The History of Nintendo EAD. Part 3: 128 Marios, Puppies, and a New Nintendo

Mario 128 Was Not a Lost Super Mario Game

Let us begin with the era’s most famous ghost. Mario 128 really existed, but in August 2000 it was not an announced game. It was a technical demonstration for the GameCube, created by Yoshiaki Koizumi. Its purpose was simple: to show that the new console could process many separate three-dimensional characters at once while changing the geometry of a surface in real time.

The name was deliberately reminiscent of Super Mario 64. Viewers therefore made the reasonable assumption that they were seeing the first glimpse of a sequel. Miyamoto later used the words Mario 128 several more times when discussing new experiments. As a result, several different prototypes merged in the public imagination into one large “lost game.”

There is no evidence, however, that Nintendo spent years developing a single project under that name. We do know where some of its ideas went. Koizumi continued thinking about a platform game in which the hero ran across spherical surfaces. A few years later, that problem led his team to the planets of Super Mario Galaxy. But that belongs to the next chapter.

The Mario 128 technical demonstration shown to the public. The crowd of tiny Marios was meant to showcase the capabilities of the new console, not to present a finished game.
Nintendo / 社長が訊く『スーパーマリオギャラクシー』Vol.1 プロデューサー・ディレクター 篇 / nintendo.co.jp

At the GameCube’s launch, Mario’s role unexpectedly went to his brother. Luigi’s Mansion clearly demonstrated the console’s lighting and transparency effects, but in scale it was nothing like the successor to Super Mario 64 that buyers expected. After the Super Nintendo and Nintendo 64, this launch broke what had become a familiar pattern: a new Nintendo home console arrived without a major new Super Mario game.

This episode matters not because EAD “failed to finish Mario in time,” but because it revealed a new problem. A technical demonstration can be built in a few months and presented to the public with impressive results. Turning it into a game with a clear goal, dozens of levels, and carefully balanced difficulty is much harder.

Pikmin Did Not Grow Out of Mario 128

Pikmin is often connected to the Mario 128 technical demonstration because both placed a crowd of small creatures on the screen. This led to the theory that Miyamoto had simply replaced 128 Marios with Pikmin. Magazines, websites, and even some books about Nintendo’s history repeated the story.

In 2023, the Pikmin developers finally explained the game’s origins in detail. Work had begun with experiments on the Nintendo 64. A young programmer named Yuji Kando created prototype creatures with interchangeable behavior modules: some determined their actions, while others defined their personalities and emotional responses. They could choose what to do on their own instead of merely replaying a single preset animation.

The Nintendo 64 could not handle large numbers of fully modeled characters. The early creatures therefore looked like flat images that always faced the camera. The GameCube made it possible to replace them with separate three-dimensional beings. This is where the projects began to look similar: both teams were testing how many moving characters the new console could support.

But the Pikmin team knew nothing about Mario 128 and did not use its technology. They were two independent responses to the same hardware capability.

The most interesting part came later. The basic mechanics already worked: the player could throw Pikmin, the creatures carried objects and multiplied, and they came in several types. Yet there was still no game as such. Why should the player throw these creatures? What were they trying to accomplish? How would each day begin and end?

Miyamoto therefore returned to the director’s role for several months. He gathered all the disconnected actions on a single sheet of paper: the Pikmin help Captain Olimar find parts of his ship, each day has a time limit, lost creatures must be replaced, and the entire escape from the planet lasts thirty days. The separate mechanics finally came together as a game.

Pikmin had to be shown at E3 two months later. The team had only one demo level, but the developers now understood the direction they needed to take. The Japanese release followed on October 26, 2001—six weeks after the GameCube launched.

Shigeru Miyamoto’s diagram of the Pikmin gameplay loop. The sheet still uses early names for the creatures: Piki and Picky.
Nintendo / 開発者に訊きました：ピクミン４ / nintendo.co.jp

This is why Pikmin cannot be described as the result of one successful idea. First, the developers learned how to display many autonomous creatures and decided what those creatures could do. But much more time went into turning that collection of possibilities into a complete game. EAD’s teams would face the same problem more than once while developing for the GameCube: the new console allowed them to realize more ideas, but choosing the essential ones became increasingly difficult.

In Super Mario Sunshine, the Water Came Before the Island

Development of Super Mario Sunshine also began not with a story or a world, but with a new gameplay mechanic. Yoshiaki Koizumi came up with a way to use the GameCube controller’s analog triggers and proposed a game with a water pistol whose pressure changed depending on how hard the player pressed.

Miyamoto liked the idea, but a pistol in Mario’s hands looked too aggressive. The weapon became the FLUDD water pack. Once water became central to the game, the team added the sludge, the sunny Isle Delfino, and the task of cleaning stains from the island.

This approach differed from earlier Super Mario games. Previously, the team first created a world—a plain, a desert, a sea—and then designed suitable actions and obstacles. In Sunshine, one gameplay mechanic shaped the entire world. To help the developers visualize the island’s geography, Koizumi even modeled it in clay.

The clay model of Isle Delfino that the team used while creating the world of Super Mario Sunshine.
shmuplations / Super Mario Sunshine – 2002 Developer Interviews / shmuplations.com

Here too, freedom quickly became a problem. The team designed ten interchangeable nozzles for FLUDD, but kept only three in the finished game: the Hover, Rocket, and Turbo Nozzles. Together with the standard spray, that was already enough to give every area several possible routes.

The Hover Nozzle caused particular trouble. It could save Mario after an inaccurate jump and let him reach places that would have been inaccessible in earlier games. This meant the designers could no longer guide the player along a single obvious path. Every roof, wall, and alternative route had to be tested. That freedom became both one of Sunshine’s greatest strengths and a major source of difficulty during development.

Sunshine was released in Japan on July 19, 2002. Its responsive controls, lively resort setting, and unusual summer-holiday atmosphere brought the series something new. Yet the awkward camera, certain missions, and sudden difficulty spikes often spoiled the experience. Miyamoto later admitted that the team had made the game too difficult for the average player.

The problem was not that EAD had forgotten how to make Super Mario. The developers had excellent ideas for what players could do with water. They failed to answer a different question: how many of those possibilities should be available at the same time?

Cartoon Link Solved a Practical Problem

At the same Space World 2000 presentation, Nintendo showed a short scene featuring realistic versions of Link and Ganondorf. The demonstration ran in real time on the GameCube and looked like a more detailed version of Ocarina of Time. Like Mario 128, it was meant to demonstrate the console’s capabilities, but the public took it as a promise of the next Zelda.

At first, the team did try to develop the realistic direction. The result was polished and completely predictable. According to the developers, few new ideas emerged: they were simply building another Ocarina of Time with better-looking models.

Designer Yoshiki Haruhana proposed the solution. He drew a small Link with a huge head and expressive eyes. This cartoon look allowed the hero’s movements to become livelier and more exaggerated. Satoru Takizawa then drew a Moblin in the same style, and the team began imagining how the two characters would fight. Ideas followed one after another: the new visual direction not only separated the game from Ocarina of Time, but also shaped its animation and combat.

The first sketch of cartoon Link, drawn by Yoshiki Haruhana. It helped the team find the visual direction for The Wind Waker.
Nintendo / 社長が訊く『ゼルダの伝説 風のタクト ＨＤ』 / nintendo.co.jp

The cartoon style was not merely decoration. In a realistic world, it can be hard to tell at a glance which of dozens of detailed objects can be moved, cut, or set on fire. In The Wind Waker, interactive objects stood out from the scenery through their distinctive shapes and animation. Link’s expression communicated his mood, while the direction of his gaze guided the player toward points of interest.

When Nintendo showed the game at Space World 2001, not everyone liked the new style. Fans who had expected a realistic sequel considered cartoon Link too childish. But returning to realism would simply have broken the game: the characters’ exaggerated movements and comic-book effects would have looked absurd in that kind of world.

The Wind Waker was released in Japan in December 2002. Even today, its graphics do not look outdated. Rejecting realism was exactly what allowed the team to create the movement, combat, and visual guidance on which The Wind Waker was built.

When One Miyamoto Was No Longer Enough

Pikmin, Sunshine, and The Wind Waker are often described as “Miyamoto games,” but by the early 2000s that label no longer captured how EAD worked. He could suggest a direction, evaluate a prototype, or force a team to remove unnecessary elements. But he could not manage every project from day to day.

The department’s veterans gradually took responsibility for their own series and teams. Eiji Aonuma oversaw Zelda, Katsuya Eguchi handled Animal Crossing and new projects, and Hideki Konno worked on Mario Kart and future online games. Takashi Tezuka and Shigeru Miyamoto increasingly evaluated the work of several teams and set their overall direction instead of concentrating on one specific game.

In May 2002, Satoru Iwata became president of Nintendo. Over the following years, the company reorganized its production divisions, while EAD established several groups led by producers. Nintendo never published every detail of this structure. It is clear, however, that team leaders gained more independence and Miyamoto no longer took part in every decision.

We described EAD’s structure in more detail in a separate article. What matters here is why Nintendo decided that change was necessary. Over its entire life cycle, the GameCube sold 21.74 million units, compared with 32.93 million for the Nintendo 64. The old solutions were no longer enough. Nintendo needed new leaders and independent teams that could find their own paths instead of always looking back at Super Mario 64.

EAD Opens a Second Studio

In 2003, EAD created a separate group in Tokyo for the first time. The Kyoto division remained the center of development, but the Tokyo team worked independently and gave younger employees a chance to grow within a new studio. Takao Shimizu and Yoshiaki Koizumi became its leaders.

EAD Tokyo’s first game was Donkey Kong Jungle Beat. Released in Japan in December 2004, it used the DK Bongos controller: the player struck the left or right drum, clapped their hands, and made Donkey Kong move in time with each action.

Donkey Kong Jungle Beat, the first game from EAD’s Tokyo team. Donkey Kong’s movements were designed around the bongo controller.
Nintendo / 社長が訊く『スーパーマリオギャラクシー』Vol.1 プロデューサー・ディレクター 篇 / nintendo.co.jp

The team deliberately began with a compact project. It did not yet have the experience needed for a large game, and the DK Bongos required simple, predictable controls. Hit the right drum and Donkey Kong moved right. Clap and he grabbed an object. The result of every action was immediately visible on the screen.

After finishing Jungle Beat, the staff asked to work on a major Super Mario game. Nintendo’s agreement proved that the Tokyo studio was not considered secondary. The company wanted a second independent development center that could be trusted with Nintendo’s most important series.

But while the Tokyo studio prepared to move from the compact Jungle Beat to a major Super Mario game, EAD’s Kyoto teams were tackling a different problem: finding ways to make the Nintendo DS’s new capabilities understandable to people with any level of gaming experience.

Nintendo DS: First a Familiar Mario, Then Clouds

The Nintendo DS launched in Japan on December 2, 2004, for 15,000 yen. Its two screens, stylus, microphone, and wireless connection clearly set it apart from earlier handheld consoles, but buyers still had to be shown why those features mattered. The developers therefore began with something familiar: Super Mario 64 DS was EAD’s only game in the launch lineup.

The familiar castle and levels from 1996 remained, but players could now control Yoshi, Luigi, and Wario, while the lower screen offered mini-games and a virtual joystick. Super Mario 64 DS demonstrated the new console’s capabilities through a game buyers already knew.

The first Nintendo DS customers at a Bic Camera store in Tokyo on the morning of December 2, 2004. One of them bought Super Mario 64 DS with the console.
Minoru Funatsu, Kazuo Ishida, Kenji Saeki / 新型携帯ゲーム機「ニンテンドーDS」販売開始。都内各所で行列ができあがる / impress.co.jp

The next game could not have existed without a touch screen. In Yoshi Touch & Go, Yoshi ran forward automatically while the player drew cloud paths with the stylus, circled enemies, and changed the route in real time. Instead of a conventional platformer, the result was a short arcade game built around accuracy and speed.

In Yoshi Touch & Go, the player draws a cloud path between enemies. A line on the touch screen immediately becomes part of the level.
Naoto Yamamoto / ニンテンドーDSゲームレビュー「キャッチ! タッチ! ヨッシー!」 / impress.co.jp

In practice, the controls were not as self-explanatory as the developers had hoped. At the Touch! DS event in November 2004, director Hiroyuki Takemoto demonstrated the game to visitors and noticed a child who was around two years old. He expected the child to understand naturally that a line should be drawn across the screen, but the child had no idea what to do.

The team realized that a touch screen alone was not enough to make a game understandable to everyone. Drawing with a stylus was easier than memorizing button combinations, but the rules still had to be explained and the link between a gesture and its result had to be obvious.

Nintendogs Asks You to Carry the Console With You

The idea for Nintendogs appeared after Miyamoto got a dog. He was interested not in competitions or training for points, but in everyday interaction: giving the dog a name, calling it, petting it, and teaching it a simple command. The team first tried to create the game for the GameCube, but the console’s controller did not offer enough flexibility. The Nintendo DS stylus and microphone, however, made it possible to realize the idea in full.

The player touched a puppy with the stylus, and it reacted to the direction of the movement. Names and commands were spoken into the microphone. The stylus and microphone did not merely add features to Nintendogs—they supported the entire game. The developers later considered a Wii sequel, but abandoned the idea: the Wii Remote had no microphone, and petting a puppy by waving a controller in front of a television was not the same as touching it with a stylus.

A puppy in Nintendogs learns to recognize the name the player says into the Nintendo DS microphone.
Nintendo / 子犬たちとのふれあい / nintendo.co.jp

Another feature changed not the controls but the habit of using the console. In Bark Mode, a closed DS searched for another nearby system running Nintendogs. When it found one, the dogs exchanged gifts and messages. All the player had to do was carry the console around.

Even members of the team doubted that people would walk around with an active console just for a random encounter. But during Japan’s Golden Week, owners began reporting that exchanges happened almost everywhere: in shops, on trains, and in crowded neighborhoods. EAD realized that an accidental meeting between two handheld consoles could itself become a game mechanic.

Nintendogs was released in Japan in three versions on April 21, 2005. Its worldwide sales eventually reached 23.96 million copies. But EAD did not stop with one game for a new audience. In the same year, the studio released Big Brain Academy, Mario Kart DS, and Animal Crossing: Wild World.

These games were all structured differently, but each one suited a handheld console. Big Brain Academy offered short exercises, Mario Kart DS allowed players to compete online, and Wild World let them return to their town or visit friends at any time. Players could come back to all four games every day, even when they had only a few spare minutes.

New Super Mario Bros. Rejected the Touch-Screen Revolution

By 2006, touch controls had become the defining feature of the Nintendo DS. But they did not suit every game. The developers of New Super Mario Bros. tested whether Mario could be controlled comfortably with the touch screen, then rejected the idea almost entirely.

For precise running and jumping, the directional pad and buttons worked better. Nintendo’s official website explained the controls with a single image: the directional pad and two main buttons. The lower screen displayed the map and held a reserve item, but it did not interfere with Mario’s every movement.

The official New Super Mario Bros. control diagram for Nintendo DS. The main game once again relied on the directional pad and two buttons, while the touch screen was given a supporting role.
Nintendo / New スーパーマリオブラザーズ / nintendo.co.jp

The novelty was still there. Mega Mario smashed through bricks and pipes, Mini Mario ran across water and slipped through narrow passages, and the developers brought over moves from the three-dimensional games, including the wall jump. But Takashi Tezuka asked the team not to add ideas merely for the sake of quantity. The main goal was to preserve a familiar feeling: even after a failed jump, the player wanted to try again immediately.

New Super Mario Bros. was released in Japan on May 25, 2006. It was the first new two-dimensional game in fourteen years to feature Mario in the leading role. Worldwide sales reached 30.80 million copies.

For comparison, the best-selling GameCube game was Super Smash Bros. Melee, with 7.41 million copies sold. Each of the three Nintendo DS games—Nintendogs, Mario Kart DS, and New Super Mario Bros.—sold more than three times as many copies.

Epilogue: Not Every Feature Belongs in a Game

During the GameCube and Nintendo DS years, EAD’s approach changed. The developers no longer added new features simply because the console supported them. When a familiar solution worked better, they abandoned the experiment.

In Nintendogs, the stylus and microphone defined every interaction with the puppy. In New Super Mario Bros., the touch screen was limited to the map and reserve item because running and jumping worked better with buttons.

The Wii became the next test. Motion controls once again changed familiar ways of playing, and EAD now had to use them in Super Mario, Zelda, and Nintendo’s other major series.