
Fukio “MTJ” Mitsuji – 1989 Developer Column
This 1989 feature, originally published in Gamest magazine, is part 3 of a 12 part series covering various aspects of console and arcade game design. We published part 5 (a three-person interview) earlier this year; this section focuses on enemy behavior algorithms, how to measure player improvement, and many other subtle points of game design. Filled with gems of wisdom that still hold valid today, it serves as both essential reading for aspiring game developers and sage critique of Japanese game development of the era.
There’s something I’d like to state up-front: I don’t intend to offer instruction on relatively unimportant matters like “tips for writing game specs”, as such trifling skills are best learned after joining a company. With this pre-amble, I’m ruling out superficial topics like those from the outset and diving straight into substantial topics that will be useful well into the future.
==exploring the fun of “the battle of wits”==
Every game promises some degree of conflict between player and enemy, but what would happen if the enemies simply ignored the player and refused to “play” with him? The player will assuredly leave the game due to boredom. In other words, it’s only when the game fights back that the player is motivated to continue. Conversely, the greater the depth and intensity of “enemy-centric games”, the more polished and interesting the experience will be. How to imbue the enemies with a sense of intelligence and life: that’s the primary purpose of the “enemy algorithm”, and it’s one of the secrets to making arcade games with more engaging behavior…
Pac-Man is a good early example of a game that successfully harnessed this idea; around that time, each enemy type began to behave in a more individualized and human-like manner, and from that point, enemies were designed with distinct personalities as a matter of course. However, it seems to me that this process has reached a stumbling block: put simply, once the enemy patterns have been committed to memory, those once-deadly enemies have no more tricks to ensure their survival!
If you were to enter the world of a video game as an enemy, think about how you might engage the player: wouldn’t you apply different tactics or feints in order avoid being defeated the same way twice? At the very least, you wouldn’t repeat a recognizable pattern. Some of you might be told, “it’s just a game, you don’t need to overthink it”, but I personally think it’d be nice to have games that provide a deeper, more intellectual level of interactivity.
My most recent game, Syvalion, was borne of this concept; in this game, the enemy puppet-master reacts to the player’s improvements and shifts its tactics by doing things like sending tailored enemy spawns within the same round, with the overarching idea being to adopt new strategies or feints according to the actions of the player. Since we were able to set a precedent with this experiment, I’d like to think more games will be released that utilize this system… Pattern-based games are simpler and easier to make, so you see this method is a pain in the neck, but as a player I'm hoping to see more games like this.

Of course, it goes without saying that the game rules and fundamental enemy behavior need to be rock-solid; it should also be noted that even a game that eschews patterns needs to be something the player is able to decipher and learn, or else they can’t be expected to improve. I think I was able to prove with Syvalion that even an “anti-pattern” game can be learned, and the high number of score-players who’ve emerged around the game make that fact clear. The more intelligent and life-like the enemies, the more fun they are; from this point on, we may no longer be able to say that self-improvement is the sole privilege of the player. Isn’t it perfectly fine for the enemy to improve, too? I think that’s what makes for a true “battle of wits” — but remember, simply giving enemies more health and faster bullets is not the solution…
==thoughts on game difficulty==
In the “de-patterned” style of game I mentioned earlier, the enemies’ attacks change every time you play, so you might be thinking, doesn’t that make the game more difficult than a regular, patterned game? Not necessarily — the game isn’t automatically tougher just because the enemies attack in a different way every time, and even if the enemies’ attacks are constantly changing, if you can intuit their actions, then you can take the appropriate counter-actions.
You can also take measures to make the game easier: for example, a boss that continually changes its attack patterns but can be defeated in a single hit couldn’t be easier. “Difficulty” is not something that can be evaluated by just one criterion.
In some cases, games based around raw memorization are far more difficult, and in that case, what’s the purpose of making a game difficult to begin with? Certain players might say, “the fun comes from overcoming each obstacle as you improve, so difficulty exists in order to motivate the player to keep at it.”
“Coin operators” — in laypersons’ terms, arcade operators, managers, clerks, etc — and arcade game manufacturers might might say, “Players need to be satisfactorily entertained for a certain length of time, and the game also needs to generate profits… that’s our challenge, and difficulty is the solution.” and so on, and I don’t think either of those answers are invalid: the player’s not going to find a game interesting unless it freely provides a certain degree of playability, and on the buyer side, they’re operating as a business, so games that allow the player to play for too long (without inserting more coins) may be seen as commercially unviable and disappear from circulation, even if the players find them enjoyable.
Arcade games always feature rigid, finely-targeted difficulty in order to generate customers; that tack is motivated by commercial concerns, but in the case of home games, it’s less of a priority because you no longer have to worry about play time. Among those in the industry, 3 to 5 minutes is commonly understood to be the ideal play time for an arcade game, and every arcade game developer is trying to figure out how to make their games more fulfilling within this short window of time. That’s all well and good, but there seems to be an endless amount of arcade games released that feel like they’ve just stolen all your money and given you no satisfaction or desire to try again. What’s more, it’s fair to say that there are many instances of games failing in the market due to over-zealously pursuing player turnover. Setting the correct difficultly level is, in fact, extremely difficult.
At the moment, the player base at arcades is polarized between highly-skilled players and newcomers, and it’s safe to say that making something that satisfies both groups of players while also generating good income has become much more difficult than ever before. Unfortunately, the current game center landscape is one where arcades are dominated by games that only a narrow group of advanced players can play, to a degree that even they are finding hard to stomach, but because it continues to be profitable, that’s just the way things are.
I don’t think my saying so will have any real impact, but I think game designers and distributors should think about the future of arcades in the long term —their focus shouldn’t solely include the diehards and super-players but also the general public: salarymen, women and the children who will grow up and become dedicated players. Those who only think about short-term profits are doomed, as their short-sightedness will eventually come back to bite them… this is something that shouldn’t just be taken into consideration by game makers and distributors, but also the operators at each venue who are determining the settings for difficulty, lives and extends on each board.
==reducing polarization==
As I just mentioned, the polarization of players has made it very tough to design games, but is there a way to bridge this gap?
I experimented with a system in Syvalion that monitors the skill of the player in real time and adjusts the difficulty curve based on the player’s response times and intuition, allowing players who cross a certain skill threshold to meet new, powerful enemies that they’ve never encountered before, which also serve as a sort of benchmark for improvement.
Additionally, we also added a “beginner’s course” for newcomers; this beginner/advanced system is something seen in many Atari games, but it seems that even Atari struggled to make their advanced modes sufficiently appealing that the majority of players would only want to play the advanced mode, and when you think about it, it’s only natural — people rarely volunteer to make things harder for themselves, after all, and if that’s the case, then the value of those distinct modes, not to mention all the work that went into making them, has been nullified.
With Syvalion, I tried to add as many exclusive, attractive features to the advanced mode in order to make people want to play that mode, even in spite of its difficulty, and in that respect, I believe I was successful. Another game, Rabio Lepus, was the first Japanese arcade game to implement this kind of course selection system, but I get the impression that most people declined to try the expert course. Why is that so? Go ahead and analyze this as a research theme with your friends.
As a creator, I’ve already devised one slightly odd method of breaking down the walls of polarization; all of you out there should rack your brains to find your own methods, too.
==half fun==
Each form of play is governed by its own set of rules; players accept the existence of these rules, make mistakes one after another and gradually intuit the correct course of action, and once they reach the sudden realization that they’ve gotten better at the game, they’re met with surprise and elation. In order to make the player feel the interestingness of a game, I think it’s very important that, even when they’re defeated, the player comes away thinking “oh, that was my fault, I should’ve done x just now… maybe if I try y it’ll work? Alright, let’s give it another shot!”.
In order to make a game truly feel fun, it’s crucial to impart upon the player the feeling of always being able to improve and always being able to visualize a way to improve the next time they play.
With this in mind, for Syvalion, I implemented a feature for the first time that gives the player written hints on how to progress whenever they die — needless to say, if the player’s defeated in a manner that doesn’t suggest even the slightest possibility on how to progress, their interest level will be immediately slashed in half. If one thing’s for sure, it’s that the game must always present the player with hints that are comprehensible, which in turn means the game designers have to calculate every step of the game in great detail. An obvious example of this approach is Dragon Quest: every inch of the world is thoroughly and meticulously designed to ensure that whatever hint the player requires to advance is always within their grasp. Why shouldn’t arcade game designers take a page from Dragon Quest’s book and similarly present possibilities to players with the same degree of doggedness?
Then again, what if you were making a puzzle-solving game — at one point or another, you’d probably be tempted to be a little cruel and deliberately toss in a puzzle with zero hints on how to progress, as if to say to the player, “oh, you don’t have what it takes for this one, huh?”, right? Whatever your approach, once the big rush from nailing down your concept for the game has subsided, it’s the game designer’s job to thoroughly and consistently apply that approach across the entire game, and even Dragon Quest maintains that same level of consistency from beginning to end. If you’re going to do something, go all the way, or not at all! I hope you’re able to come to your own conclusions about player improvement and possibility and make full use of these concepts in your future game projects.
==don’t be swayed by hardware==
Fundamentally, the first step in making a game is to formulate an idea, and then software has to be created in order to realize that game on hardware. Making games to a particular hardware specification is typically the domain of home games, but even a lot of arcade games are made for specific hardware nowadays — this is a by-product of the business model of using the same standardized board (circuitry) across multiple games, which we call “system boards”. (By the way, the industry term “conversion” refers to games that subsequently reuse existing hardware or cabinets.) For the buyers, these are all appealing traits, as they allow arcades to be able to purchase several games at a reduced price… but, the question we need to ask is, how long will the quality of new games be able to hold strong against the constraints of being designed for conversion kits?
I don’t know if it’s because developers are being ordered to make games that rigidly demonstrate the specific characteristics of a given board, or that business requires a certain amount of games to be made for one piece of hardware, but if we as an industry are still being forced to make games in order to meet arbitrary quotas, what’s the point of making games to begin with?
This isn’t just an issue faced by newbies, either, and so as a true lover of games, I’d like to encourage everyone to avoid making games in this manner — if possible, I’d like for prospective designers to think openly about their game ideas without being steered by hardware; also, during the concept stage, don’t worry about whether or not your ideas can be realized via software, as that’ll only put a damper on your creativity. Technology continues to evolve at a rapid pace, and any ideas that can’t be realized right now will surely be possible soon.

If your boss is constantly saying things like, “these ideas are pie-in-the-sky nonsense!” or “that’ll never work, you idiot!”, then that person is failing to properly nurture the talents of his subordinates; I think one of the most important talents one can nurture is being able to think seriously and deeply about how to elaborate on “fanciful” ideas. Either way, if you take somebody who hasn’t even learned to develop ideas without restraints and then ask them to come up with something for hardware that forces very strict restraints, they’re really going to struggle, so that’s why one ought to always give priority to their ideas, and worry about the hardware later.
It’s also easy to smother your game ideas if you’re too enthusiastic about showing off particular features of the hardware — to give a specific example, there have been dot-eater games that went crazy with screen scrolling, so even though they’re keeping in line with advances in technology, being unable to see the entire maze at a glance meant that the core gameplay was significantly weakened. Whatever the case, don’t overlook the forest for the trees by getting carried away with showing off new hardware gimmicks.
==measuring improvement==
Have you ever wondered why video games have score counters to begin with? Micom BASIC’s Kohji Kenjoh once wrote an article on this topic, in which he said something like, “Originally, scoring systems were meant as a means of quantifying and visualizing a player’s technique and advancement; however, there are many games produced today for which the purpose of the score counter is extremely unclear…”, and I remember thinking he was spot on.
In fact, I’d say the vague and sloppy scoring systems present in most games nowadays can’t be considered the slightest bit useful as a gauge on player skill — some games give way bigger bonuses to secret items than genuine technique, rendering their game completely pointless to play at a serious level, and others allow for massive score by merely repeating some simple, monotonous task, and if a player who doesn’t know about this exploit is suddenly made aware of it, they’re likely to completely drop the game due to it being “solved”. You might consider this an inconsequential aspect of your game, but it’s important to understand that neglecting aspects of your game that might seem minor to you can, in the worst case scenario, bring down the overall quality of your game, as in the example mentioned above.
Even games without scoring systems will feature some sort of “yardstick of improvement”: in one game, it might be an experience system, while in another, it could be story progression. Whatever the case, whether your game does or doesn’t feature a scoring system, make sure that whatever “yardstick of improvement” your game uses is the correct one for the task, and that your “yardstick” isn’t bent or even broken.

If you've enjoyed reading this interview and would like to be able to vote each month on what I translate, please consider supporting me on Patreon! I can't do it without your help!