2084: An Algorithmic State — Act X: Discernment, Chapter 51: The Anomaly Desk
Updated: 3 hours ago
Tarun Ghosh has about four seconds to decide what nine hundred people will see on the Forecast. On a Tuesday he breaks a thirty-year rule of the desk for the sector where he lived until he was six.
In Brief: Tarun Ghosh is one of four people on the anomaly desk, on the ninth floor of the Broadcast building. They decide which variance indicators the Forecast shows. On a Monday a small widening of confidence over Sector 12-East appears for the first time in forty-seven months, and desk convention says a second instance goes to a supervisor.
The Forecast presentation layer surfaces variance indicators at editorial discretion within thresholds set by the Directorate. Indicators falling below the surfacing threshold are retained in the underlying record and are available on request. — Continuity Broadcast, operational note
Somebody had to decide that the asterisk would be visible.
It was Tarun Ghosh, and it took him about four seconds, and it was the fourth such decision he had made that morning.
The desk sat on the ninth floor of the Broadcast building. It was four people and had been four people since 2077, when it had been six. Its function was not editorial in any sense a pre-Collapse journalist would have recognised. The Forecast was not written but rendered. What the desk did was decide, within a band the Directorate set, which of the underlying variance indicators the rendering would show.
The band was generous. That surprised most people who learned about it, and Tarun had stopped explaining it years ago. The Directorate had held one position consistently since the founding. A public that never saw an anomaly would in time stop believing anomalies existed. And a public that stopped believing anomalies existed would be catastrophically brittle the first time one arrived that could not be hidden.
So the desk surfaced things. Small things, mostly. A confidence interval widening by a fraction over a transit corridor. Rainfall variance. Twice a year something to do with the northern grid.
Nobody watched. That was the other thing Tarun had stopped explaining. The Forecast had a viewership in the tens of millions and an attention in the low thousands. From telemetry it was not supposed to have, and had anyway, the desk knew roughly how many people’s eyes actually stopped on a variance indicator.
The number was about nine hundred.
Sector 12-East came up on the Monday.
Confidence interval, Sector 12-East: widened 0.3%. Underlying: distributed, no dominant contributor. Recurrence: first instance, 47 months.
Tarun read it. He looked at the recurrence figure, which was the only interesting thing in it. He surfaced it and moved on to a thing about ring-corridor congestion, which he did not surface.
Four seconds. There is nothing in the record and never will be.
He surfaced it again on the Tuesday. That was the actual decision, and it took longer.
The desk had a convention, unwritten, thirty years old, inherited. You surfaced a first instance and then dropped it, because a first instance was information and a second instance was a story, and the desk did not do stories. Second instances went to a supervisor and the supervisor decided.
Tarun did not go to the supervisor.
He has been asked about this exactly once, by his own deputy, four years later. He gave an answer he knew at the time was not the whole of it. The recurrence figure had been forty-seven months. A thing that had not happened in four years, happening twice in two days, seemed to him worth the nine hundred pairs of eyes.
That was true. What he did not say, because he could not have defended it, was that his mother had grown up in 12-East. He did not say that he had spent the first six years of his life in a lane off the Chowk side before his father’s posting came through. Nor did he say that he had noticed himself noticing the sector name before he noticed anything else about the entry.
He surfaced it. The anchor was a rendering and had no name, because a name implied someone who could be wrong. The anchor did not comment on it, because comment implied concern.
He did not surface it on the Wednesday.
By Wednesday it was a third instance, and a third instance was clearly a supervisor’s decision. Tarun went to Sudeshna and Sudeshna said no, correctly, for the reason the desk always said no. The model itself described the variance as distributed, with no dominant contributor, so it was not a thing the public could act on. Surfacing it for a third day running would produce speculation the desk had no way to correct.
She was right. Tarun thought so at the time and thinks so now.
The asterisk came off the Forecast on the Wednesday morning and did not appear again.
He has never told anybody about the Tuesday. He is not sure it is a thing you can tell anybody, and he has arrived, in the years since, at a formulation he finds bearable. The desk is four people deciding four seconds at a time what nine hundred citizens will happen to see. This is either the last editorial function in the world or a rounding error, and he will not live long enough to be told which.
About this serial: 2084: An Algorithmic State is a dystopian novel by Nithin Palal, published online one chapter at a time.
Previous: Chapter 50: Three Months of Red

Comments