Sailor creator Yoot Saito on the IA Dreamcast ahead of his time

Sega's Dreamcast was ultimately a failure, Sony dominated the early 2000s market with PlayStation 2. But Sega's machine left behind a particularly innovative and influential software library. And perhaps no title has been as memorable as Yutaka's iconoclast "Yoot" Saito Marine, a virtual pet simulator in which you used a microphone to converse with a sullen and sarcastic fish-man, with the help of a narrator voiced by Leonard Nimoy.

Yoot Saito was previously best known as a Mac-based developer who created Tower, a tower management game that SimCity Maxis studio published worldwide as SimTower. (His suite, Tower II, came to be known as Yoot Tower out of Japan.) After Marine and its PS2 suite, Saito designed Odama, a pinball strategy hybrid for the GameCube where players used voice commands to direct soldiers to the table.

This summer marks 20 years since MarinePress release, so I contacted Saito by email to discuss his development and his legacy. His answers were so thorough, with different passionate tangencies in topics such as the history of video games and linguistics, that I decided to reproduce them here with minimal editing.

This interview has been slightly modified for clarity.

About how Marine has come to be after SimTower

"If you just separate the themes, one concerns the game of building skyscrapers and the other, raising a fish with a human face. They have a completely different aspect. [But] if you trace my starting point in the game industry, it's because the masterpiece that is SimCity had a profound effect on me and led me to want to play games. So, my main interest at the time and even now is that of simulation games.

If I had to be even more specific, I would say that I have a huge interest in communication and language. For me, the most basic aspect of a computer game is to create a new world of game designer via the 'language' of the game. To do this you need to remove unnecessary items from the real world and spread what you choose to keep as key features of the game or elements with which you can interact as elements or via game controls. This is the backbone of simulation games.

So at this point, Marine is a game where the challenge was to use the language filter to bridge the gap between players and non-players. Trying to turn the current language into a language element / filter for games is a big challenge, but it's actually like converting natural language expressions into playing cards and playing a deck of cards with another person .

That means Marine is actually like playing poker. The keywords are the "cards" that are played between the end user and the game character (Seaman). Most players would not realize it, but Marine is a game with a lot of push-pull game elements in the same way as poker. This back and forth is a driving mechanism of the game. SimTower when you add a security room in the skyscraper, Marine a different keyword leads to an entirely new conversation path, and perhaps it is the easiest way to convey that concept.

I have conceived the idea of Marine at lunch one day. I spoke with a developer working with me on a new Tower series game. We had several projects underway, and one of the teams was working on a simulation game in which you were raising tropical fish in an aquarium. We laughed, but I said, "If I had to do a game like this, I'd do something crazy!", To which my colleague replied, "Yeah, like what, in particular?" I felt like I was on the spot and had to say something quickly.

"I would make the camera face the user and the animal speak in human terms, something like," Who was this woman that you brought here yesterday? "An animal like that would be "A different colleague said," That would be really strange, "which I replied:" Yes, strange and awesome! "Of course, we could not call Sea Monkey … we must call it Marine! And that went on to be the code name of the project and the final name of the actual game.

Everyone laughed at the ridiculous nature of the idea, but this weekend I told my wife about it and she found it strange and awesome. So it was a universally strange concept for men and women. She then added, "It's a really strange and disappointing concept, so you absolutely have to do it!" It was then that I realized that even women could enjoy and enjoy a strange and disgusting game. These words helped my decision to play the game seriously. "

On the influence of the virtual pets of the 90s and Shigeru Miyamoto

"When you say that virtual pets were popular in the '90s, I guess you mean Tamagotchi, but this trend did not really interest me. As I have already mentioned, I like complex simulation games. The more complex the theme, the more no other creator has done, the more I'm interested. I did not want to follow the trend of someone else.

Most game creators find a winning recipe in a popular game and give the chance to some elements or design elements, to a story, etc., and then call it creating a game. But I do not understand this state of mind. Do not focus on sequels and want to do something really strange and new, or you could say, something that will probably yield nothing, is quite rare. I only met another person who could understand my basic feeling, it's Shigeru Miyamoto from Nintendo.

Miyamoto-san really likes strange and different concepts. So during Marine was released on Dreamcast, behind the scenes, we almost chose to develop the game on the Nintendo 64DD. In fact, around the time Marine Dreamcast, Miyamoto-san, was published in a magazine with a phone number. Marine shirt. Marine was finally released by Sega, but Miyamoto-san was an extremely important person in the process of obtaining Marine done, especially for someone like me with very little console gaming experience.

In addition, although Sony's robot dog, Aibo, came out at about the same time as Marineno product was influenced by the other. Coincidentally, I had conversations about a biped robot from several companies. Although 20 years have passed since Marine given the light of day, I make an artificial intelligence engine that can speak without the need for pre-existing scripts or text strings. The road was long to get here. Inventing it has taken a lot of time and has required a lot of different investments, but if you ask me, it's more interesting to join leagues than to develop a normal and certainly more exciting game. "

How Marine came to be on the Dreamcast

"In fact, a man named Irimajiri-san, who was Sega's vice president at the time and then became president, was the one who drove Marine to be published on the Dreamcast. Kenji Eno-san, another creator, told me that he had someone that he really wanted me to meet and introduced me to Irimajiri-san.

At the time, the Dreamcast was still referred to by its code name Katana and was a new type of gaming console. Irimajiri-san was previously president of Honda in the United States and felt that Sega needed to focus on new ones. types of players. It was the time when Sony was king and so I decided that it would be interesting to be part of the market leader from a lower position and join Sega. Transforming a prototype built on Mac into new consumer hardware was quite difficult, but we were able to do it in about a year and a half. It was thanks to the hard work of the team at the time that we were able to achieve this goal. "

On the Dreamcast material

"The Dreamcast microphone was developed by the Sega device team. I remember that they already had a prototype in use. However, bringing a game developed on PC to the console posed a lot of problems. Games developed for PCs were focused on the use of a mouse and a keyboard. Basically, you take your time to play most PC-oriented games. You play a few games / turns / turns on the keyboard and mouse, then go on to work on these same input devices. These were games you could just spend time on, but did not need all your attention.

Conversely, the console games require you to sit in front of the TV, fully focused on the game. You have entered the game controller powered by electricity and you basically said, "It's time to play!" And the work was not going to interrupt him. You quickly pressed several buttons, made different movements and you focused on this experience. For the console industry, they wanted gaming experiences that focused on interactivity and concentration. Since our game was developed on PC, our main focus and experience did not match the target console platform. We needed to develop new ways to interact with the game, which required more activities (button presses, etc.) from the end user and elements to constantly focus on the game while playing, which resulted in major changes in the design.

On location Marine in English

"I asked Sega of America (SOA) to handle almost all of the localization of the English version of Marine. The reason is that there is a flagrant cultural difference between different countries. When a person in Japan asks you what your blood type is, it comes down to asking someone what is his zodiacal sign. So the translation would not have been enough on its own to develop an English version of Marine. It was almost necessary to rewrite the script for it to work. SOA knew a very skilled third party to rewrite scripts. So they focused on what allowed our team to build the system and the basic game without limitation. The proposal I had was to use Leonard Nimoy as the guiding voice of the user.

Japanese is a very nuanced language, complex and based on the feeling. It is a language that allows you to share subjective feelings with emphasis. The Japanese really connects to the deeper emotions of the speaker. Since I do not speak English, I can not comment on the deep ties one feels when speaking English, but I can say when I read the game comparison between Ecco the dolphin and Marine in Rolling stone magazine, it was clear to me that the editors did a great job of understanding the basic concept of the game. "

On the state of the art voice recognition technology at the time

"The voice recognition engine we used with Sega in Marine was dismissed by a Belgian company. We used their voice recognition engines as users and not as creators. The size of the memory being limited at this time, the performances of the speech recognition were completely different from the current ones. Specifically, we used a function called navigation. Currently, the common function is called dictation. When you create a type of navigation conversation, it must predict in advance what the user is likely to say and choose it. If the user pronounces a word that is not an option, an error will occur. Therefore, the writer must indicate in advance the type of words that the user will say and engage in a conversation that leads to one of them. For this reason, you have had a structure in which conversations are conducted by exploring intentions. It was an interesting experience that had not been done in a game before. "

On the design of the game around technical limitations

"At that time, there were no other games apart Marine who have had voice conversations, and not many, even now, so it's hard to compare. But voice recognition technology itself is changing rapidly right now – if a smart game designer was working on something, I think it could create a really interesting game. However, when you speak of speech recognition as a technology, when you try to cross a certain threshold (that of the human being), you must have an intelligence and not a hearing ability . In other words, what is essential is not speech recognition technology, but the understanding of artificial intelligence. The speed of progression of AI is very fast right now, so I think it's a fun time.

To come back to the subject, I was able to finish the development of Marine specifically because of the limit in technology. If there was not one, I would never have finished it. As I said at the moment, when we propose a game designed on a PC to the consumer, there is a big difference in the environment called time density. How could one overcome what has tested my wisdom, one might say.

Sure MarineThe personality

"The reason MarineThe personality and the background are so distinctive because they contain my philosophical idea and not a specific message. At the beginning of Marine project, I firmly decided that it would be the opposite of classic games in three respects:

• It would not be cute
• He would look at the world of players from inside the TV
• The theme of the conversation would be everyday things, not an imaginary world

I was convinced that if I stayed true to these three aspects, the game would be completely different from any other game – something different, not often seen. That was my strategy and what I'm still trying to do, even now. "

On the reaction to Marine

"The answer of MarineThe release in Japan was much larger than that in the United States. It has become a social phenomenon, such as comedians appearing in Japanese television shows imitating and wearing Marine costumes, or the news saying, "The games have evolved so much!"

To return to my personal philosophy, when a Japanese creator creates for Japanese users a world that has a considerable impact on Japanese, the social phenomenon is that it attracts many media and society. In the United States, the phenomenon looked like this: "A relatively rare game is out. It's for the new Sega material that they pushed into the console war. So I think recognition is very different.

After 20 years, I see in Japan TV artists who are currently talking about Marine. They said it was because it had a big impact on them when they were kids. They joked and said:Marine traumatized in my early childhood. 'As a creator, it's a great honor.'

Sure SeaMail, the derivative of the Windows desktop

"SeaMail is a kind of prototype on a more versatile PC platform that exchanges words a little more complicated. Due to budget constraints, the work did not go very far, but it was a very ambitious attempt. So Marineunlike the user, lives in the world of the computer of a versatile PC and examines the contents of the user's emails, the frequency of interaction with specific people and comments on the user's relationship with others. It looked a lot like the current Siri. This knowledge exists in the artificial conversation engine I am currently working on. "

Using the microphone in Odama

"The purpose of the microphone in Odama is completely different from Marine. Odama is a game that simulates a fight of soldiers. I was interested in this so-called "samurai crowd". The reason for this microphone was that the two hands of the user were occupied by the pinball machine. The voice was used only to give orders. I was also motivated to use it because I wanted to recreate the feeling of a general who was giving orders to a scrum of soldiers participating in feudal battles. I wanted it to be like a football coach who gives orders like "push" and "right". That's all.

For me, Odama was the greatest masterpiece of my life, but the launch was delayed again and again. I think I really pissed off Nintendo. At the time, programmers had no idea of ​​crowd control thanks to autonomous robots or artificial intelligence. Development has been slowed down.

On the current IA and technical assistant of the voice

"Artificial intelligence is at a very advanced stage right now, but the Japanese conversation method has not changed in 20 years. Marine It was a game where the planner wrote pre-existing scenarios and where people enjoyed their game.

An infinite conversation was impossible. So, I am no longer interested in this project. Scenarios can only be increased two or four times over the previous game. That's why I launched the Artificial Intelligence Lab to create an engine that keeps talking even without a script. If we can complete this engine, we will start developing a 'Marine The new thing will be the sailor's voice generated and spoken on site in real time, on the spot, not a prerecorded voice. This requires a lot of inventiveness, but for now, we are just beginning to see the light at the end of the tunnel.

Japanese is a very special language in that it seems to have a solid grammar, but it is not. In fact, it is not fair to say that this is not the case. The Japanese grammar we learned in textbooks is far removed from the Japanese we use in conversations, which is not a true grammar. This divergence and contradiction have been ignored for decades. There was no need to do anything so expensive, like affixing a new grammar. In Japanese, there are two or three subjects. However, the SVOC style proposed by the United States after the war was used. Linguists have long argued in this direction, but none of them, who insist on writing memoirs, has created a motor for AI and tried to prove that his hypothesis is correct.

And now it is the era of conversation engines. It may not be necessary to redefine grammar for schools, but it is necessary to redefine the Japanese logically for a robot to speak. That's what we're working on.

It's difficult. Trial and error, about 10 years. When I changed the grammar into two or three subjects through scanning and symbols, I started to succeed. My experience in the development of Marine and with the nuance of playing the voice made it possible – especially the fact that the meaning of a word changes depending on how the melody is applied. We have noticed that Japanese grammar resides in melodies and not in letters. It was not easy at the age of paper and pen. I had overlooked this as a nuance, whereas in fact, it was a pillar of Japanese grammar and the voice actor of the Marine series, I became aware of this notion and decided to develop a melodic recognition engine. "

On current projects

"Right now, I'm creating a conversation engine that keeps talking even if there is no scenario. To this end, we have created an organization called "Marine Artificial Intelligence Lab". When this engine is finished, I think voice recognition in conversations will change dramatically.

I think that a talking robot will be marketed in Japan in about six months. Once the ball starts to roll, I would like to license this engine to many machines and create an era where microwave ovens, cars, cameras and robots made in Japan speak fluently. We look forward to seeing this happen and are looking for people and investors to help us in English. A website in English is still in preparation, but you can contact the site: www.seaman.ai. "