

The level generator is pretty slick in how it creates locked door puzzles that are always solvable, and it produces special dungeon zones, which are an underused feature in random dungeon games. Its use of food is also innovative, limiting not time but rests.

13. Diablo series

Random dungeons, enemy layouts and treasure, multiplayer exploration

Developed by Blizzard Entertainment

Senior designers: Erich Schaefer and David Brevik

Reason for inclusion:

The Diablo series is the most popular commercial random dungeon game in the world, and still has fans even ten years later. While the play is less mysterious than NetHack, there are interesting things about it.

The game:

Diablo's random dungeons and look, and non-respawning monsters, produce another mining game. Unlike the other random dungeon games discussed here, the check on exploring the dungeon at the player's leisure is entirely the monsters: there's not much in the way of environmental obstacles to get in the way.

Frankly, the Diablo games diminish many of the cooler things about random dungeon exploration games. Unknown items don't exist, as overall types share similar appearances but random functions. One is given to wonder why identification is even in the game: unknown items can be used, but only get their basic effects. The result is simply to force the player to return to base more often to get stuff ID'd. There is no chance of figuring out an item through use or experimentation, and there are no unknown potion or scroll types -- not in the roguelike style.

In their defense, it's possible the developers did this because to implement true roguelike object identification requires permadeath to avoid reload abuse, and that's a change many mainstream players won't enjoy. And in a heavily multiplayer game, the penalty for using a bad magic item isn't so bad since there are other players to help out. And it's good that they do away with Rogue's food system. It's not so bad in Rogue, but food has no place in these games' structure.

Diablo's dungeons are relatively simple compared to those of other random games, and that's a statement, I realize, that demands explanation. In architecture, Diablo's dungeons are actually very complex -- far more so than Rogue, NetHack, Dungeon Hack, or many other games. And scattered through them are monster rooms and shrines, in addition to quests and unique monsters. But despite these things, there is a kind of sameness to them. Levels are defined, in game terms, as monsters and loot. To a great degree, the layout of walls in a random dungeon is not that interesting. All of these games have automaps, after all, so there's no danger of getting lost. It's not all that exciting, in Diablo, if a corridor is straight, has a bend in it, or a branch, or is actually a room. It's just space.

It's all just space, and its shape doesn't matter for much. It's true that players do need to take it into account to avoid getting surrounded, and keep a path open through which to flee if necessary. But by abandoning many of the daggers in Rogue's design, the import of the dungeon itself is lessened. In true roguelikes the player must explore efficiently, to gain the most loot while using the least food and taking the fewest risks. Diablo is by no means the only random dungeon game to abandon this aspect of Rogue, but it does leave me to wonder, sometimes, why one would bother putting random maps into a game if their structure has so little real consequence to the player.

Design lesson:

The incentive for playing the Diablo games primarily comes from finding random loot. The games' dungeons are a bit less interesting than loot-hunting because of the lack of consequences for the act of exploration. There are no traps on the floor, there's no food, and there's no randomly-appearing monsters. For a single-player game this is less interesting, but for multiplayer it works better. Perhaps this is why Diablo's system is basically the template upon which most MMORPGs use.

14. Barcode Battler

Hash-based character generation

Developed by Epoch.

Reason for inclusion:

A Pokemon-style game that generates monster stats in a unique way: it uses a hash function applied to data contained in UPC-style barcodes.

The game:

It's really an awesome idea.

There is this game called Barcode Battler, and it once caused quite a stir in Japan. The way it worked is, the players each swiped ordinary UPC-type codes, culled from any product in the player's house, into the machine's attached barcode scanner. This data, ordinarily used for retail purposes, would be read by the machine and, after putting it through an obfuscation process (like applying a hash function, or using it as a pseudo-random number generator), used to define the essential statistics for a "monster." The monster could then be pitted against another player's monster, generated using the same process but probably using a different code.

The result of this is that, since there is no real randomness in the monster generation mechanism, scanning in the same code always produces the same monster. The result is that, just as the data that defines a character in a game is, in a sense, identical with that character, the use of a physical barcode to generate statistics for an "animal pal" kind of character gives the code and its visual representation a kind of shared identity.

It also sets up a real-world, Magic: the Gathering-style economy around the codes. The story goes that Barcode Battler enthusiasts, upon finding a code that produced a good monster, would end up causing sell-outs of the code's product as many players rushed out to obtain their own copy. Yet unlike in CCGs, the makers of the Barcode Battler system obtained no financial compensation for their system, a powerful incentive to run such games.

And unlike the card games, the rarity comes not from limiting the availability of the signifying objects, for there is no good way to ensure a popular product won't produce the best possible monster, but simply through making the powerful attributes less likely to come up. It may well be that 20-ounce bottles of Coke could produce the SuperMegaDragon, but the designers can make this unlikely by using a much smaller portion of the species-space for the dragon.

Design lesson:

The obfuscation system is an important part of all this. If it were easy to tell which barcodes produced good monsters then players wouldn't actually have to scan them in, or even cook up their own monsters using homemade barcodes. The fact that the link between codes and monsters is unknown to most players makes the process mysterious, and thus, it seems more real.

It was copied by the U.S. game Scannerz, which uses a similar kind of system. Another series that uses this system is the PlayStation versions of Monster Rancher, which instead of barcodes uses the data off of ordinary CDs put into the console's disc tray.

Links:

Wikipedia's article on the game is informative.

An enthusiast site for the UK version of the device.

Games That Hide Important Play Information

15. Street Fighter II

Undocumented movesets, relying on players to discover how to play on their own

Developed by Capcom

Planned by Akira Nishitani

Reason for inclusion:

Street Fighter II is chosen as emblematic of a range of fighting games, and being the first breakaway hit of the type most games end up doing it like it did. (Technically the original Street Fighter had command-based special moves too, but no one much likes that game anymore.)

The game:

The first step in becoming good at any fighting game is to learn the moves. Of course most people just read a FAQ nowadays, but how do those FAQs come to be written? Before, most arcade games went out of their way to demonstrate how to control them, printing the information right on the control panel. Imagine what must have happened in order to have performed that first Haudoken? Especially the first Shoryuken, which isn't exactly an intuitive input? The breaking from the idea of direct-control to command inputs is a significant one, and for many fighting games the moves aren't printed out for the player beforehand.

What purpose does hiding the special moves have? Well, it adds to initial approachability at the cost of making it harder to master. And that's not necessarily a disadvantage to a fighting game, as mastery is supposed to be difficult. That helps to make matches more interesting since, early in a title's life at least, the players must work without full understanding of a game's options.

And really, mastery of a fighting game has to be difficult, because I don't think it's a very well-kept secret that there's not much to a fighting game. There are no maps to explore, there are no power-ups to find, a left-right line doesn't give much room for maneuverability, and secrets tend to be limited to playing a certain way (don't lose a match for an extra ending) or in entering controller codes ("Sub-Zero wins... Fatality"). The fighting game play model is purposely simplified to focus on the aspects of the game the designers see as important, but being simple, other means must be added in order to make it explorable. In this case, what's being explored is the capabilities of the characters, and how they compare with each other.

So how does news of the moves get out initially? Well these days there's strategy guides and such, as ostensibly complete catalogs of all the interesting stuff in a game they would naturally have complete movesets. Home versions of fighting games will often include a selection of moves in the manual, and some (like the Soulcalibur games) have a training mode that purposely spoils all the moves. And these days most games end up on sites like GameFAQs that pool the information gleaned from all these sources.

Back in the early days these options were not available, but some games would print a few moves on the cabinet art. Since the computer player has complete knowledge of all the possible moves, people could at least find out what kinds of moves were available from watching the opponents at work. Fighting games often lend themselves to discoverability by having the move animation match up, in an interpretational manner, with the motions of the joystick and the buttons, so the animation of a move is often a clue as to how to perform it. In Street Fighter II at least some degree of discoverability is important, since with a digital joystick and six buttons, there are a huge variety of controller inputs to search in order to find a new move.

Design lesson:

Fighting games like Street Fighter II simplify the game world in order to focus on aspects of video game play that, before they came along, were important but not the main focus. It also, by hiding access to the more powerful abilities, adds a schoolyard mystique to the game, and increases the effort players must put in to master it. Yet, it's interesting to speculate as to what would happen to a game less charismatic than SFII were it to hide access to options in this way...

Links:

David Sirlin holds forth with fascinating notes on fighting game design on his blog.

16. Pokemon

Hidden stats.

Developed by Game Freak.

Designed by numerous people -- Satoshi Tajiri seems to be one of the main minds behind it. The recent DS games list Shigeru Ohmori as Game Design Leader.

Reason for inclusion:

A little-known aspect of the Pokemon games among casual players that comes to rule the strategy of experts is a set of hidden variables that carry each monster's battle history forward in unexpected ways, and greatly influences stat growth upon gaining levels.

The game:

Pokemon has gotten so complex across the five incarnations of the series, with trades between them, new features added each game, and the dozens of ways of interacting with them, that it's really in a class by itself. It requires far more out its grade school enthusiasts than a casual observer may suspect. It's popular to ridicule this franchise in general, in many cases justifiably, but the games are beyond reproach. And huge amount of Pokemon's appeal comes from the mystery that suffuses the games. The lore that surrounds the games is the current preteen version of knowing the special moves in Street Fighter II and Mortal Kombat back in the day.

I could easily fill this whole article with stuff about the Pokemon series alone. They are games that require poring over megabyte-sized FAQs to completely understand. The only other game I can think of that's of similar complexity is NetHack. Between breeding, evolution paths, time and day differences, berries, accessories, trading, daily events, contests, one-time-only encounters, Nintendo-only monsters and many other things besides, it defies understanding without serious effort. It almost seems as if it's designed to build character. Yet of all these things, there is still one aspect that's generally little-known. It's the dividing line between the Pokemon amateur and the super-amateur: do you know what Effort Values, a.k.a. EV, are?

Effort Values are the fan name for them. The official name is not known. They are an aspect of Pokemon that none of the companies responsible for the game have ever officially acknowledged. All of the game functions that relate to them speak in imprecise terms. It is only through the work of people hacking saves and using other extra-game means that we can put any hard numbers to EV.

EV is a set of hidden stats tied to all pokemon since those captured in the black-and-white Game Boy originals. In addition to the base stats of HP, Attack, Defense, Special Attack, Special Defense and Speed, each also has a hidden EV value. EV goes up when pokemon fight, like experience points, but this varies according to species fought. The level of the monster doesn't matter -- only its species. Fighting Bidoof earns a player's troops HP EV, while pitting them against Golduck gets them Special Attack EV. Fighting a lot of low-level monsters will help a bit, but there are limits to the amount of EV that can be built for each stat, and possessed in total, so to fully optimize a given pokemon the player must shepherd it along a managed program of fights.

EV points are not reported in the battle victory message, and do not appear on a pokemon's stats display, but they build up invisibly as a pokemon fights. When a monster gains a level, in addition to the base stat growth and a random factor, it also cashes in some of its EV points in exchange for additional points of its matching stat. In this way, a pokemon's battle history comes to influence its advancement in ways other than just gaining levels.

There exist ways to grant pokemon "free" experience. Giving it a Rare Candy grants a free experience level. But those experience points are just empty calories, for while the pokemon gains its level bonus to stats, unless it has been fighting anyway to build up EV its stat growth will be less than it could be. Over 100 levels of advancement these bonuses add up, with the result being that an EV-trained pokemon will always be superior to one that was built haphazardly.

In true pokemon fashion, there exist other ways to build EV, although the workings of the system are never explained in-game. Giving a pokemon vitamins will boost a specific EV value a bit. In Diamond and Pearl, equipping a pokemon with the Macho Brace will cause it to earn double EV, a significant bonus. There is also a very rare effect in the game that happens randomly, which causes a pokemon to get sick with a "Pokerus." Getting sick is actually a very good thing, because it doubles EV earned, its bonus stacks with that of the Macho Brace, and it can be spread among the player's collection through casual contact, Corrupted Blood-style.

Design lesson:

EV may be the most mysterious aspect of a game... that may have too many mysterious elements. One interesting thing about it is that there's an aspect of punishment to it. It seems as if it exists to make players who use many Rare Candies pay for it. But it also serves to cause a pokemon's battle history to carry forward in ways other than its experience count, helping it seem like slightly more than a collection of pixels with numbers attached.

Links:

GameFAQs EV FAQ for Pokemon Diamond.

17. Gauntlet II

Secret rooms

Developed by Atari Games

Designed by Ed Logg

Reason for inclusion:

Gauntlet II is overall a better game than the original, with more diverse play, more opportunity for strategy and generally more stuff. Its coolest feature might be the secret rooms, little-understood special levels that are sometimes entered after completing a level, and even let the player enter a T-shirt contest if successful. But how are the rooms found? No one quite knows....

The game:

The original Gauntlet certainly had enough terrain to explore, but once the player is familiar with all the maps he can settle down to the task of devising his strategy in each of them. But the sequel has a really strange, and awesome, feature above the original...

Every so often, when a level is completed, before the next level there will be a message: "To find a secret room:" followed by some instruction. Typical instructions include: "Don't get shot," Don't be fooled" and "Be pushy." While it's thought that the instructions aren't actually unrelated to the requirement necessary to enter a room, it doesn't seem that the message issued is necessarily good at that moment.

What is known is that, if the right trick is done on the right level, the player who performed the trick will show up, alone, in a special level, portentously titled a SECRET ROOM. There are multiple possible rooms that could turn up, and they're timed just like treasure rooms with a very strict limit. They may contain multiple permanent ability potions, and are very difficult to complete in time. Not reaching the exit in time causes no negative consequences for the player, but if the player does make it, finishing in time presents a special code and an address at Atari Games the player can write to and enter a contest and a free T-shirt.

The T-shirt offer expired long ago, but the room's allure persists. The ability potions can turn any character into a powerhouse, and since the player has no competition for them, even in a multiplayer game, all he has to do is get to them. But more than that, the secret rooms are special. One could play many games of Gauntlet II and never see the first secret room.

These days there are few remaining mysteries when it comes to what games do: we know there's no way to stop Aeris from dying, we know how to get to Bubble Bobble's "Happy End," and we know where all the warp zones are. But we still don't know how to get into a Secret Room. I suppose someone could disassemble the source code and find out. They probably will someday. Here's hoping it's awhile before it happens.

Design lesson:

Even today, no one (other, I suppose, than the people who worked on the game themselves) quite knows how to enter one of Gauntlet II's secret rooms. It's an interesting experiment in adding a super-goal to an otherwise straightforward arcade game. Atari did a number of games with those: Toobin' and Vindicators also have T-shirt contest entry screens.

18. King's Bounty and Heroes of Might & Magic

Piecing together the map.

Developed by New World Computing/3DO

Designed by Jon Van Caneghem

Reason for inclusion:

The game that inspired the Heroes of Might & Magic series is a relatively little-known game from New World Computing called King's Bounty that, until HoM&M, had little to do with its more famous siblings. Both are games that involve exploration, economics, some chance, magic and strategic, turn-based combat... but there's also a cool over-puzzle to both the original and some of the later games that makes discovery, and a good memory, perhaps the most important player skills of all.

The game:

Might & Magic games sometimes pose interesting challenges to the player upon conclusion. Might & Magic II ends with a timed, randomized cryptogram, and after finishing World of Xeen's quest, the player can go back in to explore a dungeon that's also a crossword puzzle.

King's Bounty is another game that New World Computing created around the time of Might & Magic I and II. Originally it had little to do with the M&M games, but it was used as a basis for the spin-off Heroes series when its creators decided to diversify the franchise. And the coolest thing about King's Bounty, which also made it into some versions of HoM&M, is the over-puzzle. I'm most familiar with how the puzzle works in King's Bounty, so that's the game I'm going to describe.

The object of the game is to find a scepter that has been stolen from the king of the realm, and by restoring it to said ruler, end the rebellion that resulted from its loss. The evil dragon boss what swiped it has hidden it somewhere in the realm, and it is the player's task to find it. He's taken the map showing its location, divided it into a grid of five by five squares, and distributed the pieces to his 24 lieutenants, keeping the center piece for himself.

The realm is composed of four huge continents, each with a good number of lieutenants to fight. Each has taken up residence in one of the land's many, many castles with an army themed around that boss's personality. The player's job is to recruit mercenary troops from the many towns and lairs in the land, take over castles to provide periodic income with which to pay them, then go out and capture the villains. Some defeated villains yield artifacts which aid the player's progress, but most importantly, each villain removed from the game gives up a piece of the map.

When the big boss hid the scepter, he didn't follow standard Evil Overlord protocol, stashing it away in the deepest dungeon or strongest fortress he could find. Instead, he stowed it on a random square on one of the four continents. It's well-hidden, so the player doesn't see it while walking around; it could be stashed in the least accessible corner of the last continent, or it's possible that he'll walk over its location a hundred times during the game.

The only way the player can find it is to search the square it's hidden. But not only are there far too many spaces to just try every spot, there is an overall time limit to the game. If the player runs out of days the quest ends in failure, and searching without the center piece of the map (the one held by big boss Arech Dragonbreath himself, by far the toughest villain) causes a search to take 10 days instead of 1.

If the player searches on the right spot he'll find the scepter and win the game immediately. Not only is this an instant win, it's the only way to win. If the player collects every map piece but can't match the picture to a spot on one of the game's continents he still loses. But if the player can figure out where it's hidden after getting the map pieces from just the weakest villains, he can win the whole ballgame without much bloodshed. The combat, the economic game, the exploration, recruiting armies and buying spells, and so on -- all of this exists only as a means.

The real game to King's Bounty is a picture-matching puzzle. Everything else just makes it easier. But the right spot is practically impossible to find without finding at least one map piece, and to usefully narrow it down requires several. The harder villains also happen to be the ones with map pieces in the center. The game, thus, is a game about gathering information. Or put another way, it's about solving a mystery. How about that?

Design lesson:

King's Bounty engages insome clever misdirection, with all the game's combat mechanics built in service of a kind of shaggy dog design. In practice players will usually have to defeat most of the villains anyway, but a player who doesn't like to fight could conceivably win a different way. It rewards players who are good explorers by letting them use that skill for something other than navigation. "Role playing game" is nearly always effectively a synonym for "fantasy combat simulation," so it's nice to see a game where the mechanics of exploration play a larger role.

19. Baten Kaitos

The deck, maintaining it, and the luck of the draw

Developed by Monolith Soft

Battle design by Hiroya Hatsushiba and Yoshiharu Kuwabara

Reason for inclusion:

Since Magic: The Gathering hit it big, card-based battle systems haven't been uncommon either in real-world or RPG settings, but few did it as well as Baten Kaitos and its sequel, which use a deck, customizable by the player, to essentially replace its combat mode's random number generator.

The game:

The Baten Kaitos games are strange in many ways. Instead of plain human beings, all the characters sport wings of some type, which is a bit funky. Polygonal backgrounds have been left out in favor of the PlayStation Final Fantasy technique of using pre-rendered artwork for all areas. And then there's the combat.

Combat is bizarre. Every fight the player gets into is played out as a card game. When a character's turn comes up, the player attacks by playing cards from his hand. She can play one card to do some damage, or she can play a sequence depending on the numbers in the corners of the card. Doing so does considerably more damage, but relies on the player having the right cards both in both his deck and her hand. Playing a run requires cards from a wide range of values, and higher numbers are introduced slowly through the game.

I really can't emphasize the card fighting system enough. Characters have little in the way of statistics because the cards are the whole game. These cards, called "magnus", are collected by the player and organized into decks, which can be rearranged any time outside of combat, in a manner that CCG players might appreciate. But combat itself doesn't really play like any real CCG. The game is more a combination of rummy, with its searching for runs, and snap, because battle still takes place in real time. A turn ends when a player can't play any more cards or a very strict clock runs out on his turn.

Perhaps the biggest effect of the card battle system is that it's possible to get royally screwed over if you draw a poor hand at the start of a fight. Played cards get replenished automatically from the deck, but if the player can't play any cards he only gets to swap out one before the next turn. If the player has unusually bad draw luck even basic encounters can become difficult. Hand size extensions earned through the game make this less likely, but it's hard to ever rule it out completely.

Design lesson:

These are very odd games. It's difficult to imagine it could ever have hit it big in the current marketplace, yet it forges ahead with its oddness as if it were the most natural thing in the world. And it is difficult to dislike for having the sheer balls to replace combat with cards. It's enough to make me wonder, in fact, if the game could have been made on a more popular system than the Gamecube. As an exclusive for the tiny purple box, it was probably a foregone conclusion that the game wouldn't be profitable, so they just damned financial success and went and made what they wanted to make. There's nothing like certain failure to loosen inhibitions!

20. ADOM

Still mysterious after many years.

Developed and designed by: Thomas Biskup

Reason for inclusion:

In addition to its roguelike roots and capable copying of many of the best features of NetHack, in addition to introducing a number of similar features of its own, and the whole "ultra victory" system that's in place, there exists an item in Thomas Biskup's popular roguelike that, to this day, no one other than Biskup understands the purpose of, the "weird tome." How has it remained a mystery for so long? Well...

The game:

Ah, it's another roguelike. And probably the one that's taken the most inspiration, of them all, from NetHack. But hear me out on this one, because it's not here for containing random treasure and an identification game.

While NetHack is the product of dozens of people, hundreds if you count patch contributors, ADOM is mostly the work of one guy. And yet, it shows an amazing degree of ingenuity. Considering it's the work of a single designer, it's genius. It's right around the time people discover that the herbs on the dungeon floor grow according to the rules of Conway's Life that they get the notion that Thomas Biskup must be some kind of god-being. If one forgives some style issues (reading much of the NPCs dialogue makes me cringe), it makes for an extremely mysterious game.

There are very few computer games that still have the power to mystify us, and for the most part developers have given up the fight. When many of us started playing video games, we were still in school. Remember what it was like to hear about Super Mario Bros. warp zones for the first time? These days, I'm sorry to say, publishers have realized they can get themselves a tidy extra income by purposely dispelling any aura of mystery that might surround their games. Nintendo themselves took the lead in selling the secrets of NES games from the publication of the Official Nintendo Player's Guide, and now game strategy guides are expected to be sold alongside nearly any major game.

With ADOM there is no official cheatbook, and unlike NetHack there's no available source code to search, so despite intense study by a legion of players and the work of extraordinarily dedicated FAQ authors, there remain some things about the game that are mysterious. Not that a lot of progress hasn't been made in decoding the game's deeper mysteries: for a long time the details of the "ultra" endings, special victory possibilities that let the player ascend to godhood instead of just saving the world, were just rumor. People looking at the executable discovered that Thomas Biskup has even included unused, red herring strings in the code to throw people off the track.

One of the remaining mysteries concerns an item called the "weird tome." The stated purpose of the book is simply a package in a fetch quest -- deliver it to another character in exchange for valuable items -- but the player is able to read it. Apparently consisting of pages of magical music, attempting to read it without extremely high (far in excess of normal) stats and perfect scores in the relevant skills will just confuse the player for a long period of time. Although general consensus is currently that the book's contents is just a huge red herring, a couple of other characters in the game say interesting things if the book is given to them.

Design lesson:

To keep a game secret hidden in the face of overwhelming player interest is profoundly difficult. With consoles, developers have the advantage of relatively secure hardware, but on computers it's much more challenging without resorting to unusual trickery. A game's popularity works against it here; the more fascinated people are by the mysteries in a game, the harder they'll work to figure it out.

Links:

ADOM's remaining mysteries are summarized in a section of the community-written Guidebook.

Google Groups threads in which members of the ADOM player community discuss efforts to discover the purpose of the tome:

Disassembling the source code (2004)

Recent hacking of save files & building characters with maximum stats

Some older executable investigation from 2003