Like most people in this industry I think I know what makes a good game. Of course also like most people I've never released a game that is actually considered good. Something always conspires to get in the way. As Mark Cerny once said, "Everything has to go right to have a hit, but only one thing has to go wrong.". I do strongly believe that if all the rules below are followed a good game will come out if it.

Actually that's not true.  I'm proud of a few games.  Gunship, M.C. Kids, parts of Gex and Crash Team Racing.

Of course a point I like to make is that there is a big difference between a good game and a hit. There are several hit titles that are not good games. A simple example would be any almost any Simpsons' game for the NES or SNES. Most of them did very well in the market because of the popularity of the Simpsons but most of them were very bad games.

Good game design is of course different to different people and different depending on the type of game you are creating. I'm not a big sports game fan and so I really don't have any great ideas on how to make a great sports game. Also, some sports people like simulations and others like arcade style play. Those two types of people or going to find different types of games to have the 'better' design.

My favorite games are platform games, action adventure games and platform shooters. My basic philosophy in game design is to look at my favorite games and compare them to bad games of the same type. The differences should point out why one game is good and another not.

Platform Games

For me, I define a platform game to be one were the player runs around and jumps on platforms but also for the most part does not shoot. This distinguishes between games like Mario and Mega Man or Sonic and Turrican. It also points out that games like Aladdin and Earthworm Jim are more like Mega Man or Turrican than Sonic or Mario. The cute graphics may make you think they are similar to Mario but if you were to replace the graphics of all these games with rectangles you'd quickly see which ones are more like a Platform Shooter than a Platform Game.

Here are some of my design ideas on Platform games. Many of them may also apply to other types of games.

• All levels should be separated into action, rest, action, rest areas. One easy way to do this is the have 'blocking' graphics. Graphics that allow you to separate one situation from another. Like the pipes in Mario. The pipes in Mario not only separate one area from another in terms of play but also separate groups of enemies so that the player can't get too many enemies on the screen at once which would push the game engine to it's limits. (i.e., slow down)
  
  Separating areas like this helps you look at an area and gauge how hard it is in a smaller context than the entire level. Can the player make it thought this one area to the next rest spot. Longer areas are generally harder to survive so you can often get a progression through a level by making the next area a little longer than the previous area.
  
  
• The game should have JUMP plus ONE an ONLY ONE action. More than one is too complicated... A Second action can be added but only if it is seldom used. e.g. Mario Spin Attack/Donkey Kong Switch Players.
  
  Example:We noticed this in Gex. There are three actions in Gex, JUMP, Tongue Lash and Tail Whip. It used to be that Tongue Lash was a mandatory action. You had to use it. We changed it near the end of the product to make it optional. You can play the game and never use Tongue lash because most players found it too complicated. In the heat of play they couldn't remember which button to press and neither could anybody on the team.
  
  
• 1/3 to 1/2 of the game should be secret. This makes the game last longer and gives it replayablilty. It also gives the players lots to discuss with their friends. All of this adds up to great word of mouth and keeping the game in the player's mind for weeks of play instead of days and therefore much much better enjoyment, memories, and sales. (This idea from Miyamoto)
  
  Super Mario World and Wario are good examples of this with Mario having the Star World and the Special World and Wario having the Ice World all being totally secret.
  
  
• Bosses should be simple. The action of a boss should be based around the basic dodge then shoot, dodge then shoot. The dodging can get increasingly difficult in later stages of the game but should be relatively simple with earlier bosses. Variations like Dodge/Dodge/Shoot and Dodge/Collect/Dodge/Shoot are okay but choose the variation before designing the boss.
  
  
• Design half your characters around what actions you need them to take. In otherworld, if your design NEEDS a character that goes up and down then come up with a character that looks like it would go up and down. If you need a character that is indestructible then make a character that looks indestructible (like the spiked rollers and the octopus in DKC).
  
  Design the other half by look and then give them a behavior appropriate for their look. Example: You are doing a horror level, you decide you must have a mummy. Design the mummy to do a mummy like thing like just walk left/right and growl (This idea from Miyamoto)
  
  That may sound stupid but consider the all to common alternative. Enemies that either all do the same thing or enemies that do something they don't look like they should be doing.
  
  
• Music must 'drive' the player in an platform game. Like Techo/Dance it must make the player feel like moving in the same way that parade music makes you feel like marching and dance music makes you feel like dancing. It's okay to make exceptions (like water levels in DKC) but the initial music must make you move.
  
  
• Enemies must graphically distinguish whether or not the player can kill theml. In otherwords don't just make a bunch of enemies and arbitrarily decide that some are killable and some aren't. An enemy that you can't kill should be obvious by how it looks like maybe it's covered in armor or looks like it's made out of stone. Something visual that lets the player know that it's probably indestructible.  Another good thing to do is have a good and obvious visual effect and/or sound when you try to kill an indestructible enemy. Example: The metal 'tink' in Zelda makes it obvious that you can't kill a particular creature with a particular weapon.
  
  
• Come up with playability guidelines for level building. Example: A Bonus puzzle level should take a good player no more then 6 tries to perfection. No regular level should take more than 4 lives to complete. Then test your levels with new players. Your goal should be to make a game that's fun not frustrating. If most people can't make it through your level without throwing the joypad your level is too frustrating and needs to be fixed.
  
  
• Design the game to present levels in a non-linear themed order. In otherwords Sonic and Gex are boring because in Sonic you have to play three levels that look almost the same before you get to see anything new. In Gex you have to play 5 levels before you get to see anything new.
  
  A Good example: DKC. Jungle 1, Jungle 2, Cave 1, Jungle 3, Water 1, Forest 1 Boss 1. Forest 2, Forest 3, Tree 1, Jungle 4, Water 2, Boss 2.
  
  
• Set specific limits during design so that proliferation of objects and situations those objects must deal with is kept much smaller. Choose a specific numbers of enemies, items and other things that have to be drawn and programmed. That way you can have few well designed, well used, well programmed objects instead of lots of rarely used, poorly designed objects.
  
  Example: there are over 450 separately programmed objects and enemies in GEX vs. about 50 in DKC (which game is better? I think DKC is better.)
  
  
• Level design needs to come before art. Specific design. This is to avoid the problem of the designer incorrectly using the art to make structures the art was never meant to make. Example, all graveyard levels in Gex make no sense. Graveyard dirt flies up in the middle of nowhere and for no reason. If the artists know what the designers plan to do then the artists can draw art what will work for the situation. Note: This also means the designer MUST commit to design ideas. Art cannot be redone!
  
  This also assumes that your team is split into designers and artists.
  
  
• Make a list of levels and for each level list what elements are introduced in that level.
  
  What new graphics:
  What new enemies
  What new objects
  What new game play elements
  What new sounds
  
  Quick example from Super Mario World.
  
  Level 1 has:
  Turtles, Yoshi, Help Box, Mushroom
  
  Level 2 has:
  Big Shot, Ducking areas, cannon pipes, invisible mushrooms
  
  Level 3 has:
  Swinging platforms, vertical levels, Secret blocks, falling off bottom.
  
  Level 7 has:
  First time player sees cape. Flying bugs, Triangle Wedge, 1up game, Secret level, Key
  
  Level 8 has:
  Water, Fish.
  
  The point here is to make each level NEW and DIFFERENT. If you use all if your items in the first level what is left for the player to look forward too. Also it gives you a chance to make a list of things you need for the entire game and see if it's a reasonable size.
  
  
• Each level must have a strong THEME. Pick 2 or 3 things and build a level out of them. Good examples of this are in DKC in almost all the levels. Each level chooses two or three things and then makes a creative level out of them. Start with simple setups and then progress them with more and more complicated things. This serves several purposes. One is that it defines the the level so that you can get it done. Another is that it distinguishes the level from others. This is the hanging from the rope dodging vultures level. This level is the barrel level. This level is where you have to jump over bees. (vs the more common, This is the level with a bunch of stuff in it.  This next level also has a bunch of stuff in it just in a different layout.  Yuck!)  Another reason is that it limits the level to a few things. If you don't pick a theme with a few things then usually you'll just keep adding more and more stuff and never actually finish the level because you won't have defined what it means to be finished with it.
  
  
• Have a world map and make sure the player can take at least 2 routes through it at all times. This brings up a whole host of issues. In order to have an open map you must have enough levels. 8 or 12 levels is probably not enough. Most games that have only 8 to 12 levels get a certain amount of game play by making their levels hard. They also usually only allow a certain number of continues and then you have to start the game from the beginning. They do this and tell themselves that they're making the game last longer and therefore more fun but really they're just making the game frustrating. Sometimes extremely frustrating. Having a world map with multiple paths through it means that if the player is having trouble getting through a level they have other levels they can try. Having enough levels to support multiple paths through a world map means that you won't have to resort to making your levels frustrating because you'll have enough levels that the game won't seem too short. Also, you should design the game so that after completing the game you can go back and play the levels on the paths you didn't take giving your game more replay value and making the player happy that the game is not completely over when they beat the final boss.
  
  
• Some enemies should be indestructible. Don't make all enemies take damage from your basic attack and don't have some just have more hitpoints than others. Some good examples, in Mario you can't kill a piranha plant (unless you are powered up) You can't kill the spiked turtles unless you powered up. This forces the player to deal with the enemy in a different way than other enemies. Gex didn't follow this rule and therefore is boring as almost every enemy is dealt with exactly the same way and therefore might as well be the same enemy.
  
  
• Power ups must be useful and different. Any power up you have must be required somewhere in the game. Mario has very good examples. The ones in the last paragraph. (i.e. you need the powerups to kill some enemies. Also you can't get to several secrets without being able to fly. Gex would be an example of not following this rule. Every enemy can be killed with a tailwhip. You never need any of the powerups ever. Therefore there is no compelling reason to get them.
  
  
• Have more than one ending. Having more than one ending to the game gives the player a reason the play again. Especially if one ending is better than another and or vastly different. A bad example of this is Super Metroid. It has three different endings but they are so minorly different that there's no point in trying to see them. Gex, and Mario 64 are great examples of this in that they have very different endings and word of mouth and the press has really pushed the need to keep playing until you've seen the better harder to achieve endings.
  
  
• Have multiple goals. DKC, Yoshi's Island, M.C. Kids all have multiple goals. In DKC for example you can just get to the end of the game or you can try to find 100% of the secrets. Yoshi's island you can just get to the end of the game or you can try to finish each level with 100 flower points allowing you to access other levels. The problem here though is that you have to make the goals interesting and not frustrating or tedious. Getting 100 flower points in Yoshi's Island is not fun for anybody it's just tedious and lame. If almost no one will take the time to see your extra work what's the point. Games should be fun not frustrating. Super Mario world is one of the best examples here. You can go though the game and finish. Then you can try to see all of the game and although some of the secret levels are hard the basic premise of completing all the levels is not frustrating. Don't make it frustrating to get to see all of the game. Tricky maybe but not frustrating.
  
  
• Never force the player to take a hit. Some designers that will make setups that require the player take hits. Sometimes multiple hits. They usually justify this by saying that the player will just have to search the level for extra hitpoints so that when they get to that setup they can survive it. Instead of searching for the extra hitpoints I think most players will just stop playing your game.
  
  
• Consider a two or three hitpoint system. In Mario, Sonic and DKC you have 1 or 2 hitpoints. That means that the player basically switches from 'I'm safe' to 'I'm in danger' mode all the time. For example on Sonic if you have coins you are 'safe' because you can take a hit from anything and survive. Then when you take that hit and you have 0 coins you are in panic mode. You must run and find another coin quickly so that you can get back to safe mode. The same is true in Mario, you are big Super Mario and you are safe or you are just Mario (small) and you are in panic mode running to get another powerup to be Super Mario again.
  
  Bad example is Gex because you can have up to 8 hitpoints. That means that most of the time you don't care if you take a hit. It's not important until you are down to one hitpoint.
  
  It also encourages your designers to make setups that require hits because if you've got 8 hitpoints what's one hitpoint to you. If instead you decide early in your design that you will have a 2 hitpoint system then you can make your designers aware that they must be designing things that can be played without taking a hit and that they must balance the safe/panic modes throughout a level so that if the player is in panic mode (1 hitpoint) he doesn't have to go to far to get another powerup/coin and back to safe mode.
  

Action Adventure Games

Action adventure games are often grouped together with RPGs (Role Playing Games). I wish this didn't happen because the markets for the two types of games are as different as sports simulation vs. sports arcade game. An Action Adventure is a game where all the actions of the main character happen in an arcade action style. They require quick reflexes and speed. An RPG is a game that doesn't require any quick reflexes. Instead the player chooses from a menu which action each character should make and then watches as that character performs the action either successfully or un-successfully.

I should say that I find Zelda 3 (A Link to the Past:SNES) to be the ultimate example of a good Action Adventure and so most of my examples will be from Zelda

• Transportation: In an action adventure game the area you can explore typically gets larger and larger. If you are going to require the player to get around that larger area then at several points in the game add some new transportation that makes traveling less tedious. Zelda 3 is a good example of this. You start with just walking. Then you get the dash boots which allows you to move somewhat faster. Then they add the water whirlpools that will take you from one to another and finally they add the flute and bird. Each one is slightly more convenient than the last so that has you have to go from one end of the world to another it gets less and less tedious.
  
  
• Toyness (pickup, throw, environmental damage): Have lots of things for the player to do that 'don't really matter' and make them easy to do. In Zelda 3 examples are: You can cut bushes, pickup bushes, cut the grass, dig almost anywhere, there are lots of rocks to pickup and pots in most buildings to pickup. Almost none of these things are required but they are fun and make the world seem more alive.
  
  Note: This requires some for thought because if you can pick up one thing in the game then you need to make sure there are not lots of things that look like they can be picked up but can't.  For example lets say the player can pick up trashcans.  Lets say you didn't give this any thought and your artists went and drew rocks and bushes and other ornaments that are about the same size.  Now you make the game and you put trashcans and rocks and bushes all over the place.  Now you focus test the game and you find that every new player is frustrated because he can't tell which things he can pick up and which things he can't.  The solution is to not draw anything that same size as the trashcan unless you are going to allow the player to pick it up.
  
  
• Anticipation: Make sure to have lots of anticipation. There are two things I mean by this. One is for example if you have a river that the player can't cross but he can see the other side he'll wonder what is on the other side and wonder what he needs to do to get to the other side. He'll anticipate crossing. He'll want to know what's over there and keep playing until he can get there. The other kind is where you put some kind of barrier in front of things that the player can later remove. For example in Zelda 3 the first one is the cracks in the wall. You pass a few going through the castle dungeon but you ignore them because you can't do anything about them. Later in the game you get bombs and find out that you can blow a hole through the wall anywhere you see a crack and then you try to remember all the places you saw cracks before so you can go back and check them out. This also happens with light rocks, dark rocks, the mirror and pegs.
  
  
• Different Monster Behaviors. Try to have all monsters behave differently. Define differently by how the player has to deal with them. For example a monster that walks straight for the player and another that does that same thing but has two hitpoints instead of one might as well be the same monster. It's boring. Go through Zelda 3 and notice how different each monster is. Some have shields and can only be hit from one side. Some try to seal your shield. Some shoot right at you, some jump away from you.
  
  
• Most Monsters don't pay attention to the player: If you go through Zelda 3 you will notice that almost 2/3s of all the enemies don't pay any attention to the player. Instead they just follow some pattern of movement and basically get in the way. Those that do follow the player almost always move slower than the player or pause every few seconds to give the player a chance to get away.
  
  
• Always make sure the player knows what to do next. Try to give enough clues that the player always knows what to do next. That doesn't mean that you have to tell the player about everything there is to do but at least tell him about the required things. Imagine Zelda if it didn't mark the map for you. You'd just have to wander around and stumble upon the entrances which would be really tedious. In Zelda 3 also there are several people you can talk to that will give you clues as to where you'll find other things.
  
  
• Give the player more than one thing to do: Try to make sure there are always several things the player can do. For example the player can go to level 3 or find the flippers or find the ice wand. All of those options are available so that if one of them is too hard he can try something else that may help him with the other parts or maybe just playing for a while longer one one of the other goals will get him better at controlling the character for when him comes back to his original goal.
  
  
• Always make 'secrets' findable: Don't put in any secrets in that require the player to read about it in a magazine. For example something requiring the player to stand in some unmarked spot and press a, b, c and d is not cool (unless somewhere in the game some character gives you those instructions). Great examples of secrets in Zelda are the several places where there is a strange yet obvious mark on the ground. In Zelda 3 if you see a circle of plants or stones or a cross in the dirt you can usually bet that something secret will happen there. Many of them require to you stand in the center and then use the mirror to teleport to the light world were you will end up somewhere you can't get any other way. A great one that incorporates both this concept and anticipation is the medallion you get in the desert on the light world. From the light world you can see the structure that contains the medallion up on the cliff but you can't find the way up. Later you find that if you stand in the right MARKED place in the dark world and teleport to the light world you can get there.
  
  
• Design by events not by levels and monsters and bosses: Don't design the game by deciding that you are going to have 10 levels with ten bosses and then try to populate the levels with interesting stuff.. Instead design the game by the events that the player can/will do. For example the player will first try to get in the castle. Then the player will get the sword and shield from his uncle. Then he'll rescue the princess. Then he'll escape through the dungeon. Then he'll get told to visit some old lady in the town, then he'll be told to find an old man in the town. Then he'll be told to find another man on the other side of the world. That man will tell him to go to level 1 and then come back. That man will then mark 3 more places to go. Also while in the town will be told to visit the fairly at the head of the river.
  
  The point is that each of those events has to be programmed/scripted/created.  If you do it from the level/boss paradigm you'll find that you've only done about 1/8th of the game because that game is not about levels to the player it's about the events that happen while trying to get from level to level..
  
  
• Have logical puzzles: Any game where flipping a switch on one side of the level opens a door on the other side is LAME! That would never happen in the real world and it won't make any sense to the player. All of your puzzles should make sense and should be able to be figured out by the player either on his own or with clues in the game. This was the problem with Zelda 4 on the Gameboy. It's a wonderful game but it has at least 2 puzzles that don't make any sense. One is that in one room you must kill three monsters in a certain order to solve the puzzle. There is a clue. It says something like 'goobla, crickle then fook' How is the player supposed to know that those three monsters in the room are 'goobla, crickle and fook'. A better clue would have been to show pictures of the monsters with arrows pointed from a to b to c.
  
  
• Trading is cheap and effective: There isn't alot of this in Zelda 3 but there is in Zelda 4. Basically have a character ask for something in exchange for something else. This is easy to implement and gives that player a goal and an anticipation.  e.g. I've got to get the dog food for the dog (goal) so I can get the bone key from him that opens the bone door (anticipation).
  
  
• Music must be scored: The music for an adventure game should be scored kind of like a movie. That means that each piece of music should make the player feel appropriate for the situation. If you are in a scary dungeon the music should make you feel trepidecious. If you are fighting a boss the music should make you panic. If you are visiting friendly characters the music should make you feel happy. Zelda 3 is great in this area. On the opposite end is Linkle Liver (or Wrinkled River) from Japan for the Saturn. The music is awful. You never feel threatened by a boss because the music doesn't get you in panic mode.
  
  

Platform Shooters

A Platform shooter is a game where a character on the screen runs around on platforms and shoots. Examples: Mega Man, Elevator Action, Turrican, Terminator, Bionic Commando. This as opposed to the more common style of shooter where you fly or drive a ship either up screen or across screen. Examples: Axley, Space Megaforce, Spy Hunter, R-Type, Scramble.

It's harder for me to come up with specific rules for a shooter. They all seem the same at a certain level and yet some work and some don't. Here are some ideas on what I notice about my favorites.

• Mega Man 2 vs all other Mega Man games has the player using all of the weapons or at least many of them very often. Most players once they get the Saw Blades or the Boomarang use them as their new standard weapon unless they run out of ammo. On all other Mega Man games the default weapon is basically the only usable weapon which is a frustrating weapon.
  
  
•