The Last Engineers: A Record of Vanishing IT Traditions (1)

Episode 1: Fuzzy Control

Is 'fuzzy' a dead term? Or will it come back to life in the age of AI?

This article is a documentary-style creative column that begins with a reader's question. While not based on an actual program interview, it is structured by layering the atmosphere that once existed in Japanese technical workplaces with questions about the current AI era.

The first installment began with a question from a note reader.

I remembered that 'Fuzzy [Something]' used to be popular. Is it a dead term now? Or will it come back to life in the Reiwa era?

The question was submitted by ekda.

Indeed, there was once a word in Japan called 'fuzzy'.

Fuzzy rice cookers.
Fuzzy washing machines.
Fuzzy air conditioners.
Fuzzy vacuum cleaners.

At the beginning of the Heisei era, this word was everywhere in home appliance stores.

It was not just a marketing slogan.
It was the watchword of an era when machines were trying to understand human 'vague' intuition.

Now, we live in the age of AI.

Artificial intelligence writes text.
It creates images.
It synthesizes audio.
It summarizes meetings.
It replies in human-like language.

However, this is where ekda's question quietly takes effect.

Was 'fuzzy' just a dead term?
Or has it returned in a different form in the Reiwa AI era?

Using this question as a starting point, this edition of 'The Last Engineers' traces the memories of the fuzzy control that once supported Japanese home appliances.

Program Promotion
The Last Engineers

It is said that artificial intelligence will change human work.

Writing code.
Writing text.
Drawing pictures.
Creating voices.
Supporting human judgment.
Predicting the future.

However, at the foundation of that glamorous future, something else is happening.

Technologies that once supported Japanese industry.
Old control systems that have moved the background of society.
Field intuition that cannot be conveyed by specifications alone.
The engineers who understand these things are quietly beginning to leave the workplace.

There were those who read the breath of machines in assembly language.
There were those who protected the bloodstream of financial systems with COBOL.
There were those who saw a future in TRON where every device would be connected.
There were those who wrestled with the quirks of the Japanese language inside dedicated word processors.
There were those who rescued garbled names from the gaps between character codes.
There were those who protected data while rewinding magnetic tapes in an era before the word 'cloud' even existed.

Those technologies were once the stars of their time.

However, society changes.

The bursting of the bubble economy.
The internet revolution.
The Lehman Shock.
The spread of smartphones.
The COVID-19 pandemic.
And the arrival of generative AI.

Every time a new technology gained attention, old technologies were relegated to the past.

But have they truly become old?

Society still runs on systems that someone built long ago.
Even convenient apps and massive AI are supported somewhere by old code, old standards, and old design philosophies.

In this series, starting from questions sent in by readers like 'Whatever happened to that?', we will visit vanishing IT technologies, forgotten design philosophies, and the old systems that modern society still depends on.

The first episode is fuzzy logic.

It is a story from an era when Japanese home appliances tried to understand the human 'somehow'.

Episode 1: Fuzzy Logic

Once upon a time, machines tried to understand 'somehow'.

From the end of the Showa era to the beginning of the Heisei era, a certain word was lined up in Japanese home appliance stores.

Fuzzy.

It means to be ambiguous.

One or zero.
Correct or incorrect.
Hot or cold.
Dirty or clean.

Computers have excelled at such clear-cut judgments.

However, human life is not that simple.

A little hot.
A little dirty.
A bit too much water.
Steam it a little more.
Lower the heat a little.
Wait and see.

Humans naturally make such ambiguous judgments in their daily lives.

Fuzzy control was a technology that attempted to teach machines how to handle that sense of 'vaguely'.

The place I am visiting this time is a small town in the northern Kanto region.

Passing through the shopping street in front of the station, there is an old residential area beyond the rice fields. In one corner of it lives a former engineer who was once involved in control design at a major home appliance manufacturer.

Mamoru Kodai, seventy-two years old.

In his workshop, old measuring instruments, yellowed design notebooks, and control boards from rice cookers of the early Heisei era remain.

As he traced the board with his finger, Mr. Kodai smiled a little shyly.

'Well, back in the day, if you said 'fuzzy,' it was a bit of a star attraction. There was a time when customers would stop in their tracks just because a rice cooker had fuzzy logic in it.'

In those words, the pride of a former engineer remained.

At that time, Japanese home appliance manufacturers boasted of their strength.

More convenient.
Smarter.
Kinder to people.

Home appliances were spoken of not just as tools, but as entities that would change lives.

Rice cookers read the amount of rice and the temperature of the water.
Washing machines read the degree of dirt and the amount of clothing.
Air conditioners read the room temperature, the outside temperature, and human movement.

Unlike today's AI, they did not collect vast amounts of data.
They were not connected to the cloud.
They did not send the user's life to a server.

Within a small circuit board, they read limited information and made judgments with limited processing power.

'It wasn't like today, where you can use brute force with a massive computer. So, we were always thinking about how to pass on the intuition that humans use in their daily lives to the machine.'

Mr. Kodai opens an old design notebook.

In it, handwritten tables are lined up.

Temperature is slightly high.
Moisture is a little high.
Reduce heating slightly.
Extend steaming a little.

'A little,' 'slightly,' 'quite a bit.'

These are words that machines should not be able to understand.

Fuzzy logic transformed those ambiguous words into mathematical formulas and rules.

It wasn't about whether it was hot or cold, but how hot it was.
It wasn't about whether it was dirty or clean, but how dirty it was.
It wasn't about whether there was a lot of water or a little, but how to handle the middle ground.

You don't cut boundaries with a line.
You connect them smoothly.

That was the core philosophy of fuzzy logic.

"It's not about black or white. Human life is mostly gray. How we cook rice, how we do laundry, whether it's hot or cold—everything is decided within that gray area."

Mr. Kodai said this as he opened the lid of an old rice cooker.

"Rice isn't the same every time, either. There's new rice, and there's old rice. The water temperature varies. Every household has different preferences. If you try to handle all that with just rigid formulas, you end up moving further away from the human experience."

In that era, fuzzy logic was the technology of the future.

Brightly colored brochures lined the shelves of appliance stores.
They depicted a dream of machines that were becoming more human-like.

It wasn't about machines managing humans.
It was about machines drawing closer to humans.

There was an atmosphere of that kind in the air at the time.

However, that tailwind did not last long.

As the Heisei era began, the bubble burst.

The era of competing for high functionality shifted to an era of cutting costs.
The era of making good products even if they were a bit expensive shifted to an era of making things cheaply, quickly, and in large quantities.

The language required within companies also changed.

Instead of "How can we make the rice taste even better?" it became,
"Can't we remove one of these parts?"

Instead of "How far can we adapt to the person?" it became,
"How much can we shorten the development period?"

Mr. Kodai narrows his eyes slightly as he speaks.

"It wasn't the technology that changed. It was the look in the company's eyes. Before, they asked me how to cook the rice well. After the collapse, they started asking if I couldn't just remove one of these parts."

Eventually, the era of the internet arrived.

The world's interest shifted from small controls inside home appliances to personal computers, networks, and software.

Technology that changes flashily on screens began to receive more attention than technology that operates quietly within the home.

The word 'fuzzy' gradually disappeared from retail shelves.

In its place, terms like broadband, digital, smart, and cloud began to line up.

Even so, engineers remained on the front lines.

People who watched the rice finish cooking. People who listened to the rotation sound of washing machine tubs. People who checked the heat of parts with their fingertips. People who noticed subtle differences in prototypes.

It has become numbers. But, you cannot understand it with numbers alone.

They were people who knew that kind of world.

However, after the Lehman Shock, a great wave surged into those front lines as well.

Research departments were downsized. The number of prototypes was reduced. Veteran engineers disappeared one by one through early retirement or reassignment.

What remained were specifications, circuit diagrams, and design notebooks.

'The papers are still here. The specifications, the circuit diagrams, and the charts, too.'

Mr. Kodai takes a yellowed file out from the back of the shelf.

'But you see, phrases written here like 'a little stronger' or 'observe the situation' are just words if there's no one left who can read them.'

Specifications preserve technology.

However, there are things that cannot be preserved by specifications alone.

Which sound is normal. Which smell is dangerous. Which temperature change is a quirk of the part. Which level of cooked rice is 'delicious' beyond what the numbers say.

Such things have been passed down on the front lines.

However, the COVID-19 pandemic arrived.

Study groups where old engineers used to gather were interrupted. Factory tours were restricted. In-person handovers decreased. In meetings through screens, sounds, smells, and textures cannot be conveyed.

Fuzzy logic remained in academic papers and documents.
However, the intuition for applying it to everyday appliances quietly faded away.

And now.

The world is talking about ambiguity once again.

AI.
Large language models.
Human-like responses.
Machines that read emotions.
Systems that read the room.

The questions that fuzzy logic once faced have returned under different names.

How should machines handle human ambiguity?

However, the direction of the era has changed significantly.

Machines of the past were trying to move closer to humans.
Today's AI is trying to absorb humans as data.

Rice cookers of the past were quietly cooking rice in the home.
Today's AI consumes vast amounts of electricity in data centers around the world while returning human-like words.

Mr. Kodai looked at an old rice cooker and said softly,

"I don't know. Today's AI sure talks a lot. But whether it can cook rice well is a different story."

That statement sounds like a joke, but it is not.

AI writes text.
It draws images.
It creates music.
It offers advice.
It even seems to predict the future.

But what does it really mean to understand human life?

Observing temperature.
Observing humidity.
Observing texture.
Listening to sound.
Sensing scent.
Adapting to that day, that home, and that person.

Fuzzy logic was addressing those questions within household appliances.

It was not a flashy technology.

It did not become a massive investment opportunity.
It did not incur cloud usage fees.
It did not hoard user behavioral data.

It simply cooked rice in the corner of the kitchen.

Yet, inside that small machine, there was a philosophy that did not look down on human ambiguity.

Mr. Kodai says.

"Fuzzy logic wasn't so much a technology to make machines smarter, but rather a technology that didn't look down on human ambiguity."

With the collapse of the bubble economy, the place where value was placed changed.
In the age of the internet, it was robbed of its leading role.
With the Lehman Shock, the front lines were cut back.
With the COVID-19 pandemic, the places for succession were severed.
And in the AI era, that very question has returned under a different name.

Fuzzy control.

It is not a technology that has disappeared.

It is a technology that the times have pushed into an invisible place.

Things that cannot be divided into black or white.
Things that cannot be expressed in ones or zeros.
Things that exist between correct and incorrect.

That is where human life resides.

Once upon a time, Japanese engineers were trying to pass that ambiguity on to machines.

It was a challenge to create human-like machines, long before it was called AI.

Mr. Kodai said, while putting old circuit boards into a box.

"Fuzzy logic didn't end. Its name just disappeared. As long as humans continue to live with ambiguity, that kind of thinking remains somewhere."

Technologies sometimes disappear as buzzwords.

However, the questions do not disappear.

How should machines handle human ambiguity?

That question is still not over.

In fact, now that we are in the AI era, it is something that should be asked once again.

Episode 1: Fuzzy Control.

This is a story from an era when Japanese home appliances were still listening to the human 'vague feeling'.

In this series, we are accepting questions from our readers.

"Come to think of it, what ever happened to that technology?"
"Things that were once commonplace but have recently disappeared"
"Looking back now, perhaps that was the future"

If you have any such memories, please let us know in the comments section.

Next time: Assembler.

We will visit the engineers who listened to the heartbeat of computers from the place closest to machine language.

Text by Rintaro Takechi