Chapter 124 – The Duel Operating System

The days following Alexander's acquisition of the Duel AI Knowledge proved to be surprisingly fulfilling.

His normal routine hadn't changed much. He still attended school, spent time with Ginny, handled his responsibilities at Industrial Illusions, trained his body and Spirit Energy, and participated in the Duel Monsters club whenever he had the opportunity.

There was simply one additional responsibility now occupying a significant portion of his free time: organising the knowledge he had acquired.

The YGO System had given Alexander an enormous amount of understanding regarding artificial intelligence, machine learning, recognition systems, game-state analysis, predictive algorithms, and countless other subjects. While the knowledge itself had been implanted directly into his mind, that didn't mean it was automatically arranged into something that another person could simply read.

Alexander therefore spent much of his spare time compiling everything.

He had to take the interconnected understanding inside his mind and translate it into diagrams, explanations, technical documents, and development proposals that the R&D team could actually use, and it was a surprisingly satisfying process.

At the same time, Industrial Illusions was becoming increasingly busy.

The company's expansion into Canada was progressing well, with distributors and retailers continuing to establish the groundwork necessary to introduce Duel Monsters to the Canadian market.

That alone would have been enough to keep the company busy; unfortunately, it wasn't the only thing happening as the Duel Monsters Cup circuit was also being organised.

With the Grand Prix having demonstrated just how much interest there was in competitive Duel Monsters, Industrial Illusions was now preparing tournaments throughout the country. Competitions would be held across the various states during the year, giving Duelists opportunities to compete for qualification toward the next Duel Monsters Grand Prix.

Moreover, interest in the game was surging once again as the Grand Prix had attracted millions of viewers, and the tournament had apparently convinced a significant number of people who had previously treated Duel Monsters as nothing more than a children's card game to take the competitive side more seriously.

Industrial Illusions now had to keep up with that interest.

On top of all that, the Inter-School Duel Monsters Championship organised by Emma Frost's Frost International was approaching, and while Industrial Illusions wasn't actually organising the tournament, the company still had plenty of reasons to pay attention to it and ensure that it succeeded.

Duel Monsters was growing rapidly, and every major tournament was another opportunity for the game to reach new players.

Needless to say, Alexander's weekdays were becoming increasingly packed; His weekends therefore became even more valuable and, whenever possible, he used them to simply relax.

Sometimes that meant spending time with Ginny, Chuck, and Julia as a group; other times, it meant going on a date with Ginny.

Their relationship was continuing to develop naturally, and neither of them seemed particularly interested in rushing things.

They were still discovering what it meant to be in a relationship, and for the moment, they were perfectly content with kissing and making out over their clothes without pushing things further. There was no pressure and no expectation that they needed to reach some particular milestone. They were simply enjoying being together.

For Alexander, that was more than enough and perhaps because everything else in his life seemed to be moving faster and faster, he appreciated those quiet moments more than ever.

~~~▽~~~

The day after Alexander finally finished organising the Duel AI Knowledge, he attended another meeting with the R&D team.

The meeting began much like the previous ones, with Phineas reviewing the current state of the Battle Box project while the engineers listened and occasionally added their own observations.

The physical construction was progressing reasonably well. The hardware wasn't finished, but the team had continued solving problems one at a time. The software, however, remained the biggest obstacle.

Naomi explained the situation in greater detail.

The basic control systems weren't particularly difficult; the problem was everything that came afterwards, as the Battle Box needed to recognise cards; it needed to know where those cards were located, it needed to understand their orientation and movement, it needed to maintain the current state of the duel, it needed to understand card effects, it needed to enforce the rules, and eventually, it would have to use all of that information to control an artificial intelligence capable of actually following a complex game of Duel Monsters.

Each individual problem was manageable, but the problem was trying to make all of them work together.

Alexander listened quietly for several minutes, then he finally interrupted, "I think we're looking at this the wrong way."

The engineers looked toward him while Alexander leaned forward slightly and added, "During the last meeting, we talked about how complicated the Duel AI was going to be. I spent some time thinking about that."

He reached for the folder he had brought with him and stated, "I think the problem is that we're treating the Duel AI as one enormous software project."

Alexander placed several documents onto the table while continuing with his explanation, "Every component depends on information from another component. The recognition system needs to communicate with the game-state system. The game-state system needs the rules engine. The rules engine needs card information. The AI needs all of them. So every time we improve one part, we potentially create another problem somewhere else."

He paused for a moment and pushed the documents toward them before concluding, "I think we should separate them."

The first page contained a large diagram, and at the top were several boxes arranged in layers. Across the top of the page, Alexander had written: Duel Operating System.

Naomi immediately picked up the document while Alexander began explaining, "The lowest layer communicates directly with the physical hardware."

He pointed toward the bottom of the diagram. "That handles the sensors, projectors, and everything else the software needs to communicate with."

His finger moved upward. "Above that is recognition. This layer would identify cards, determine their positions, track their orientations, and monitor their movements. The important thing is that the recognition system doesn't need to understand the duel, but it simply needs to tell the rest of the software what it is seeing."

Alexander pointed to the next layer, "This is the game-state engine."

That system would maintain a constantly updated representation of the duel.

Life Points. Cards on the field. Cards in the Graveyard. Turn order. Phases. Active effects. And everything else necessary to describe what was currently happening.

"Above that comes another layer: The rules and effect engine," Alexander emphasised the distinction. "This system doesn't make decisions. It determines what is legally possible. It will process card effects, determine their consequences, manage chains, and ensure that the actions being attempted are legal within the current state of the game."

Above that sat the Duel Controller. Its responsibility was to actually manage the progression of the duel.

And finally, at the top, was the AI.

"The AI doesn't need to understand everything about Duel Monsters itself," Alexander explained. "It just needs to receive an accurate representation of the duel and decide what it wants to do."

Naomi's eyes widened slightly while Alexander looked around the table and pointed toward the entire diagram. "That's the important distinction. We don't need to create an artificial intelligence that understands everything about Duel Monsters. We need to create a system that understands Duel Monsters and then give an AI access to that system."

For several seconds, nobody said anything, then Naomi looked back down at the diagram and muttered, "That would mean..."

She began tracing connections between the different modules before she continued, "If the recognition system changes, the game-state engine doesn't necessarily have to change."

"Exactly."

"And if we change how card effects are processed..."

"The AI doesn't need to be rewritten."

Naomi's expression became increasingly interested as she realised, "Because it only receives the resulting game state."

Alexander nodded, "Right."

The next moment, Naomi immediately began asking questions.

How exactly should the modules communicate?

What information should each module expose?

How should simultaneous effects be represented?

How should chains be passed between the game-state and rules engines?

What happened when the recognition system wasn't completely certain about the identity or position of a card?

Alexander answered most of her questions confidently.

However, there were several occasions when he admitted that certain implementation details would need to be tested experimentally, but that didn't seem to disappoint Naomi; if anything, she became more interested.

The other engineers soon began examining Alexander's proposal from their own specialities.

Victor was the first to raise a different concern as he pointed toward the connection between the hardware and the software, "The amount of information being transferred could become a problem."

He was talking about latency because if the system had to constantly communicate with sensors and process every piece of information immediately, even a small delay could create problems during a duel.

Alexander nodded and said, "I've been thinking about that too."

He pointed toward another section of the document. "We should distinguish between information that has to be processed immediately and information that doesn't."

Some information would require real-time processing; other information could be handled asynchronously. That would prevent the entire system from becoming dependent on the slowest process.

Victor seemed satisfied with the answer and began making notes.

Elias had another concern. He focused on the interface between the physical Battle Box and the software. "The architecture assumes the sensors are reliable."

He gestured toward the prototype diagrams before he declared, "Our current system isn't."

Alexander looked at the diagrams and said, "That's fine."

Elias raised an eyebrow and asked, "Fine?"

"The software shouldn't assume the hardware is perfect." Alexander tapped the game-state section. "If the sensor data is incomplete or contradictory, the system shouldn't blindly act on it. It should flag the state as uncertain and determine whether more information is needed."

That prompted another round of discussion.

Priya eventually raised the question of card recognition, "Are you expecting the system to visually recognise every card through the Duel Monster microchip? Because we are nowhere close enough to actually complete them."

Alexander shook his head. "Not for the first prototype, but we can make the problem easier..."

That answer seemed to surprise her.

He explained that the cards used during the early testing phase could include machine-readable markings or another controlled identification method; that way, the system didn't have to immediately solve the much harder problem of visually recognising every card through the use of the microchips. Because of that, they could solve recognition in stages.

The answer earned Alexander another approving look from several members of the team because he wasn't trying to force the most advanced possible technology into the first prototype; he was trying to get a prototype working.

Amelia eventually asked the question everyone had been waiting for, "How much does this actually change our development schedule?"

Alexander hesitated as he didn't want to give them a number he couldn't justify.

"It doesn't eliminate the hardware problems," he began, "And it doesn't give us a finished AI. We still have to build everything, and what this changes is how we build it... So maybe it is better if Phineas is the one who makes this judgment."

He looked toward Phineas and handed him another document that contained a revised development schedule for the software.

The original plan had treated much of the software development as a largely sequential process, but Alexander's proposal broke it into progressively functional stages just like they did for the Battle Box development plan.

Hardware communication and basic projection control. Card and field recognition. Basic game-state tracking. Life Point and turn management. Rules enforcement. Card-effect processing. Interaction with holographic entities. Simple scripted opponents. Increasingly sophisticated AI decision-making.

Phineas stared at the list, then he looked back at Alexander. "You mean we can start testing before all of this is finished."

Alexander nodded. "That's the idea."

A crude Battle Box could be built first. It wouldn't have a sophisticated Duel AI; it might not even be capable of handling every card, but it could recognise cards, track their positions, maintain a basic game state and eventually produce a holographic monster in the correct location.

Then the team could test it. Find problems. Improve the hardware. Replace individual software modules. Expand the rules engine. Add more cards. And only afterwards move toward more sophisticated AI decision-making.

Phineas slowly leaned back in his chair as the implication was becoming obvious: They didn't have to build the entire Battle Box before they could test the Battle Box.

In the end, he said, "With this... The estimated time for a working prototype will be cut down by quite a bit."

In response to those words, the room became noticeably quieter as everyone was reaching the same conclusion.

The software was still difficult, but it was no longer an incomprehensible problem.

It was a collection of smaller problems.

Problems that could be assigned to different people. Problems that could be tested independently. Problems that could be solved one at a time.

Naomi was the first to speak, "I want to restructure the software team around this architecture."

Victor immediately began discussing the hardware changes that would be necessary. Elias started sketching modifications to the control systems in his notebook. Priya was already asking about the card-recognition prototype.

The meeting gradually stopped being a presentation, and it had become an engineering session.

Alexander mostly stayed quiet but answered questions when necessary, corrected assumptions when they came up and occasionally added a suggestion, but he deliberately avoided dominating the discussion.

This was their field.

He might have the knowledge necessary to design the architecture, but the people sitting around this table were the engineers who would ultimately turn those diagrams into working machinery, and Alexander trusted them to do it.

Eventually, Alexander brought the discussion back under control.

He looked at the revised schedule, then at Phineas and said, "All right."

The room gradually became quiet again before he asked, "How soon?"

Phineas studied the numbers for several seconds; it was clear that he didn't want to overpromise as he said, "Assuming everything goes well?"

Alexander nodded.

Phineas pointed toward the earliest milestones and declared, "Considerably sooner than the previous year-plus estimate."

Several members of the team exchanged looks.

"But," Phineas continued, "we need to be clear about what we're talking about."

He pointed toward the later stages. "This won't be a finished Duel Monsters system. It won't have a perfect AI. It won't be ready for commercial production."

Instead, their first real milestone would be something much simpler, a primitive Battle Box.

Something capable of recognising a card. Understanding where that card was on the field. Maintaining a basic game state. And producing a holographic monster where it belonged.

~~~▽~~~

By the time the meeting finally ended, the engineers were already carrying copies of Alexander's documents out of the room.

Several of them were still arguing about which component should be tackled first.

Naomi was discussing the software architecture with Victor. Priya was already thinking about how quickly they could construct a controlled card-recognition test. Elias had filled several pages of his notebook.

Alexander watched them leave with a small smile, but Phineas remained behind.

He stared at Alexander for several seconds, then he shook his head. "I still can't believe I'm saying this."

Alexander looked at him. "What?"

Phineas gestured toward the stack of documents. "You just walked into my R&D department and handed my engineers an entirely new software architecture."

Alexander shrugged. "I had a productive weekend."

Phineas stared at him, and there was a brief silence before he asked, "You expect me to believe that's all you did?"

Alexander smiled innocently, and Phineas simply shook his head.

For all the absurdity of the situation, however, Alexander couldn't help feeling satisfied.

The Battle Box wasn't finished, and there were still countless problems ahead, but for the first time, those problems no longer felt like one massive wall standing between Industrial Illusions and its next technological breakthrough. They had become visible individual steps, and now they could start taking them.