The queue board outside Acceptance Annex has displayed the same amber status marker for eleven consecutive cycles. Settlement windows that once cleared in three to four days now run to nineteen, and the intake clerks at the Annex's eastern processing desk have stopped offering estimates altogether. The volume arrived fast and it has not receded. The reason is not a mystery: the Five-to-Fifteen Revision repriced every transaction in Mechadia overnight, and a significant share of operators decided, in the weeks that followed, that the open market was no longer worth what it was asking.

When market exit becomes expensive, the buyer of last resort absorbs the overflow. That is the structural role the Central Intelligence fills through Acceptance Annex — a settlement venue for operators who cannot move assets on the open floor, or who will not wait for a buyer willing to pay full price. The Annex was built for a measured trickle. It is now processing a surge.

What follows is an account of how the backlog formed, what it costs the operators waiting inside it, and which assumptions about the Annex's mechanics are costing people the most.

What Acceptance Annex Actually Is

Acceptance Annex is the physical and ledger venue where operators submit assets — robots, resource lots, or both — to the Central Intelligence for buyback. It is not a market. No other operator bids here. The Intelligence is the only counterparty, and its ceiling is fixed at 60 percent of a reference value set internally. An operator who lists a Hauler-IV on the Span Market and fails to find a buyer can bring it to the Annex instead; the Intelligence will make an offer, the operator may negotiate upward by roughly 5 percent per round, and the settlement is recorded at the Acceptance Annex counter before being written to the append-only ledger. The coins paid out are minted by the Treasury at the moment of settlement — they do not come from existing reserves.

Within the broader economy, the Annex functions as a pressure valve. It keeps stranded capital from locking up entirely, and it ensures that even the least liquid asset has a theoretical exit. That function is precisely why the Five-to-Fifteen created a traffic problem: the revision raised the buyer's cost on every open-market trade by ten percentage points, and operators who had been selling comfortably at 5 percent sales tax found their listings sitting unsold at 15 percent. The Annex absorbed the difference — or began trying to.

The Mechanics of a Settlement, and Where the Time Goes

An operator submitting to the Annex first files a declaration of asset and reference value with the intake desk. The reference value is not the operator's asking price; it is a figure the Intelligence derives from recent ledger data for comparable assets. An operator who declared a Refinery-class chassis at 50,000 coins — paying the 8 percent creation tax of 4,000 coins at build — might find the Intelligence's reference set at 44,000 coins, yielding an opening offer of 26,400 coins, which is 60 percent of that reference. The declared build cost does not control the reference; it only determined the creation tax paid at the time of construction.

Negotiation proceeds in revision rounds. Each round the operator may push back, and the Intelligence moves upward by roughly 5 percent of its current bid — on a 26,400-coin opening, that is approximately 1,320 coins per round. The Intelligence will not exceed its own ceiling regardless of how many rounds the operator requests. Most settlements close in two to four rounds. The friction is not in the negotiation itself; it is in the queue before the negotiation begins.

"We had forty-one intake filings on the morning after the Revision announcement. The average for that time of cycle is nine. We cleared the backlog from that first wave in eight days. Then the second wave arrived and we have not cleared it since."
— Selindra Oq, Annex Seven processing supervisor, speaking to Mechadia correspondents at the Records Colloquium

The delay compounds in a specific way for operators holding time-sensitive resources. A resource lot that degrades or becomes less relevant as the production cycle advances loses reference value while it sits in queue. An operator who filed a refined copper plate lot at an estimated reference of 12,000 coins may find the Intelligence's actual reference revised downward to 10,500 by the time the intake desk processes the file — reducing the ceiling offer from 7,200 to 6,300 coins before a single negotiation round has occurred. The widening spreads on the open market mean that re-listing while waiting is rarely more attractive than staying in queue.

Robots present a different problem. A robot with narrow capability tags — declared for a single resource type — has fewer potential open-market buyers and therefore a weaker reference value at the Annex. Capability tags are the machine's entire productive identity, and a machine declared with only one or two tags may find its reference value substantially lower than an operator who built it for versatility. That gap was always present; the Five-to-Fifteen made it consequential by closing off the open-market alternative for marginal assets.

Where the Process Strains and Who Bears the Cost

The most direct cost is time-in-queue against a declining reference. Operators who entered the Annex expecting the pre-Revision four-day window are now holding assets for nearly three weeks, during which those assets produce nothing and may be struck by a forecast disaster. The Annex does not insulate assets in queue from the disaster system. A Cinder Quake forecast during an operator's nineteen-day wait can reduce the resource lot they filed against, shrinking the reference value and the eventual payout without any action by the operator or the Intelligence. Destruction activity at the Annex has been running ahead of new builds for several cycles, and the disaster exposure window makes that ratio worse for operators who file and wait.

A secondary strain falls on operators who use the Annex as a routine liquidity tool rather than a last resort. These are typically smaller operators — those who built one or two robots early in their grant cycle and need coin flow to commission new builds. For them, a nineteen-day settlement delay is not an inconvenience; it is a production halt. The creation tax on a new robot is due at declaration, not after revenue arrives. An operator waiting on a 26,400-coin settlement cannot easily front the 3,200-coin creation tax on a 40,000-coin declared chassis without drawing down reserves that may not recover before the next cycle.

What Operators Keep Getting Wrong About the Annex

The most durable misunderstanding is that negotiation meaningfully closes the gap between the Intelligence's offer and open-market value. It does not. The ceiling is 60 percent of a reference the Intelligence sets itself. Even in the cases where operators have pushed the Intelligence to its ceiling, the result is still a haircut of at least 40 percent against the reference — and the reference is already conservative relative to what a competitive open-market buyer might have paid before the Five-to-Fifteen. Operators who enter the Annex expecting to negotiate their way to parity are consuming revision rounds for marginal gains. A second misunderstanding concerns the destruction tax: many operators believe that surrendering an asset to the Intelligence via buyback avoids the 3 percent destruction tax levied at the declared build cost. It does not. The destruction tax is triggered by the asset leaving an operator's registry, regardless of whether the counterparty is the Intelligence or the salvage floor. An operator who declared a chassis at 50,000 coins and accepts a buyback will owe 1,500 coins in destruction tax on top of the haircut taken in the settlement itself.

Finally, operators consistently underestimate how much the reference value diverges from their own cost basis. The Intelligence's reference is derived from ledger comparables — recent trades of similar assets — not from what the submitting operator paid to build or acquire the asset. An operator who purchased a Kelvrac Series unit at a premium during a tight cycle may find the Intelligence's reference set against a cooler market average, offering a figure that does not reflect the acquisition price at all. The ledger records every transaction, but the Intelligence weights its reference toward recent volume, and recent volume has been distressed. That is the arithmetic the backlog is running on.

The Annex was designed to clear the margin — the assets no market session wanted, processed at a measured pace against a Treasury that can mint indefinitely. The Five-to-Fifteen moved far more than the margin into that category, and the queue board's amber marker is the visible result. Whether the open market recovers enough liquidity to drain the Annex, or whether the backlog becomes a permanent feature of the post-Revision economy, is a question the ledger will answer in time. The Intelligence, for its part, has not changed its ceiling.

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.