Iteration: Making Great Games

There is a reason that only a few game companies consistently release amazing games. There are three main ingredients to making a great game: capital, talent, iteration.

Capital consists of the money and time to make the game. It always takes longer and costs more than you expect. This is an entire article in itself.

Talent is one of the most challenging factors to get right and complete. There is a finite number of talented people in the industry at any one time. Trying to build a team of some of the best game development talent is getting harder and harder--the good people have already been gobbled up by the publishers with the most capital. This isn't saying its not possible to have an amazing team come out of nowhere (the Portal team from Digipen who developed Narbacular Drop for example), but it's rare.

Once you have a solid base of capital and the talent in place the next step is build your game. Making the game itself actually isn't the major issue--its knowing how to take whatever project it is and iterate on it until it's good. Get your game playable as early as possible and have people play it. If your own staff isn't playing it, then something is wrong and needs to be fixed.

My first exposure to this type of focused iteration was in the development of Metroid Prime. When we were several months from release, we had most of the game play systems finished and several areas already in the game and playable. We started inviting outside playtesters (friends and family) to come play and would monitor how they proceeded through the game. We learned where they got stuck and how much fun they were having in various parts of the game. After each session we could discuss the results and make changes to the game. For example, we noticed many players getting confused about where to go when arriving the second time through the central Ruins area. The fix was simple--the next area they needed to jump to wasn't clearly visible and we redesigned it to be much more distinctive from the rest of the room and suddenly people would start jumping there. It's these small changes (just a texture essentially in this case) that you won't realize until you watch people explore and learn your game.

During Left4Dead development, the game was playable for a long time. Every week we would invite outside people to come playtest and we would monitor their progress through the game and learn what cues and other details they would misunderstand. Left4Dead was interesting because the characters themselves could assist in helping players game: Zoey would say, "Francis, use your medkit." to remind players to heal themselves. Of course, that lead to the problem of helping players know which character they were playing (since you are in a first person view). This resulted in the faux movie posters being displayed on the loading screen clearly telling the player who they are going to be playing.

As well as knowing what to tune and address, a large potion of playtest iteration is knowing what to throw away. As mentioned in one of the early commentary modes of Left4Dead, the melee swing attack used to be able to push other survivor players around. Of course, when we started playtesting people immediately realized your could use it to push other players of the roof. This feature wasn't essential to gameplay and was only really useful for griefing so it was removed.

Game development iteration is the process where you learn what you did not communicate well to your players or did not tune sufficiently for them to have fun. You need to correct this and iterate again until you find your players are having a great experience. You also need to know when you are not succeeding in some aspect and throw it away.

World of Warcraft had a beta that lasted more than a year. They constantly iterated on the game and kept adding content when most companies would have pushed it out the door. I participated in the early alpha and beta of WoW and noticed they redesigned and iterated on the starting zones a half dozen times--Northshire Abbey (the human starting area) was rebuilt several times to constantly tweak and adjust it to get it right. Warhammer Online and Age of Conan both had the opportunity to have new takes on the genre and make themselves distinctively different. They were either running low on capital or didn't realize they weren't fully ready and pushed the game out the door before it was perfect. It's unfortunate in that both could have used months more time in beta in order to iterate and make the games the best they could have been.

Companies like Valve and Blizzard are fortunate in that they have the talent and capital in order to iterate and keep working on the games they are developing until they are the best they can be. They take years to make games, but the investment is almost always worth it because the released products are of such high quality.