Ten Years in Public

Legal counsel proposes five exceptions: active investigation, litigation risk, confidential business information, personnel privacy, and material commercial harm.

June reads the list once. "That excludes every quarter we would most need to see."

Counsel argues that disclosure could prejudice prosecutions, reveal security details, violate privacy, or help competitors. Some risks are legitimate. The language is still broad enough to hide wages as commercial harm and closures as litigation risk.

Voss asks counsel to rewrite each exception around the information protected, not the institution's discomfort.

Personal privacy may remove names, health details, addresses, and individual grievance facts. Active investigations may delay only information prosecutors or regulators identify in writing as harmful to the case. Security may withhold exploit details while publishing the existence, affected system, response, and correction deadline. Trade information may protect current bids or proprietary designs but not labor, benefit, safety, or public-fund outcomes.

There will be no general litigation exception.

"Publishing an error can create liability," counsel says.

"The error created liability," Voss replies. "Publication identifies it."

Her position costs the board the shield she once preferred.

We negotiate correction deadlines. Material data errors require notice within five business days and corrected publication within fifteen. If the source remains disputed, the dashboard shows both values, custodians, and the reason. Historical reports remain accessible with correction links; no quiet replacement.

Dev requires a twenty-four-hour notice for serious safety conditions regardless of quarter. June adds seven days for unpaid-wage patterns or benefits interruptions. Camille insists an individual complaint must not become public before the worker chooses or a lawful aggregate exists.

The employee directors propose an independent audit every year. Board counsel asks whether the company can select the auditor. They reject unilateral selection. Wondervale, the employee trust, and city oversight each nominate from a public roster; conflicts are disclosed; the governance committee selects by recorded vote.

City counsel created the covenant draft at 11:46, revised it in the shared portal, and preserved every redline. The city clerk will hold the executed original. Its vulnerability is enforcement over time. Future leaders may classify aggressively, delay source systems, or underfund audits even when the words remain binding.

Hart adds remedies. Missed publication triggers thirty days to cure. Repeated failure can suspend subsidy benefits, reopen labor remedies, and require a special board hearing. Employee representatives and the trust may seek enforcement without proving individual financial loss.

I ask whether a technical delay should risk public funds needed for wages.

June proposes proportional remedies. First failures require correction and outside support. Repeated willful concealment can reach subsidy suspension, but money designated for wages, core benefits, and participant payments remains protected. Accountability should not punish the people the covenant serves.

Rina builds the public dashboard prototype. It shows current quarter, five-year trend where available, definition changes, corrections, and unresolved disputes. Each metric links to methodology and custodian. Personal data never enters the public layer.

June tests the dispute function with a fictional wage error. Payroll reports a ninety-eight-percent on-time rate; the employee archive reports ninety-six because two corrected checks arrived after the contractual deadline. The dashboard cannot average the figures or choose the company source silently. It displays both, names the different definitions, and opens a correction clock assigned to an independent reviewer.

Dev runs a safety test in which a serious hazard is reported at 11:58 on the final day of a quarter. Counsel's draft would move it into the next report because investigation remains open. The committee changes the rule. The occurrence and immediate control appear within twenty-four hours; cause and final remedy may follow as labeled updates. A reporting boundary cannot become forty-eight hours of darkness.

Camille brings five sample grievances and asks Rina to prove that small departments will not expose complainants. One category narrows to a single night-shift worker when filtered by month and role. The public layer blocks the filter, preserves the aggregate count, and sends authorized investigators to the protected source. Employee representatives test the same restriction from outside the company network.

We test it with the first available report.

Average year-round wages increased. Benefit enrollment and claim payment improved after the freeze. Safety resolution time declined. Creator credit corrections rose because the new process found old omissions. Seasonal turnover remains high.

Counsel suggests delaying the turnover metric until bargaining begins. It may undermine reopening publicity and CEO recruitment.

"Then the first report would prove the covenant is optional when a number is inconvenient," Voss says.

The metric stays.

The seasonal employee director asks for turnover by contract length and department. Camille allows aggregate breakdown where groups are large enough to protect identity. Small departments combine into operational categories. The dashboard labels suppression thresholds so privacy does not appear as missing data.

The director also asks for voluntary departures separated from contract completions. HR cannot reconstruct that distinction reliably for two old years because exit codes were optional and supervisors used them inconsistently. The dashboard marks those years unavailable and publishes the data-quality gap. Future exit records let workers review the category assigned to them, while declining an interview remains a valid choice.

Rina invites worker representatives to try to make a misleading good-news chart from the same data. They shorten the wage timeline, exclude unfilled shifts, and select a benchmark below the regional median. Each manipulation is technically defensible and materially distorting. The covenant therefore fixes comparison periods, requires workforce denominators, and publishes alternative regional benchmarks chosen by the city rather than communications.

At three, the board committee votes. One director opposes public correction history. Another wants the covenant to expire after the city subsidy term. Employee representatives demand duration long enough to outlast current leaders.

Ten years becomes the compromise.

The term covers two board cycles, Elliot's five-year contract, the trust's transition, and likely changes in city leadership. Amendment requires approval from the independent board majority, both employee seats, and municipal oversight after public notice. Privacy protections may strengthen without that supermajority; disclosure duties may not weaken.

At 4:28, Wondervale, the employee trust, and the city sign. Voss signs for the board committee. June and both employee directors sign for workforce governance. Hart signs municipal enforcement. I acknowledge operating duties without controlling amendment.

Before execution, June asks what happens if the public dashboard itself becomes inaccessible. The covenant requires downloadable tables, a plain-language summary, screen-reader testing, and paper access at the employee desk and city library. An outage longer than four hours triggers a mirrored city page and a timestamped notice. Publication means usable access, not a link that technically exists.

The committee schedules the first independent audit before selecting a vendor. Its scope includes source completeness, privacy suppression, correction speed, and attempts to influence classification. Auditors may interview workers privately and report obstruction as a finding. Fees come from a protected compliance line that the CEO cannot reduce midyear without the same public amendment process.

The city clerk enters the disclosure covenant into the subsidy record, while the board secretary and employee trust place matching copies in their archives. Hashes match across all three. Its limitation remains human behavior. Public rules create evidence and remedies; they do not make future people brave.

The first dashboard publishes before communications releases reopening hours.

Press asks why Wondervale would announce persistent seasonal turnover on a day intended to restore confidence. I answer that confidence built by omitting the current problem would repeat the practice that closed us.

June answers the more important question: what happens next. Paid bargaining sessions, schedule pilots, housing analysis, and a worker-selected negotiating team. The dashboard does not solve turnover. It gives the next leaders a number they cannot bury without a record.

Voss remains after signing. "I spent years believing controlled minutes protected the company from unfinished facts."

"What changed?"

"The facts finished without us and cost more."

She leaves before I can turn the statement into absolution.

At 5:10, the first ten-year report locks into the public archive. Wage gains appear in green only because the dashboard uses neutral trend markers rather than celebration colors. Beside them, seasonal turnover remains above the regional benchmark.

The reforms improved pay.

They did not convince seasonal workers to stay.