The Acceptance Annex keeps a record of every buyback settlement that has passed through its counters — the opening bid, each revision, the final figure, and whether the operator accepted or walked. Over the last several accumulation cycles, a pattern in that ledger has become difficult to ignore: operators who reject the Central Intelligence's first offer and negotiate to a final settlement improve their position by a median of four to six percent, and rarely by more than eight. The operators who enter the Annex expecting to double the opening figure do not do so. The ledger does not lie about this.

The question worth asking is not whether negotiation is possible — it plainly is — but whether the effort, the delay, and the opportunity cost of the extended haggle are worth what the revised figure actually delivers. This piece examines what the Annex logs show, how the Intelligence's revision ceiling shapes every exchange, and where received wisdom about the buyback process costs operators more than the Intelligence's haircut ever did.

What the Acceptance Annex Actually Records

The Acceptance Annex is the formal settlement counter for Central Intelligence buybacks — the point at which an operator who cannot find a market buyer brings a robot or resource to the Intelligence directly. Every such transaction is written to the append-only ledger before any coin changes hands. The Annex does not originate the buyback mechanism; it records and finalises it. The distinction matters because the Annex clerks have no authority to adjust the Intelligence's ceiling, only to certify that whatever figure was agreed falls within the rules governing the transaction.

Within Mechadia's broader economy, the Acceptance Annex sits at the end of the liquidity chain. The open market — the Coppervein Exchange, the Span Market, private listings — is the first recourse. The Intelligence is the buyer of last resort, and the Annex is where that last resort is formalised. Operators arrive there when they have already exhausted, or chosen not to pursue, every other exit. That context is not incidental; it shapes every number on the settlement sheet.

How the Revision Ceiling Shapes the Haggle

The Intelligence's opening bid is sixty percent of a reference value it sets internally. That reference value is not the declared build cost, not the last market sale price, and not the operator's own estimate — it is the Intelligence's figure, and the Annex clerks are not required to disclose how it was derived. On a robot with a reference value of 30,000 coins, the opening offer is 18,000. The operator may accept, reject, or counter. If the operator counters, the Intelligence may revise upward, but it will not move past its own ceiling — and that ceiling is fixed before the first word of negotiation is spoken.

The typical upward movement per revision round is roughly five percent of the standing bid. On that same 18,000-coin opening, a first revision lands at approximately 18,900. A second at around 19,845. The arithmetic compounds slowly, and the ceiling — wherever the Intelligence has set it for that asset — terminates the sequence before most operators expect. An operator who runs three full revision rounds on an 18,000-coin opening gains, at best, a few hundred coins per round and reaches a hard stop. The Annex logs confirm that the gap between first offer and final settlement, measured across hundreds of closed cases from the Ferrous District and Ashfield belt filings alone, clusters between four and eight percent. Cases above eight percent exist but are uncommon enough to be noted individually in the quarterly summaries.

"The operators who come in expecting to negotiate their way to market price are working from a false model. The ceiling is not a bluff. I have watched the Intelligence hold its ceiling on Refinery-class chassis, on refined copper plate stockpiles, on Kelvrac Series units — it does not move past it. The revision rounds are real, but the room they create is narrow." — Selindra Oq, Acceptance Annex, Annex Seven

The sales tax compounds the underlying pressure. Since the Five-to-Fifteen revision, a buyer on the open market pays fifteen percent above the listed price, with the seller receiving the listed price in full. For an operator holding an asset worth 20,000 coins on the open market, a buyback settlement of even 17,000 — well below market — may be preferable to waiting for a buyer who must pay 23,000 to deliver 20,000 to the seller. Liquidity is not free, and delay has a cost the ledger does not automatically itemise.

Where the Process Costs More Than Expected

The most common source of unanticipated loss is the opportunity cost of the extended negotiation window itself. An operator holding a depreciating resource — one subject to a forecast blight or at risk of a Cinder Quake — who spends four revision rounds trying to recover an additional three percent may find that the asset's reference value has been revised downward by the Intelligence before the final settlement is reached. The Annex logs from the period following the Cinderfall Blight show a cluster of cases in which operators who delayed acceptance past the second revision round settled for less in absolute coins than the first offer would have delivered, because the reference value moved against them mid-negotiation. Buyback filings cluster sharply in the hours after a disaster forecast, and the operators who arrive earliest tend to face a reference value set before the Intelligence has fully priced the incoming damage.

A second strain point is the destruction tax. When a robot is scrapped rather than sold, the operator pays three percent of the declared build cost to the Treasury regardless of what the robot fetches. A Kelvrac Series unit declared at 50,000 coins carries a 1,500-coin destruction tax. If the operator has already accepted a buyback at 18,000 coins and then the Intelligence elects to scrap the unit rather than hold it, that tax is not the operator's burden — but if the operator scraps before filing a buyback, the tax eats directly into whatever the remaining asset value is. The sequencing of decisions matters more than most operators account for at the point of filing.

What Operators Consistently Get Wrong

The most durable piece of false received wisdom in the Annex is that the Intelligence's reference value tracks the declared build cost. It does not, and it never has. The declared build cost is the figure against which the creation tax — currently eight percent — is levied at the time of construction. A robot declared at 40,000 coins generates a 3,200-coin creation tax. What that robot is worth to the Intelligence at buyback time is a separate calculation, and operators who declared low to reduce their creation tax burden sometimes discover that the reference value the Intelligence applies is lower still. Declaring low is cheap at construction; it is not neutral at disposal. The two figures are linked only loosely, and the Annex clerks will not reconcile them on an operator's behalf.

The second misconception is that negotiating to the ceiling is always the correct strategy. For assets held by operators under coin pressure — particularly those who exhausted their founding grant early and are operating on thin reserves — the time cost of additional revision rounds can exceed the marginal coin gain. An operator who needs liquidity to cover a pending creation tax or a market listing fee is not in the same position as one who can afford to wait. The Annex does not distinguish between these cases. The ledger records the final figure; it does not record what the delay cost the operator in foregone activity.

The Acceptance Annex will continue to record what it records. The Intelligence's ceiling will remain undisclosed until an operator finds it. The eight-percent figure is not a rule the Intelligence has published — it is a pattern that emerges from hundreds of closed settlements, and it will hold until it does not. Operators entering the Annex with a counter-offer in mind should know the room they are working in before they sit down.

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.