This is part two of an x-part series summarizing the design of one of the greatest games of all time, SOTC. Part 1 can be found here.

Welcome back to Design Theatre! Today, we’ll take a look at how Shadow of the Colossus (SOTC) was designed in its first major sections of gameplay.

There are sixteen bosses in Shadow of the Colossus, each with a unique twist on the “stab its vitals” formula. The first boss, a giant wandering colossus with a club, is by far the easiest, only requiring the player to climb up its leg and stab its head to defeat it. However, at this early point in the game, the player doesn’t yet know they have to do this, and since the player has likely never faced something this large in a game before, they panic. The battle serves as a tutorial, teaching the player the fundamentals of how to defeat a colossus, without giving them all the answers. For example, it is hard to imagine that the latter bosses of the game require the player to utilize their environment in creative ways to defeat them, yet the goal – climb the colossus and stab it with your sword – remains the same. From this section we learn:

#5. In the beginning of a game, introduce a short-term goal that shouldn’t change for the duration of the game.


Most games need a clearly-labeled short-term goal for the player to understand where they need to proceed. A lot of the problem with early point-and-click and text adventure games is that a short-term goal doesn’t exist. The player may even have to discover the long-term goal in these games, which usually answers the question: “Why am I here, playing this game? Why should I care?” In Super Meat Boy, the short and long-term goals are simple: reach bandage girl to beat a level, and beat the game to save bandage girl from Dr. Fetus. Clear, and effective. But not all games are like this. The Dishwasher: Dead Samurai was a game with great gameplay, but varying long and short-term goals, making it unclear what the player should do and what the end result of their toil would be. In Portal 2, the expanses during the middle section of the game – where the player isn’t sure where they have to go to proceed – disobeyed the short-term goal rule for too long a time, making the game needlessly frustrating and less enjoyable. Left 4 Dead also has great gameplay and a clear short-term goal, but lack of a narrative thread to tie each chapter together really ruined the game for me; unlike Portal, which started out with great puzzles and clear short-term goals but gave the player “story nuggets” that tied the game together. The effectiveness of long-term goals in single-payer games warrants our sixth lesson:

#6. Have a long-term goal established.

Very few people, if any, want to play a single-player game without end-game rewards. This should be pretty obvious, but that opening cutscene at the beginning of Metroid seems to escape many new designer’s minds, choosing instead to jump right into gameplay, as if their gameplay spoke for itself. This, however, is simple not true — everyone enjoys a little story. Most of all, it prolongs the fan-base of the game, which as Half-Life and Mother 3 have shown us can survive long past the normal production time of game sequels.

In SOTC, the player controls a boy named Wander. Unlike most game characters, who appear strong and have great dexterity, Wander is pale, fragile; he stumbles and struggles his way up colossi. My friend got frustrated at Wander, yelling “Why won’t you do what I tell you to do?!” during the final boss fight of the game. While this lack of control could be attributed to the inaccuracy of PS2-era adventure games, Wander’s animations are definitely not arbitrary. In addition, Agro, your horse, is very hard to control – while the developers could definitely have made Agro go exactly where you want him to, in SOTC your horse is not a transportation device, but a main character, with its own freedom of movement. Because of Agro, SOTC’s horseback-riding seems the most realistic of any game, frustrating at times and glorious at others. Here we move to the seventh lesson from SOTC, perhaps the most controversial:

#7. Games do not always have to be fun.

Oh god, here we go. Many critics place extraordinary weight on how the game controls, how fun it feels. However, because of Shadow of the Colossus, I realize that sloppy controls can be a design choice just like any other, provided they fit the context of the game. Wander is supposed to look weak; his skin grows paler and more bruised as the game goes on. Agro is supposed to be hard to control; horse-riding is not a simple endeavor, even though many games make it seem so. The sloppy controls diminish the fun of the game, but really, is SOTC supposed to be fun? Is tacking down 16 colossi by oneself without much food or water while one’s soul is destroyed fun? Not really. So what sense does it make for the controls to feel unrealistically responsive, all for the sake of fun, when fun doesn’t fit the role Wander plays?

I remember Cactus’ talk at GDC, where he brings up this point. It is overlooked in an industry devoted to entertainment, but has it roots in more established mediums. Take a look at the IMDb Top 250 for a second. The Shawshank Redemption? Not what I’d call ‘fun.’ Schindler’s List? The Godfather? You can see where I’m going with this. Most of the greatest films ever,  like most of the greatest books, are not ‘fun.’ We need to move out of the rut of making games fun for the sake of doing so. Ask yourself: should parts of my game be fun? Does the gameplay fit the context the player’s avatar is in?

After the player defeats the first colossus, he/she goes up against a colossus that cannot be simply climbed. To climb this colossus, the player must use their bow and arrow in a real situation for the first time. Introducing a boss that requires the player to use a bow and arrow this early is a smart move: if the game did not welcome the player to shoot arrows, they might not realize the bow and arrow is necessary at all.

#8. If the player can do something, show them why they should early on.

If a weapon or object is necessary for gameplay, introduce it early on. Suppose you are developing an RPG-adventure game, and the player can pick up loose items lying around to store in their inventory. Now the player can collect keys, but keys aren’t needed for the first half of the game. Then the player reaches a locked door – do they realize that they can go to their inventory and double-click on a key to open it? Okay, so maybe that’s a little far-fetched, but you get the point: if the player can double-click on items in their inventory to impact the environment, then show them that they can do so early in the game.

Well, that’s all for Part 2! See you next time! ^_^