Sunday, November 18, 1990.
At 9:03 AM, Kozlov appeared at the hotel porch on time.
He had changed his coat today.
It was still dark blue, but newer than yesterday's, and the shoulder seams did not sag.
The motorcade drove along the south bank of the Moscow River for twenty minutes.
Kozlov talked the entire way.
He spoke of the history and development of the Soviet Union Academy of Sciences, of the Academy's contributions to both defense and civil sectors, and said that "The Soviet people's enthusiasm for scientific endeavors cannot be measured in numbers."
Shuichi, holding the hot tea provided by the hotel, nodded, said "Remarkable," said "Admirable," and said "The prospects for scientific exchange between Japan and the Soviet Union are undoubtedly vast."
Kozlov would pause every few sentences to let the accompanying liaison translate, then continue speaking.
Satsuki sat by the window, listening, occasionally nodding slightly at appropriate moments.
She did not speak.
The streets outside the window unfolded in the gray morning light.
At this hour, there were a few more pedestrians on the road.
Someone was holding a paper bag, walking very quickly, as if rushing to meet some deadline.
Someone else was walking a hound; the dog's joints were prominent, and its waist and abdomen were tucked in deeply.
Satsuki withdrew her gaze.
The building of the Computing Center of the Soviet Union Academy of Sciences was at the end of a relatively secluded street.
By the time the motorcade stopped in front of a gray-white building, the snow had already stopped.
The building itself did not look shabby.
It had a wide main entrance, thick stone exterior walls, and a metal plaque with a somewhat darkened surface embedded above the lintel.
But after walking inside, the chill did not immediately disappear.
The hallway was very cold.
Unlike the hotel for foreign guests where Satsuki and the others were staying, the heating here could only be described as "lukewarm."
"Why is it so cold here... I was clearly woken up by the heat a few times last night."
Emi, following beside Satsuki, hunched her shoulders slightly and whispered.
"Perhaps it is because they need to 'concentrate resources to accomplish big tasks.'"
"The priority of the Academy of Sciences is probably lower than that of our hotel for foreign guests."
Satsuki did not turn her head, speaking softly.
"Keep up. This is the part you are most interested in."
"I have heard for a long time that Soviet mathematics is world-class!"
Emi jogged to catch up with the group.
"Your Excellency Saionji, welcome, welcome."
The person receiving them was the Deputy Director of the Computing Center.
His surname was Belov, he was forty-seven or forty-eight years old, his hair was combed back very neatly, and he wore a pair of thin metal-framed glasses on the bridge of his nose.
His old suit was clearly pressed, the shoulder seams were smooth, and his tie was tied very properly.
It was just that the cuffs were badly worn.
The fabric of the left cuff had already faded to white, and a small thread was sticking out on the right.
When he spoke, he would subconsciously pull his hands to his sides, as if worried others would see.
Shuichi stepped forward and shook his hand.
"Deputy Director Belov, thank you for your hospitality today."
Kozlov stood between them, smiling professionally as he acted as the translator.
"The Computing Center attaches great importance to this visit." Deputy Director Belov said in Russian, "We have long been committed to numerical calculation, scientific engineering simulation, and international academic exchange."
"In the past three years, the center has participated in six international cooperation projects, published twenty-nine papers, and researchers have traveled to Warsaw, Prague, and East Berlin to attend conferences."
Kozlov translated steadily.
Shuichi nodded, his tone gentle.
"Your center's achievements are fruitful and admirable."
Satsuki stood half a step behind her father, her hands folded in front of her, with a well-behaved smile that did not steal the spotlight.
She looked like she was listening carefully.
In reality, her gaze had already shifted from Deputy Director Belov's cuffs to the other side of the corridor.
The first office had its lights on, and there were three people inside.
An older researcher was hunched over the table writing something, and next to him were two young people sharing a terminal.
The second door was closed, with a piece of paper stuck to the glass that read "Equipment Maintenance."
There were no lights in the room.
The third room had its door half-open; the bookshelves inside were half-empty, and there were two cardboard boxes on the table, as if someone had just moved out, or as if no one had returned in a long time.
The bulletin board was next to the stairs.
When Satsuki passed by, she did not stop, but her eyes swept over the papers on it.
Notice on the Adjustment of Consumables Rationing for the Fourth Quarter of 1990
Temporary List of Equipment Transferred to Mothball Status
Review of Dormitory Allocation for Young Researchers
There was also a personnel transfer notice with curled edges.
She saw a few names marked with "Expatriate," "Advanced Study," or "Pending."
Pending.
Satsuki lowered her eyelashes.
Emi was holding her tool bag, initially still a bit sluggish.
Her gaze wandered between the wall plaster, the radiators, and the wooden door handles, finally landing on an old printer on a cart.
"Satsuki-chan..."
She lowered her voice.
"The equipment here feels even older than the stuff at Todai."
"I thought the machines at Todai were old enough already."
Satsuki did not turn her head.
"Yes."
"If they run simulations on these machines, they will be running them until next year."
"They are using these machines to allow the Soviet Union to compete with the United States, after all."
Emi blinked.
She looked up at the front.
The double doors at the end of the corridor were pushed open, and a low fan sound came from inside.
They had arrived at the server room.
The server room was warmer than the corridor.
But only slightly warmer.
Several rows of cabinets were arranged along the wall, gray, beige, and light green mixed together, with numbers pasted on the casings.
Some of the nameplates were worn, but the production years could still be clearly seen — 1981, 1983, 1985.
The newest one had a nameplate from 1987.
Satsuki's gaze swept along the next row of cabinets.
There were none after that.
1987... After Gorbachev came to power in 1985, was there no way to update the equipment?
As soon as Emi came in, her eyebrows twitched.
She was about to speak, but her gaze stopped at the back of the cabinets.
Cables.
The old machines themselves were not surprising, but the cabling was very clean.
Not the "pretty" kind of clean.
It was the kind of clean where, after countless unplugging, modifications, and replacements, order was still maintained with the minimum amount of material.
Every bundle of wires was neatly tied with cloth tape, and the labels were handwritten; there were traces of different people's handwriting, but the format was consistent.
The cold and hot airflow had been artificially separated; several pieces of equipment with the worst heat dissipation had been moved to positions near the windows, and a homemade baffle had even been installed under the window frame to direct cold air to the back of the cabinets.
These antiques were being well cherished.
Emi's expression slowly changed.
"...Huh."
Satsuki glanced at her.
"What is it?"
"These people really know how to make do."
Emi said softly.
Satsuki could not help but smile.
"Is that a compliment in a server room?"
"Yes." Emi stared at the row of cables. "A very high compliment."
Deputy Director Belov had already stood at the front of the server room.
"Next, we have arranged for three researchers to introduce the center's recent research directions to everyone."
After Kozlov translated, he added a sentence.
"They are all very experienced comrades in the center."
The first researcher was an old professor in his fifties.
His surname was Petrov, he had graying hair, wore a dark brown sweater, and an old wool coat over it.
The content of the introduction was traditional numerical calculation, solving partial differential equations, fluid simulation, and stress analysis of engineering materials.
He spoke in a standard, conventional manner.
The formulas on the blackboard were written steadily, every conclusion had a source, and every set of data was kept within the range allowed for display.
Shuichi did not understand the details, but he could tell that this person was very old-school.
So, after the other party finished speaking, he smiled and said:
"Your solid research is the foundation that any country's scientific system needs."
Professor Petrov was stunned for a moment.
He probably did not expect a Japanese Kazoku to say such a thing.
Subsequently, he bowed slightly.
"Спасибо. (Thank you.)"
The second was a female researcher in her early forties.
Natalia Mikhailovna.
She wore a dark gray suit skirt, her hair was coiled at the back of her head, and she held a few pages of typed manuscript in her hand.
She was doing theoretical work on communication protocols.
She spoke very logically, or rather, too logically.
Each paragraph seemed to have been scoped out in advance, the speaking speed was not fast, and the key points were clear.
"Between distributed nodes, we mainly study message confirmation mechanisms in high-fault-tolerance environments, as well as redundant coding schemes in unstable physical links."
Kozlov translated with some difficulty.
Emi was just listening at first.
When she heard a certain word, she suddenly raised her head.
Natalia stopped for less than half a second halfway through speaking.
That half-second was very short.
Short enough that Shuichi just thought she was taking a breath.
But Satsuki saw it.
Natalia's gaze shifted from her manuscript to a rolled-up diagram next to the blackboard that was not being displayed.
She seemed to want to add something, but Deputy Director Belov coughed lightly beside her.
So she continued reading her manuscript.
"...The above are the periodic results that can be introduced at present."
Can be introduced.
Satsuki kept this phrase in her heart.
The third researcher came out last.
Alexei Orlov.
In his thirties, thin build, hair a bit messy, with faint dark circles under his eyes.
The sleeves of his sweater were rolled up, and there were ink stains on his fingers.
Compared to the previous two, he did not have as much ceremonial sense when he went on stage; it was as if he had just been called over from another room to complete a task that had to be fulfilled.
When Deputy Director Belov introduced him, his tone was also much shorter.
"Comrade Orlov is responsible for some application work in network simulation and parallel computing scheduling. Now please let him perform a standard demonstration."
Application work.
Satsuki paid a little more attention.
Orlov sat down at an old terminal and typed a few commands.
Green characters appeared on the screen, and the program began to run.
At first, Emi was just watching politely.
After ten seconds, her eyes fixed on it.
After twenty seconds, she leaned forward a bit.
After thirty seconds, she fished a paper notebook out of her tool bag, took out a pencil, and started writing rapidly.
Satsuki stood beside her, unable to understand those output parameters.
But she could understand Emi.
Usually, when Emi saw boring equipment, the corners of her mouth would press down slightly, and her fingers would touch the screwdriver in her tool bag.
When she saw something interesting, she would forget where she was.
Now she had forgotten to blink.
On the terminal screen, tasks were cut into fragments, distributed among different nodes, reclaimed, and redistributed.
The machine was very old.
But the response speed was wrong.
Emi muttered under her breath.
"This is impossible..."
She wrote a word in her notebook, then crossed it out.
"That is not right either..."
Orlov's demonstration ended, and he was just about to stand up.
"Excuse me."
Emi suddenly called out to him in English.
"Your bottleneck is not computation. Where is the latency? Transport layer, or scheduling queue?"
The server room went quiet for a moment.
Orlov was stunned.
He looked at Deputy Director Belov.
Belov was talking to Shuichi, and Kozlov was also translating, not paying attention to this side.
Orlov hesitated for a moment, then replied in English.
"Neither. Physical link is slow, yes, but the real delay is synchronization barrier. We avoid global barrier when task graph allows partial ordering."
Emi's eyes lit up.
"Then you are not doing simple batch scheduling. You have dependency prediction?"
"Not prediction. Conservative estimation. If dependency graph is sparse, we pre-allocate window."
"Window size?"
"Dynamic. Based on failure rate and message acknowledgement."
"Your acknowledgement is too expensive."
"We compress it."
"How?"
Orlov paused.
This time, he did not answer immediately.
He asked back:
"If your nodes fail silently, do you trust timeout, or redundant witness?"
Emi's pencil stopped on the paper.
She looked up at him.
It was as if a narrow bridge that no one else could see had suddenly appeared between the two of them.
"Depends on cost of false positive."
Orlov smiled a little.
That was the first time since he entered the room that he showed an expression that was close to real.
"Good answer."
Emi smiled too.
"Of course."
She was just about to continue asking, but Deputy Director Belov had already turned around and walked over here.
"Time is almost up."
"Next, there is a tour of the archives."
Orlov shut his mouth.
Emi also tucked her notebook into her arms, as well-behaved as if nothing had just happened.
Satsuki looked at Belov, then at Orlov.
Judging from that director's attitude, he was not very valued; one could even say he was a marginalized person.
Of course, this did not mean he was a person who was completely buried.
Being brought here meant that at least someone knew he was useful.
It was just that this system was accustomed to writing value on a different kind of form — equipment, staffing, project level, superior approval, demonstrable results.
A machine could be placed in an exhibition hall.
A terminal could be captured in a report photo.
A blueprint for an automated management system could be hung on the wall.
But application work was different; it was not glamorous enough.
How much the synchronization barrier was reduced, how much waiting time was saved after message confirmation was compressed, how much computing power the whole system lived on after the task graph changed from global blocking to partially ordered — these things did not have a beautiful shell, and it was very difficult to write them into a sentence that would make a minister nod.
This system did not put its own gems in the display window.
The side effects of the technocracy had already caused this country to begin to stagnate.