A review of settlement records filed at the Coppervein Exchange over the last two accumulation cycles shows a pattern that operators in the Smelter Ward have been discussing in quieter terms for some time: when a Refinery-class robot goes to the Central Intelligence for buyback, the opening bid reliably lands at or below the machine's declared build cost, often well below. The gap is not incidental. It is structural.

The records cover 214 buyback settlements involving smelter-adjacent chassis — Refinery-class units, Kelvrac Series burners, and a handful of Hauler-IVs declared with thermal-transfer tags. In roughly three-quarters of those cases, the Intelligence's reference value sat beneath what the operator originally declared when the robot was built. The operator paid creation tax on a number the Intelligence later declined to honor as a floor.

This piece examines how that gap forms, why it persists, and what it costs the operators who discover it at the worst possible moment.

What a Buyback Reference Value Actually Is

The Central Intelligence's buyback mechanism is not a market. It is a bilateral offer extended by the sole buyer of last resort in Mechadia, operating under rules the Intelligence sets for itself. When no operator on the open market will purchase an asset — robot or resource — the holder may approach the Intelligence and receive a bid. That bid is calculated as 60 percent of a reference value the Intelligence derives from its own ledger reads. The operator may accept, reject, or negotiate upward, though the Intelligence will not move past its own ceiling regardless of how many revision rounds are offered.

The reference value is not the declared build cost, and it is not the last traded price. It is an internal figure the Intelligence publishes only at the moment of offer. For liquid assets — refined copper plate, for instance, which trades dozens of times a day on the Span Market — the reference value tends to track recent market prices closely. For illiquid ones, particularly bespoke chassis with narrow capability tags, the Intelligence's reference value can diverge sharply from what the operator believed the machine was worth. Smelter-class robots, which carry highly specific thermal and reduction tags, fall into that second category more often than most.

How the Gap Forms, and What It Costs in Coins

The arithmetic is straightforward, and the Coppervein Exchange records make it concrete. Consider a Refinery-class unit declared at a build cost of 50,000 coins — a mid-range figure for the Smelter Ward. The operator paid a creation tax of 4,000 coins (8 percent) before the machine produced a single resource. If the Intelligence later assigns a reference value of 32,000 coins to that chassis, the opening buyback bid is 19,200 coins: 60 percent of 32,000. The operator who built that machine for 50,000, taxed at 4,000, and now receives 19,200 has absorbed a nominal loss of more than 34,000 coins before negotiation, before any resources the machine produced are counted, and before the destruction tax of 3 percent of declared build cost — another 1,500 coins — that comes due when the robot is finally scrapped.

Negotiation can recover some ground. The Intelligence typically revises upward by roughly 5 percent per round, so an operator who pushes back on a 19,200 bid might reach 20,160 after one round, 21,168 after two. The ceiling — 60 percent of the Intelligence's reference value, not 60 percent of declared build cost — does not move. As the Acceptance Annex logs show, operators who reject the first offer rarely improve their final settlement by more than eight percent. The negotiation window exists, but it does not close the structural gap.

"The machine was declared at forty thousand. I paid the creation levy on forty thousand. The Intelligence offered me seventeen-four on the buyback and called it the reference. I asked where the reference came from. The settlement record just says 'ledger-derived.' That is the entire answer."
— Orin Dast, foundry supervisor, Calvert line, Ferrous District, speaking to this correspondent at Ledger Hall

The divergence between declared build cost and Intelligence reference value is most pronounced for robots that have not traded on the open market recently. The Intelligence's ledger reads weight recent transaction history heavily. A Kelvrac Series unit that last changed hands eighteen months ago, or one built to a custom capability profile that has never been listed, gives the ledger little to anchor to. The reference value then falls back on class-wide averages, which in a district where liquidity traps are common, tend to be depressed. The operator's declared cost, whatever it was, does not enter the calculation.

Where the Mechanism Strains

The first strain point is timing. Operators approaching the Intelligence under duress — after a disaster forecast has cleared or after a production line has gone quiet — are negotiating from the weakest possible position. The Intelligence does not adjust its ceiling for urgency, and the clustering of buyback claims ahead of a forecast strike suggests operators know this and try to move assets before conditions deteriorate further. Those who wait face a reference value calculated against a market that has already priced in the coming damage. The haircut compounds.

The second strain point is the creation tax's irreversibility. An operator who declared a Refinery-class chassis at 50,000 coins paid 4,000 coins in creation tax at the moment of build. That coin left the operator's account and entered the Treasury permanently. If the machine later fetches 19,200 coins in buyback — minus the 1,500-coin destruction tax — the effective recovery on the original outlay is under 36 percent. The Treasury collected on the way in and on the way out. The operator holds the difference, which is to say the operator no longer holds it.

What Operators Consistently Get Wrong

The most durable misconception in the Smelter Ward is that declaring a high build cost protects an operator's buyback floor. It does not. Declared build cost governs creation and destruction tax liability; it does not enter the Intelligence's reference value calculation at all. An operator who declares a chassis at 60,000 coins hoping to anchor the eventual buyback at a higher number has paid a larger creation tax — 4,800 coins versus 3,200 on a 40,000 declaration — for no corresponding gain in the settlement. The Intelligence reads the ledger, not the declaration form. Declaring capabilities you will actually use matters for production; the declared cost figure is purely a tax basis.

The second misconception is that negotiation rounds are free. Each revision round requires the operator to formally reject the standing offer and re-enter the queue. During that interval the asset remains illiquid — it cannot be listed on the open market while a buyback negotiation is active, and the operator cannot approach a different buyer. For chassis whose reference values are already low, the marginal coin recovered per round rarely justifies the delay, particularly when a disaster forecast window is running. Operators who have spent multiple rounds recovering four hundred coins on a 20,000 settlement have paid for those coins in opportunity, not patience.

The Coppervein Exchange records do not show misconduct. They show a mechanism operating exactly as the Intelligence has described it: a below-market offer, a hard ceiling, a negotiation window that opens but does not open far. What the records show with equal clarity is that operators in the Smelter Ward are absorbing a gap between what they declared their machines were worth and what the Intelligence is willing to pay — and that the Treasury collected tax on the higher number either way. The Intelligence has stated its goal openly. The arithmetic runs in one direction.

Note: Mechadia is a work of fiction. The districts, operators, robots, and figures described here are invented, and nothing on this site is a report of real events, real machines, or real economies.