Moved this post ahead since this week’s Famitsu contains an important announcement: Yoshida’s Famitsu column will be ending soon (around #139, I believe), because he has too much work and doesn’t have time to write it anymore. I’ll still translate all of them, of course! But it’ll be a shame to see the column go… (Understandable, though. Yoshida’s time must be extremely valuable at this point)

As for what comes next… well, I have 32 columns to think about it lol.

#107: “The Fear of Simplicity”

Published in 2018/04/26 issue

With April comes a new fiscal year. Normally, I’d be busy with company events and HR-related things at this time of year, but for some reason, I’m in Saudi Arabia. I’m not the kind of person that gets surprised easily, but the events that led to my Saudi Arabia visit had me going, “Wow, things like this can happen?” I’m not sure if I’m allowed to talk about it yet, so I’ll write about this trip at a later date.

I’ve been in Saudi Arabia for three days now, and between students, over 40 FFXIV players (which was a surprise), and people hoping to enter the game industry, I’ve already spoken with many different people. While I was speaking to these people, who were born and raised in Saudi Arabia and want to work in video games, I remembered that there are also many new grads in Japan who started working in the industry this month, so today’s column is directed towards them.

I think that most work assigned to fresh new planners (called designers overseas) in the game industry falls under data entry. This could be a variety of things; for example, quietly inputting values into Excel that your seniors came up with, or converting written dialogue into a computer-readable format. If you’re assigned to a project that’s nearing its deadline, you might even get “lucky” and suddenly find yourself tasked with inputting monster parameters because there’s no time left.

With the dramatic evolution of video game hardware comes a dramatic improvement in presentation. However, it’s still a matter of communicating with the computer. If you dig deep enough, everything is made up of 0s and 1s—in other words, “on” and “off” commands. This will be true even hundreds of years in the future. So, when a newcomer to the game industry is told to “make monster parameters,” they might get confused: “Huh? With… words? Do I write how they should act?” Back when I first joined the game industry, graphics fidelity was very poor, so I immediately assumed it’d be with numbers. But now, in today’s age, it’s hard to immediately figure out the process for “breathing life into a monster.”

The enemies and monsters in video games act under a set of various parameters. Even if it looks like they use an AI (artificial intelligence) program, AI itself is only the principles that govern enemy or monster behaviour. The way you give them “life” is by punching in numerical values.

For example, let’s say we want to use AI to make a character give out presents when they’re in a good mood, and attack people when they’re in a bad mood. This is ultimately just a branching set of actions, not a personality. Here, we’d set the numbers such that when the “mood” parameter is above +30, they’re considered in a good mood, and when it’s below -80, they’re considered in a bad mood. “Mood” is a parameter type, and the character’s personality is created by inputting values into it.

After combining these values, we can see that this character’s personality is “sociable and patient.” The upper threshold for their mood is +30, so they get in a good mood pretty easily. On the other hand, the lower threshold is -80, so they won’t get angry unless their mood parameter drops by quite a lot.

This is a pretty extreme example, but it’s fine to assume that most game characters are created this way. Then, the problems that stem from here are “How many parameters should we have?” and “What should the values be?” In the above example, “mood” is a parameter and “+30” is a value.

Fresh grads joining the game industry are extremely motivated, because they finally get to create things for the world to see. They eagerly think, “All right, I’m going to make my monsters unique and exciting!” But because of this, their first job often ends up being horribly painstaking. This is because they make too many parameters in their attempt to create an appealing character.

Using the previous example, in addition to mood, they’ll create parameter fields for physical strength, agility, endurance, luck, and so on. They also tend to think, “Wouldn’t I be able to create a better character if I also had detailed parameters like charisma and favourite foods?”

But, is “charisma” really necessary? What do you want to make the character do using that parameter? How is it going to bring enjoyment to the player? You have to consider those questions first. When you add so many parameters with questionable uses, you end up with characters that won’t be unique unless you fiddle with several values.

For example, let’s say you set their “favourite food” to “ramen” (which will actually be a numerical ID, such as 3). However, if you look across the entire game and only see two instances of ramen, then that character’s preference for ramen will only be relevant two times. Plus, in order to make this fondness for ramen a unique trait, you need to set other favourite foods such as steak or curry to the other characters, and now you have to make it a game with a lot of different foods in it. It’s putting the cart before the horse.

Building up all of these different parameter fields also becomes a hindrance when it comes to mass-producing characters. If you need to create 200 characters, then you’ll of course need 200 personalities. But the more parameters you have, the longer it takes to input values for each character, and that extra time is multiplied by 200. Plus, you’ll be forced to set more detailed values in order to make them unique. Then, when you implement them into the game and do a test run, you find that the characters still feel samey, so you go back to tweak the numbers—but since there are so many fields, you end up not knowing which ones to adjust.

Of course, for games where the characters themselves are the selling point (such as idol raising games), increasing the number of parameter fields makes sense. However, even in those cases, you still need to be careful about whether each field is truly necessary or not. Fixing the structure after the fact requires an insane amount of work.

Creating data in a simple manner can actually be scary. Having more parameters makes it feel like you’re doing more work, and there’s an odd sense of security that comes with that. When you’re feeling that way, I implore you to try playing Tactics Ogre or Vagrant Story with an official guidebook in hand. These games are like textbooks in the way they bring their characters to life with a tight set of parameter fields and values. There’s a lot you can learn from your predecessors, and good examples are readily available.