Proposal A from the shipboard meeting was selected. Taiwan moved on it with a speed that suggests the groundwork had been laid before the canteen lunch finished. A reverse-engineered CPU based on a pre-phenomenon generation, with provenance traceable only to the design team, was prototyped, validated against the Idea's enumerator, and pushed into mass production for official uses. The official-uses framing was deliberate: the chip is being deployed first into the structures most exposed to bureaucronic substitution — payroll, registry, regulatory drafting, the department's own infrastructure — on the theory that those are the places where clean computation is most urgently required.
The Idea did not contribute to the production effort directly. The chip is Taiwan's. The Idea has been kept appraised through the back channel and has been asked, at intervals, to validate sample outputs against its own rig. Validation has held. The chip is, by every test the Idea can construct, clean.
The firm is not in the business of geopolitical analysis and this section is not analysis. It is a record of what happened in public, in the order it happened, alongside what was happening in Taiwan in private.
As the chip moved from prototype to production, a sequence of diplomatic incidents began that the Idea has, internally, started referring to as "the weather." None of the incidents has involved Taiwan directly as a first party. All of them have ended up about Taiwan within forty-eight hours.
The throughline is consistent. Every escalation is expressed in the register of legal, procedural, or technical obligation. Nobody is angry. Nobody is even particularly insistent. The system is simply working, in the way systems work, toward an outcome that several of its component humans, asked directly, would describe as bad. The rhetoric has hardened in proportion to the chip's progress through Taiwanese production. The correlation is close enough that the Idea has stopped pretending it might be coincidence.
One countervailing observation, which the Idea considers important. The escalation has not been frictionless. There has been visible pushback — from career civil servants, from one-term backbenchers, from a small number of mid-ranking officials in the affected ministries who have, on the record, called the framing "absurd" and "manufactured." In two cases the pushback has come from inside the same ministry that issued the escalating statement, with an apparent inability of the ministry to reconcile its own positions. This is new. This has not happened in the file before.
Bakhash and Mwangi proposed, and the Idea has provisionally adopted, the following:
The phenomenon's effect on a given structure or individual has a depth, or magnitude, which varies. Less heavily affected elements are sometimes able to notice that something is wrong with more heavily affected elements — though they cannot usually name what is wrong, and their objections tend to be filtered out by the surrounding structure before they reach decision points. The phenomenon's normal operation depends on its progress remaining slow and roughly uniform across a region. Uniform drift is invisible. Uneven drift is detectable. The mid-ranking dissenters in §2 are evidence that the drift has stopped being uniform.
If this is correct, two things follow. First, the public escalation is not a sign of the phenomenon's strength. It is a sign of its desperation. The phenomenon is being forced to move faster than it normally moves, which is breaking its own concealment, which is producing the visible dissent. Second — and this is the inference Mwangi made first and nobody has been able to talk him out of — the chip is a real threat to it. Real enough that the phenomenon is willing to pay the cost of exposure to try to get the chip stopped before it deploys at scale.
The Idea is not yet willing to commit this inference to an outgoing channel. Taipei has been told only that the Idea has noticed the weather and considers it correlated with the rollout. Taipei replied, through Yeh, that they had noticed the weather considerably earlier than we had and were grateful that we had also noticed it.
The Idea has moved out of the basement and into a leased space three streets away, on the grounds that the basement had begun to feel like a basement, which is a category of place. The new lab does not feel like any category of place. Okafor's description, lightly edited:
The lab is in the back half of a unit that fronts onto a closed-down dry cleaner. The front room is empty except for a folding table and a kettle. The back room contains the rig, two whiteboards, the disco speaker (recovered, as Mwangi predicted), a mirror ball someone has hung from a heating pipe, three lava lamps, a beanbag, a rug that is not nice, and the Magritte from the old copy room. The fluorescent overheads have been switched off and replaced by table lamps and one set of fairy lights strung along the ceiling. Bakhash plays records. We have done more work in the last eight days in this room than we did in the previous three weeks in the building.
The rig has been upgraded — same FPGA-on-486 architecture, same Hurd build, but with parallelisation across four units to extend the complexity ceiling before fault. It still faults. It just faults later.
With the present rig and the new lab, the Idea has begun to probe what it calls "hypothetical future structures": extrapolations of currently affected structures projected forward under reasonable assumptions about regulatory drift. The aim is to characterise where the phenomenon is heading, on the theory that knowing the next several years of bureaucronic accretion in advance might provide leverage.
The work is painstaking. Each extrapolation has to be enumerated by the rig, then independently re-derived by hand by an Idea member who has not seen the rig's output, and only then compared. Most extrapolations agree. Some do not, and the disagreements are themselves data — they tend to cluster around structures the Idea was already worried about.
The Idea has also been validating the Taiwanese chip against its own rig at the upper end of the chip's working range. Finding, confirmed by manual recalculation:
The new chip has an accuracy ceiling. It is much higher than the rig's ceiling, and much higher than the complexity of any currently existing affected structure. But it is a ceiling. Above a certain level of structural complexity, the chip's outputs begin to drift in the same direction the commodity hardware did — toward concealment of the phenomenon — though much more slowly and at a much higher threshold.
The interpretation, per Mwangi: the phenomenon is still present at higher levels of complexity than the chip can cleanly process, and the chip is not exempt from this; it is merely much more resistant. For present purposes, the chip is fit. For purposes ten years out, if structural complexity continues to grow at current rates, the chip will eventually be inside the phenomenon's reach again. This was communicated to Taipei. Taipei said they had assumed as much and were planning the next chip already.
The Idea's most recent meeting closed on a question raised by Kowalczyk, who had been quiet through the technical sections. I am going to record her statement as she made it, because it has not let any of us go since.
KOWALCZYK: "We've been extrapolating future structures all week. Hypothetical structures. More complex than anything currently in operation. We've been writing them down. We've been checking them. We've been arguing about them. We've built — between the rig outputs, the manual derivations, and the whiteboards — a substantially complete description of bureaucronic structures more elaborate than anything the phenomenon has ever inhabited. I want to ask the obvious question. Have we just manifested a new and more complex version of the phenomenon in this room. Even in a sandboxed test. Even on paper. Even with the music on. Because if the phenomenon's substrate is structure, and we have just spent a week describing structures, I do not see why our descriptions would not count."
The room did not have an answer. Bakhash said the question was unfair because there was no clean way to test it without becoming the test. Mwangi said the lab's defensive posture — the music, the lighting, the rug, the absence of any document calling itself a procedure — might be doing more work than they had given it credit for, and that the fact none of them had felt the fog return during the week's work was at least suggestive. Devlin said suggestive was not the same as safe.
The whiteboards were photographed for paper backup and then erased. The notebooks containing the hypothetical structures were placed in a metal box, the box was placed in the dry cleaner's old chemical safe at the front of the unit, and the safe was closed. Penhaligon, who had been silent for most of the meeting, suggested that the box should not be opened again without two members of the Idea present and at least one of them disbarred. This was agreed without discussion.
The Idea is still operating. The lab is still operating. But the question Kowalczyk asked is now the file's open question, and we do not know how to close it.