The Game Where I Was Rank One Became Reality

Chapter 329: Four

The variance appeared on the fourteenth day of the quarterly calibration cycle, and Oren almost missed it.

He’d been running the standard sequence, impedance checks on the primary trunk line, Ashenveil to Korthane relay station four, the segment that carried seventy percent of commercial telegraph traffic between the two powers. The procedure was documented in the Mechanist Telegraph Office’s operations manual, page forty-one, section three: "Quarterly Signal Integrity Assessment, Trunk Lines, Priority One." He’d performed this procedure eleven times. It took four hours. He had never found anything.

The impedance check required comparing the signal-to-noise ratio of the current trunk line against the baseline established at installation. Year 471. The baseline was precise: ninety-eight point four percent signal fidelity across a standard 200-character test message, measured at three points along the trunk, Ashenveil terminal, relay station two (midpoint, Greenvale junction), and Korthane terminal. Acceptable variance: ±zero point five percent, attributed to atmospheric conditions, line degradation, and the inherent imprecision of rune-amplified signal propagation over distances exceeding four hundred kilometres.

Current reading: ninety-seven point eight percent.

Within tolerance. Oren noted it in the calibration log, the bound leather volume, third shelf, behind the impedance gauge and the box of replacement rune-crystals that hadn’t been opened since Year 483, and moved to the next segment.

He stopped.

ninety-seven point eight percent was within tolerance. But he’d been running this check for eleven quarters, and the number had never been ninety-seven point eight percent before. It had been ninety-seven point nine percent last quarter. He remembered the number because ninety-seven point nine percent was the first time the reading had dropped below ninety-eight point zero percent, and dropping below ninety-eight point zero percent had given him a moment of concern that he’d dismissed because ninety-seven point nine percent was still within tolerance.

He pulled the logbook from the shelf. Quarterly entries dating to Year 475, when the telegraph office had begun systematic monitoring under the expanded Communications Infrastructure Act. He ran his finger down the column.

Year 475 Q1: ninety-eight point three percent. Q2: ninety-eight point three percent. Q3: ninety-eight point two percent. Q4: ninety-eight point two percent. Year 476: ninety-eight point two percent, ninety-eight point one percent, ninety-eight point one percent, ninety-eight point one percent. Year 477: ninety-eight point one percent, ninety-eight point zero percent, ninety-eight point zero percent, ninety-eight point zero percent. Year 478: ninety-eight point zero percent, ninety-eight point zero percent, ninety-seven point nine percent, ninety-seven point nine percent. Year 479: ninety-seven point nine percent, ninety-seven point nine percent, ninety-seven point nine percent, ninety-seven point nine percent. And so on. The numbers formed a line, and the line went in one direction, and the direction was down, and the rate of decline was consistent at approximately zero point zero five percent per quarter.

Consistent. Not random.

Atmospheric degradation was random. Line wear was random with seasonal weighting, heavier in winter, lighter in summer, averaging out across the year. Rune-amplifier drift was semi-random but self-correcting during routine maintenance cycles; the maintenance teams at relay stations two and three replaced amplifier crystals on a fixed schedule, and each replacement reset the drift to baseline. A consistent linear decline of zero point zero five percent per quarter across fourteen years was not random and was not atmospheric and was not line wear and was not rune-amplifier drift.

Oren sat at the calibration bench and stared at the logbook. He was not a senior engineer. He was not a signals analyst. He was a calibration technician whose job was to perform the standard sequence and note the numbers and flag anything outside the tolerance band. ninety-seven point eight percent was inside the tolerance band. The trend was not inside anything. The trend was a line that went down, and he didn’t have a procedure for lines that went down inside the tolerance band.

He checked his arithmetic. Counted the entries. Verified the decline rate against the data. zero point zero five percent per quarter. Fourteen years. Cumulative decline: approximately zero point six percent from baseline. The trunk line was operating at ninety-seven point eight percent signal fidelity. It would continue to operate. The commercial operators reported no signal loss. The messages arrived. The words were correct. The zero point six percent cumulative decline was invisible to any user and to any operator who checked the current reading against the tolerance band without comparing it to the historical baseline.

He wrote the observation in the calibration log. Flagged it for the senior engineer’s quarterly review. Noted the start date of the decline pattern: Year 475, first quarter, though the decline likely began at installation in Year 471, before systematic monitoring started. Noted that the decline was consistent with a systematic alteration to the signal propagation specifications rather than with any known environmental or equipment factor.

He did not know what to do with this beyond writing it down. The operations manual did not have a section for "systematic alteration to signal propagation specifications." The operations manual assumed that the specifications were correct because the specifications had been written by the Mechanist Institute and approved by the Korthane Commercial Authority and ratified by both governments and the specifications were the specifications.

He filed the observation. The senior engineer would read it next week. The senior engineer would note it. The senior engineer would compare it to the maintenance logs and determine that the equipment was functioning within design parameters and that the decline was an artefact of measurement drift rather than a systematic flaw.

The senior engineer would be wrong. Oren was not sure he was right. He was sure the line went down. That was enough for the calibration log and not enough for anything else.

The report arrived in Neth Myrvalis’s office on the third day of Emberdusk, mixed in with the weekly intelligence digest from the Korthane monitoring section. Seven pages. Fifteen subsections. Twelve of them routine, diplomatic courier traffic analysis, trade-balance summaries, port authority scheduling variances, the quarterly naval deployment estimates that had been consistently wrong for six quarters and were now read more for their consistency of error than for their content. The thirteenth subsection was the Mechanist Telegraph Office’s quarterly calibration summary, forwarded as a matter of standard distribution protocol because telegraph infrastructure fell under the Ministry of Whispers’ "Communications Security" mandate, a mandate established in Year 474 and funded at a level that permitted one dedicated analyst and three hours per week of review time.

Neth read the calibration summary. The numbers were small. The analyst’s note, "cumulative decline of zero point six percent across 14 years, consistent linear pattern, flagged for engineering review", was buried in the third paragraph of a five-paragraph subsection. The analyst who had written the summary had included the observation because including observations was what analysts did. The analyst had not flagged it as urgent because zero point six percent across fourteen years was not urgent. Nothing that moved at zero point zero five percent per quarter was urgent. Urgency required speed. This had patience.

He set the digest down. Removed his glasses. Cleaned them on the cloth he kept in the left desk drawer, the same drawer, the same cloth, the same motion he’d performed before reading the Korthane quarterly assessment in Year 478 and before reading the directional shift report in Year 481 and before reading the vulnerability window analysis in Year 483. The glasses had a scratch on the left lens that had been there for nine years. He’d stopped noticing the scratch seven years ago. He noticed it now.

The ±zero point three percent Thendris corruption.

Year 471. The original telegraph specification had been negotiated between the Mechanist Institute and the Korthane Commercial Authority. Neth had the negotiation transcript in his archive, not the official published version, which omitted the technical arguments, but the internal Ministry copy which preserved the sequence of proposals and counter-proposals in the order they were made. Section 4, Paragraph 7: signal propagation standards. The Korthane delegation had proposed a tolerance band of ±zero point five percent. The Mechanist delegation had counter-proposed ±zero point two percent. The compromise, ±zero point three percent, had been drafted by a Korthane technical attaché named Thendris, who had introduced the tolerance band as a "reasonable midpoint reflecting operational realities and atmospheric variance factors" and who had been reassigned to a provincial post six months after the treaty was signed. Neth had flagged the reassignment in his second annotation, Year 481. A technical attaché who designs a critical specification and then disappears into provincial obscurity is not a career trajectory. It is an exit.

The zero point zero five percent quarterly decline was within the ±zero point three percent Thendris tolerance. The tolerance was designed to contain the decline. The decline was designed to exist within the tolerance. The specification was not flawed. The specification was a weapon, and the weapon was working as intended, and the weapon had been working as intended for eighteen years, and the cumulative effect was a systematic degradation of signal integrity that would eventually cross the operational threshold and render the Ashenveil-Korthane telegraph trunk line unreliable, and the unreliability would be attributed to equipment failure rather than to the specification that had defined the equipment’s acceptable operating parameters.

At the current rate, zero point zero five percent per quarter, the signal fidelity would cross the operational threshold of ninety-five point zero percent in approximately twelve years. Year 501. The decline would become noticeable to commercial operators around ninety-six point five percent, approximately Year 495. At that point, the Mechanist Institute would order a full infrastructure audit. The audit would examine the equipment. The equipment would be functioning within specification. The specification would not be examined because the specification was the specification.

Neth put his glasses back on. The scratch was there. The scratch was always there. Exactly.

He opened the compartment. Behind the active filing divider. Three entries. His handwriting, angular Kobold script, precise from a distance and uncertain up close. The first: Korthane directional shift, Year 478. The second: directional shift continued, Year 481. The third: "They didn’t come," Year 483. Three annotations across eleven years. Three entries in the compartment that was not locked because the entries were not reports because reporting unconfirmed intelligence violated the professional standard he had maintained for ninety-two years and the professional standard was correct and his application of the standard was catastrophic and he knew it was catastrophic and knowing did not change the standard.

He picked up the pen. The pen was lighter than it used to be, lighter, or his hand was weaker, or both, or neither, or the weight of a pen was not something a person measured after ninety-two years unless the person was looking for something to measure instead of looking at the thing he didn’t want to measure.

He wrote.

Four: telegraph flaw. Specification compromise (Section 4, Para 7, Thendris draft) contains systematic signal degradation consistent with deliberate introduction at design stage. Source: Korthane or allied party. Cumulative effect: trunk line reliability declining toward operational threshold (~Year 501 at current rate). Engineering review will attribute to equipment. It is not equipment.

He capped the pen. Set it in the groove where his pens always sat, the groove he’d worn into the desk across two decades, the groove that Hollin had worn into her desk, the groove that everyone who sat at the same desk for long enough eventually created. A small trench in the wood that held the pen when the hand released it.

Four entries. the compartment closed. the compartment was not locked.

He walked to the window. The courtyard was empty. Emberdusk evenings emptied the Ministry by seventh bell, administrative staff to the residential quarter, analysts to their respective offices, the night watch to the eastern corridor. The courtyard had a fountain that hadn’t worked since Year 462. Nobody had repaired it. The repair requisition had been filed and unfiled and refiled and the fountain sat dry in the courtyard of the Ministry whose entire purpose was to ensure that the things nobody was watching were, in fact, being watched.

The glasses sat on the desk. He did not put them back on. The ink dried. The margin was full.

The report was clean.