Ever since the beginning, there have been people who look at ORBITRON products, study our patents, and attempt to create something similar based on them.
If you read a patent, you can understand a certain amount.
Maybe they are using this material.
Perhaps the configuration is like this.
With this logic, maybe I can make something close.
That much is possible.
However, when it comes to actually making the same thing, it is not that simple.
Reading old patents won't result in today's ORBITRON
ORBITRON holds patents obtained in the past.
However, we are not making current ORBITRON products exactly as described in those patents.
The raw materials have changed. The combinations of materials have changed. The processing methods have changed. The manufacturing processes have changed.
As we have continued production over a long period, the products themselves have evolved.
Therefore, even if you read an old patent and build something exactly according to it, you will not get today's ORBITRON.
Furthermore, our current recipes are not public. They are not on the internet. They are not in any academic papers. They are not information you can find by searching.
The same goes for our manufacturing processes.
Even if you knew all the raw materials, you wouldn't know the order in which they are used, how far to process them, what signs to look for to proceed to the next step, or what criteria are used to judge quality. That information remains unknown.
Manufacturing is not just about gathering materials.
It is just like cooking.
Even if you gather the same ingredients, the dish won't turn out the same. There is temperature. There is sequence. There is time. There is the touch of the cook. And ultimately, there is a part where a person judges by looking at it.
ORBITRON is the same.
I will search overseas if that's where the necessary materials are
Even finding the materials is not easy.
In my case, when I think, "I might need this," I don't just look within Japan.
If necessary, I will search overseas. Sometimes I procure them from very distant countries.
Moreover, it is not just materials that anyone can buy with a standard order.
There are materials that even corporations cannot purchase unless they pass a screening of their intended use and the company itself, and receive approval.
We go to such lengths to obtain them.
However, just because a material was hard to get doesn't mean it can necessarily be used.
I ordered it from overseas. It was expensive. It was hard to get.
Even so, if it doesn't yield the expected results, I don't use it.
In fact, there are many raw materials I've obtained that I eventually didn't use. There are materials I tested but didn't result in a product. There are also many technologies I didn't adopt.
There are things for which I even made prototypes only to conclude, "This isn't it."
It’s not just successful information that goes into a product
When looking only at the finished product, this part is invisible.
What appears on the surface is only the final answer that remained.
But behind that,
Not this material.
Not this combination either.
Increasing this makes it different.
Let's try reducing it.
Change the order.
Change the processing method.
Still different.
Start over again.
We repeat this over and over.
ORBITRON is not a product made from an answer that existed from the beginning.
From a large number of candidates, we keep cutting away things saying, "This is wrong, this is also wrong, this can't be used," and the things that remain at the end are further refined.
It is a technology born as if it were truly squeezed out.
And there is one more important thing here.
ORBITRON technology contains a vast amount of information not only on "what we use," but also on "what we didn't use."
What went wrong? What yielded no results? Which materials were discarded? Which technologies were not adopted?
There is value in this accumulation of failures.
What is close to the AI "distillation" metaphor?
This story is somewhat similar to the concept of "distillation" in AI.
Distillation in AI is not about copying the entire internal structure of a powerful AI.
You use the powerful AI as a teacher and collect a massive amount of output, such as "how it answers this question" or "how it judges under these conditions."
And then, you have another AI learn those ways of answering.
In other words, instead of redoing all the years of research, failure, and trial and error it took to create the original AI, you use the finished results as teacher data to try to create something with similar performance.
This is AI distillation.
If we apply this to ORBITRON, it is like us spending years looking for materials, ordering them from overseas, passing screenings to obtain special materials, testing them, throwing them away if they can't be used, thinking of other combinations, changing manufacturing methods, and testing again.
Looking at the finished product reached through that process, one might try to reverse-calculate, "If this result is produced, this part is probably important," to try to create something similar.
The concept is close to AI distillation.
However, in the case of ORBITRON, it is not as simple as AI.
In AI distillation, you can throw a huge number of questions at the teacher AI and observe the answers repeatedly.
On the other hand, for ORBITRON, which raw materials are used? Which materials were discarded? What was tested and failed? In what order is it processed? Where is quality judged?
We don't put that kind of information out.
Moreover, the "output" observable from the finished product is limited.
Therefore, one can reverse-calculate from the finished product.
But what we tested, what we discarded, and why it ended up in its current form—that much is not visible.
To use the AI distillation analogy, the teacher data obtainable from the teacher is overwhelmingly scarce.
That is why, even if one can make a similar shape, it is difficult to bring it to the same level.
ORBITRON has a development entry point that cannot be explained by theory alone
ORBITRON has aspects that are even harder to imitate.
It is my own way of product development.
This is strange even to me, but there are times when I suddenly think, "Wouldn't it be interesting to try this?"
It is not because I read it in some paper. It was not derived from theoretical formulas.
While looking at materials or while thinking about something else, I suddenly think, "What would happen if I combined this and that?"
So I search for it.
If it is not in Japan, I search overseas. If I get it, I test it. If it fails, I throw it away. I search again.
Not all of them work out. In fact, there are more things I didn't use.
But among them, something that says "This is something different" appears.
From there, I work with the factory manager to refine it: "This is interesting," "Then let's change this part," "Let's try it this way a bit more."
That is how it becomes a product.
There is quite a lot of such serendipity in ORBITRON's development.
It is not just a coincidence; it is more like the feeling that, through years of experience, things suddenly connect at a certain moment.
This does not come up in a search.
After all, I don't know it myself until right before I think of it.
Gelhorn and Gelman Power are difficult to replicate from the raw material stage
This story also applies to Gelhorn and Gelman Power.
These are even more special from the raw material stage.
The suppliers of the raw materials do not provide them to anyone other than specified parties.
In other words, the idea of "I'll just buy the same raw materials" does not work.
What is even more interesting is that even the raw material suppliers themselves cannot make the same final product.
This is because the technology to make the raw material and the technology to make the final product are different.
The raw material supplier can make the raw material.
But how to process that raw material further? What to combine it with? What processes to put it through? How far to process it?
They don't know that.
Conversely, even on the manufacturing side, you cannot make it if you cannot obtain the same raw materials.
Raw materials. Recipe. Manufacturing process.
If these three are not together, you cannot make the same thing.
Moreover, the information for each is not gathered and made public in one place.
That is why it is difficult to copy.
However, technologies that are difficult to copy also have weaknesses
This is a strength and at the same time a major weakness.
If something happens to the manufacturer, there is a possibility that the product can no longer be made.
Even if you have the raw materials left, you can't make it.
Even if you have the equipment left, you can't make it.
Even if you have the old patents left, you can't make it.
If the technology or judgment held by the manufacturer themselves is lost, in the worst case, the product itself will disappear from the world.
This is not an exaggeration.
In Japan's manufacturing industry, how to pass on skills as skilled workers retire and generations change has been a long-standing challenge.
Even at huge steelmaking sites like Nippon Steel, the succession of skilled expertise has been discussed as an important theme.
When you say steelmaking, it looks like machines are making steel by putting raw materials into a large furnace.
But in reality, it is not that simple.
The state of the raw materials. The state of the furnace. Temperature. The quirks of the equipment. The conditions of the day.
There are subtle differences that cannot be judged by numbers alone.
A craftsman who has looked at the site for decades notices, "Today it's a little different."
And makes adjustments.
That judgment is not all written in a manual.
It is inside the person.
Therefore, when the person is gone, the equipment is there. The raw materials are there. The drawings are there. Even so, there are times when it can no longer be made stably as before.
I think that ORBITRON, Gelhorn, and Gelman Power are the same at their root.
Technology does not remain through patents alone
Technology is not just about patents.
It is not just about drawings. Not just raw materials. Not just equipment. Not just the successful recipe.
What was tested? What was discarded? Where did the failures occur? Why did we arrive at the current materials? How is it made? Who makes the final judgment and how?
That, too, is part of the technology.
Therefore, I do not think that "we are protected because we have patents."
What truly has value is the current materials. Current recipe. Current manufacturing process. Routes for procurement of raw materials including those overseas. The vast trial and error of failures. The judgment accumulated over many years. The senses that can only be understood on-site. And the person who can reproduce it.
Looking from the outside, it might be possible to make something similar.
You can read patents. You can analyze products. You can search for similar materials.
But you cannot see the path we took to arrive there.
In fact, I have heard many times from people who bought various similar products related to reactors that they "ended up coming back to ORBITRON."
Some even evaluate them as different when compared to old ORBITRON products.
For the purchaser, it is a detour.
Of course, there is the price, so I understand that there are people who think ORBITRON is expensive and want to try something cheap first.
That is that person's judgment.
However, to be honest, I think there are cases where choosing ORBITRON from the beginning would not have resulted in a detour.
That is what I think.
Hide to protect. Convey to leave behind
On the other hand, there is another challenge for the side that makes it.
How to leave this technology behind.
If you only think about not being copied, it is better not to release information.
But if you convey it to no one, it ends the moment there is no one left who can make it.
Hide to protect.
Convey to leave behind.
This is contradictory.
Therefore, what I really have to think about from now on is protecting the technology. Evolving the technology. And leaving the technology for the next generation.
I believe these are the three things.
Products will not remain through drawings alone.
They will not remain through patents alone.
In the end, it will not remain unless it is passed from person to person.
Having made ORBITRON for a long time, I feel that most strongly now.