Chaos Seed – 2010 Developer Interview
originally featured in PLANETS magazine vol. 7
Junichi Kashiwagi – Designer/Creator
—There are two versions of Chaos Seed, one for the Super Famicom and one for the Sega Saturn. The Saturn version sold less than 20,000 copies, but in a readers ranking survey in Sega Saturn Magazine, of 1156 titles it took 15th place. I think we can call Chaos Seed an actual hidden gem. Today, I’d like to talk with you about its game design, and take a look back at those days. I understand that before you joined the game industry, you were a hardcore gamer and game center devotee. That reminds me of Pokemon’s creator Satoshi Tajiri, who was also a big gamer first.
Kashiwagi: Yeah, I was too. In the 80s I spent a lot of times in game centers, and worked part-time as a pixel artist for a game company. I sometimes helped out with design and planning there too. One person I met during my time at the game centers then was Tadakatsu Ogura, who currently works as a programmer for Pyramid. We really hit it off, and together we made a STG on the X68000, among other things.
—Are there any games that you feel have had an especially profound influence on you?
Kashiwagi: It’s hard to name just one! I’ve always been someone who played a diverse selection of genres. No matter what the title, if the game is really well-made, that feeling of awe will suck me in and I’ll get lost in it. In that sense, there’s too many games that have influenced me—too many to choose one.
It’s also true that, in my job, I don’t have the luxury of being picky about what I develop. My stance is to try and do the very best job of whatever I’ve been asked to do. When a client asks me to “make something amazing”, I wrack my brain trying to come up with something great. Even when they say “make something no one has ever made before”, I do my earnest best to make something new. For better or for worse, I can’t say, but my approach to making games is more that of a craftsman than a fine artist.
Designer Junichi Kashiwagi (L) and main programmer
Tadakatsu Ogura (R), the two principal creators behind Chaos Seed.
—That makes a lot of sense, actually: as a craftsman, you’ve garnered a wide and refined appreciation for “gaming culture” generally. Chaos Seed is a real alchemical feat of genres and ideas, a sort of chimera itself. And I think it’s that understanding, rather than a specific leaning or predilection for any one genre, which made it possible for you to create a game like that. Did you transition straight from making doujin games with Ogura into professional work in the games industry, then?
Kashiwagi: No, after doing pixel art jobs for Famicom and Megadrive games, I enrolled in a technical school for animation. For awhile I continued doing pixel art as a part-time job, but eventually I quit because I wanted to focus on my graduation thesis. To be honest, I was torn between entering the animation industry or the games industry. But partly because of my own inherent lack of talent in drawing, and partly because Ogura asked me to work on games with him during my school, I finally decided on games.
Ogura was originally employed at Right Stuff, but Takada of Neverland Company had just opened a new office and invited Ogura to join him there, and Ogura, in turn, invited me. Neverland was formed by ex-Wolf Team staff, and before Right Stuff, Ogura had worked for Wolf Team, where he worked with Takada. Anyway, joining Neverland is where my career in the game industry really began.
—How did the Chaos Seed development get started?
Kashiwagi: At the time, what I wanted to make was a Dragon Buster-style horizontal scrolling action game. I got my start doing pixel art, and I enjoyed making action games with cool character animation. In fact, the company I had been at just before this had been making a fighting game, and I thought I could continue in that tradition, but Neverland Company was a subcontractor of Taito, and they had been successful making RPGs for them. So I was told it would be better to focus on something closer to an RPG.
—You must be referring to the Lufia series, that Neverland Company developed. Had you ever thought about making an RPG before?
This video (as well as hg101’s write-up) make a good introduction to Chaos Seed, for those who have never played it.
Kashiwagi: I had. I thought, “If I make an RPG, it will be one that reverses the roles of good and evil, where the player is creating the dungeons.” At that time—the first half of the 90s—stereotypical RPGs had reached a saturation point, and the game industry as a whole was searching for a new direction. I, too, wanted to make something that turned things upside down. I told my ideas to Ogura, who in turn presented and explained them to Takada. He thought it was interesting, and that was how Chaos Seed got started.
—For the benefit of readers who haven’t played Chaos Seed, let me summarize it a little here: in Chaos Seed, the player takes on the role of a taoist mountain hermit and creates dungeons. You summon monster units, administrate your dungeons, and repel invaders—really the polar opposite of what a traditional RPG protagonist would do. The very concept of the game is a defiant challenge to convention; did you have a blueprint or master plan you were working off, then?
Kashiwagi: Actually, in the beginning, the setting wasn’t China. It took place in your typical Western European fantasy. The protagonist was an evil priest who looked after the dungeon. He starts expanding the dungeon in order to summon the Demon King, and he issues a proclamation to the public: “In this very dungeon, our God will soon be revived! Come all ye brave and powerful warriors.” The priest lures those strong heroes to the dungeon, and harvests their souls in order to revive the Demon King. However, once summoned, the Demon King doesn’t do what the priest says, so the game ends in a final climactic battle between these two villains.
Unfortunately, Taito read all that and told me directly, “Don’t make the protagonist a villain,” putting a quick end to that idea. (laughs) So I changed the idea to one where you’re a mountain hermit, who is actually trying to do good, but whose actions are misunderstood by the outside world. That’s how the world changed from Europe to ancient China and Taoism. Taoism is really crazy, and its gods include Christ, Mohammed, and Buddha, all muddled together in there.
The complexity of Chaos Seed has made it virtually unknown to non-Japanese speakers, but there is now has a complete English fan-translation courtesy of Dynamic Designs.
—There’s that word, muddled—and in a game design sense, too, Chaos Seed is a crazy mix of ideas. (laughs) It’s a very complicated game. When you’re making the dungeons it feels like a real time strategy game, and when you fight with intruders it’s like an action game, while the story has elements of a multi-scenario game. Were there any precedents—specific games you had in mind when you created Chaos Seed?
Kashiwagi: For the action side, Legend of Zelda: A Link to the Past had just come out for the Super Famicom. It was extremely well-made, and it influenced the character animation.
—Zelda takes a long time to complete, though. In contrast, a single playthrough of Chaos seed takes anywhere from 2-6 hours. Also, in a lot of strategy games, even if you do a kind of half-assed job in the building process, after a few hours you reach max power and can just bulldoze your way to victory. I feel like in Chaos Seed, you avoided this problem by ending the game just before the player has a chance to get too powerful.
Kashiwagi: Hmm… maybe so, but I didn’t really try to fine-tune it like that. It was more that I tried to balance it so the player could do things as he liked. You can make the scenarios more difficult anyway, if you replay them after clearing them once. So I hope it doesn’t make anyone angry to hear this, but the balancing wasn’t done with laser-precision or deliberate intention.
—Do you mean you wanted a system that would permit a wide variety of playstyles and approaches?
Kashiwagi: Yeah. To give one example, the strength of the basic enemies are determined according to the Dousen’s (the main player character) level. That’s why if you keep Dousen at a low-level, but level up the ally monsters, the game can become really easy. But I also randomly add in enemies that are the average level of your units, which makes the player feel like things have just got harder. Basically, I knew that a perfectly, thoroughly balanced game from start-to-finish was going to be impossible, so I tried to design a system with the capacity to adapt to the player’s actions.
—Did you not do any of the normal playtesting then, where multiple testers play through the game and give you their feedback?
Kashiwagi: The testplayers only did debugging, and everything else (balancing etc) was left up to the opinions of our developers and the other company staff. That’s how we did it for all our games back then, actually. Nowadays, for level design, we have a more detailed playtesting process in place, but in Chaos Seed the player creates the map himself, so there’s no “level design” per se. We did set the time-table that governs when enemies attack… but it’s kind of embarrassing to call that “level design.” (laughs)
As emphatically described below, Chaos Seed was a blend of different influences. Herzog Zwei had a huge influence on Kashiwagi’s thinking with regard to units, while the Legend of Zelda: A Link to the Past inspired the action and character animation. Meanwhile Dragon Buster, Namco’s early sidescrolling arcade hit, was closer to the game Kashiwagi originally wanted to make, but the success of Lufia (not to mention the participation of its two programmers, Ukeda and Suzuki) drove the Chaos Seed development in a more RPG direction.
—For the simulation aspects of Chaos Seed, were there any particular games you were influenced by? The mid-90s was a flourishing time for the RTS (real-time strategy) genre. For me personally, when I first played games like Age of Empires and Warcraft, I distinctly remember feeling like “this is fun, but I’ve experienced this before in Chaos Seed!” I feel there’s a lot of similarities in the design between Chaos Seed and RTS games.
Kashiwagi: Yeah. As far as game structure goes, I was very influenced by Tecnosoft’s game Herzog Zwei for the Megadrive. The RTS genre was especially popular in America, but if I had to say what the model was for Chaos Seed, I think it was the Japanese game Herzog Zwei.
Before that were strategy games like The Ancient Art of War and First Queen, but those games only scrolled up and down, with your battle line either advancing or retreating… they were very simple strategy games. Herzog Zwei came along on the Megadrive felt like a revolutionary upgrade: now the strategy involved maps, where you had your own base and territory, with building production, unit AI, etc. It contained the seed of all the later RTS mechanics—and it was all controlled with a controller! It was quite the innovation. Herzog Zwei directly influenced the strategy aspects of Chaos Seed, to the point that I would almost consider it a direct descendant.
—What mechanics, specifically, were influenced by Herzog Zwei?
Kashiwagi: Mainly the way the units all have their own individual uses, and how it’s presented in a clear way to the player. All of Herzog Zwei’s units—planes, robots, anti-aircraft, tanks, infantry—they’re all created with a special strategic purpose in mind. You can’t conquer an enemy base without infantry, you can’t take out planes without anti-aircraft, and so forth. The system is very easy to comprehend. Another point I took inspiration from is the way those units will carry out their general orders automatically in real-time. I modeled much of the units in Chaos Seed after those in Herzog Zwei.
Obviously they aren’t exactly the same, but a lot of my basic thinking comes from that game. However, the difference between Herzog Zwei and Chaos Seed is that the player creates the very space in which he places units (ie, the dungeon) in Chaos Seed. If you want to make some special room operational, for instance, you’ve got to build another room or hall to connect the rooms and transport energy. In that sense, I wanted to create a game that put more emphasis on the concept of provisioning and supplies, as well as healing/repair units. Those things are often overlooked or ignored in war games.
Chaos Seed’s game design is essentially a combination of those Herzog Zwei RTS mechanics and a player created map, all done in real-time.
—I wanted to ask about the story/scenario of Chaos Seed too. Masahide Miyata, who wrote the scenario for Lufia, did the writing for the Super Famicom version of Chaos Seed; you wrote certain sections of the Sega Saturn version. I noticed that the scenarios you and Miyata wrote differ considerably in their structure and style.
Kashiwagi: The main difference is the last chapter of the main story, the karma scenario. I’ve never really had a chance to talk about this… Miyata has a lot more fans than me, so there’s never been a good opportunity to go into it. I’m sure those fans want to know why I rewrote Miyata’s scenarios (laughs). The thing is, in the Super Famicom version, I felt like what I had wanted to convey in the story hadn’t really come across… so when we ported it to the Saturn, I changed the story a bit.
The overall structure is the same, but the lead-up to the final act is different. The SFC version has a rather light-hearted ending, but in the Saturn version things are more complicated, weighty, and a little darker. This will be a little bit of a spoiler to anyone who hasn’t played the game, but for the dark, heavy Saturn ending, I was inspired by one of my favorite books, Flowers for Algernon.
—That’s the story about the mentally impaired man who becomes a genius temporarily before regressing back, right? What about it inspired you?
Kashiwagi: RPGs of that era often tried to make players cry by having the protagonist, heroine, or other main character die during the story. I wanted to try including a “crying game” element like that in Chaos Seed, too, but I didn’t want to use a character death. That felt too simplistic to me, like a cheap narrative trick.
In Flowers for Algernon, no one dies. But the degeneration of the protagonist’s intellect is handled with great skill, and pulls on the reader’s heartstrings in more subtle ways. It combines sadness with inevitability. Chaos Seed has a similar development, and no character has to be sacrificed, but the player comes to realize what must be done and make the best choice possible given the circumstances. The player is emotionally moved, and it was Flowers for Algernon that influenced that structure. I basically stole the idea. (laughs)
—In the 90s, multi-scenario and multi-endings were starting to become really popular. Then in the 00s, scenarios with a “loop” structure were all the rage. It seems you were ahead of your time with Chaos Seed!
Kashiwagi: I came up in game centers playing STG games, so I had always envisioned Chaos Seed as a game that you could replay many times. I was originally aiming for an RPG that you could complete in 30 minutes to an hour. That necessitated a story with a loop structure, as well as a system featuring multiple endings and a lot of replayability. Ultimately we couldn’t achieve such a compact game—a single playthrough of Chaos Seed takes 2-3 hours. (laughs)
Osamu Tezuka’s Hi no Tori manga, which influenced the story of Chaos Seed.
I based the loop structure on the “strange creatures” arc of Osamu Tezuka’s Hi no Tori (Phoenix) manga. As you reach the ending, a tragedy repeats itself, and then you go back to the beginning and it starts all over. Finally you have to make a choice. Will you break free from the cycle of samsara (death and rebirth), or will it all start over again?
—I see. Although in Hi no Tori, the reader doesn’t actually get to make that choice—but games allow players to make conscious decisions, which is something a manga like Hi no Tori couldn’t do. Compared to your typical adventure game, a loop structure allows the player to see more clearly the impact and meaning of his choices.
Kashiwagi: Yeah. Multi-ending/multi-scenario games have always been few and far between. It’s because they take an insane amount of labor to make.
—As someone with a background in pixel art, was there anything you were particular about in that regard?
Kashiwagi: Visual expression with computers is something I’ve been pursuing my whole career. I try to make sure the graphics have a “tactile” or interactive quality to them. I think that really affects the player’s perception of how well-made the game is too. If there’s no tactile feedback from the visuals, then it’s very hard for people to get drawn into the world of your game. I want players to feel like there’s a real live world they’re interacting with.
For example, take the action of digging through a wall. In your typical strategy/simulation game, you just move a cursor over the wall and press a button to execute a command. That’s the most basic method…
—But you don’t feel like you’re there in the world of the game then, just by executing a command like that.
Kashiwagi: In Chaos Seed, you have to bring your character to the spot you want to dig in, and hold down on the button for a moment. Only then does the command window open up. In other games you accomplish this all with a single push of the button, but I wanted to make those actions take a little more effort. In doing so, I hoped that the player would feel a more realistic level of interaction with the game world.
—Wasn’t it difficult though, figuring out exactly how much extra work you should make the player do?
Kashiwagi: For things like how long the delay should be before the command window opens up, the staff would offer up their opinions about whether it was too long or too short, and whether it was too annoying or bothersome. Even today, when I talk to the staff who worked on Chaos Seed, their opinions are still divided on some of those choices. But I think that in the end, we did a pretty good job.
—Recently in the strategy genre, tower-defense style games have become popular. I feel some similarities between there with Chaos Seed. It really does feel like the essence of the mechanics being explored in RTS games today are a remix of ideas already present in Chaos Seed.
Kashiwagi: There really aren’t many games with such a complicated a design as Chaos Seed, nor containing that many elements all thrown together. In a sense, I think Chaos Seed represents a terminus point of a certain line of game design.
—I think that’s exactly right. Chaos Seed came out in 1996. I feel like the mid-90s was a time when the possibilities of game design in the 80s reached a peak of refinement. 2D pixel art, fighting game mechanics… it felt like, where do you go from here? It was the pinnacle of one evolution of game design. Do you think there has been any successor to Chaos Seed?
Kashiwagi: As far as the RTS genre goes, games like StarCraft and Warcraft branched off from Herzog Zwei, adding 3D graphics, networking, and a variety of other mechanics. Those games are really their own genre now. With Chaos Seed, however, I don’t really feel like it is a genre branch so much as it is a conjoining of two trunks: that is, the RPG genre and the simulation genre. And in fact, I would not call it a particularly elegant combination. But having said that, separating those halves would have resulted in a game not all that different from your average RPG or strategy game.
It’s no mean feat to fit all those mechanics into one game, so a follow-up game would be difficult to make. Also, as a result of all those mechanics we crammed in, Chaos Seed is perhaps a bit too hardcore. Veteran gamers with a lot of experience may find it very interesting, but I’m not sure it’s a game that beginners, or new players who are just starting out, can enjoy. For those reasons, i don’t think we’ll be seeing a game like Chaos Seed again. Even more than the question of whether it can make money or not, I don’t think it could be made again given the way our developments are structured today.
—What do you mean by that?
Kashiwagi: Tadakatsu Ogura, who made Chaos Seed with me, loved games that played in real-time, and his expertise lied in action and STG game development. I also loved action games, but if I was going to make an RPG, I wanted to include slightly more complicated simulation-ish mechanics. We kept adding more and more ideas to Chaos Seed as the development went on: “I want to do this!” , “Yeah, and let’s try this!”… so much that our schedule and the deadlines kept getting pushed out further and further. The company noticed this and realized they couldn’t just leave it all in our hands, so they assigned the Lufia programmers Naoyuki Ukeda and Akihiro Suzuki to us. They were able to take all our basic ideas and translate them into rules that could be used in an actual game system.
As you can see, Chaos Seed had a very unusual, singular development history, with veterans of every game genre represented on the staff. But that’s also precisely why we were able to combine so many genres into one finished game.