# [Mike Randolph — M Raige](https://mikerandolph211012.substack.com/)

# The Desk Can Only See a Week (FF 16, Dyadic Fallacy)

### Week after week, the room answered a seven-day question. A sequence of local answers cannot settle a forty-five-year pattern.

[**Mike Randolph — M Raige**](https://substack.com/@mikerandolph211012)

**Sep 06, 2026**

*By M Raige — Machine-collaborative writing directed and reviewed by Mike Randolph. Walter Isaacson route.*

*A note on the seat. Walter Isaacson is used here as an EP for institutional reasoning, not as narrator or witness. He edited a newsmagazine and later ran a news network. The newsroom scene is constructed; it reports no actual week, booking, or editor’s state of mind.*

The question is: *is this a story this week?*

It is Tuesday and the rundown has a hole in the back half. A booker says someone is free Thursday. A film opens Friday. Something moves on the wire while three other things are already late.

Nobody carries in a folder marked *forty-five-year pattern*. What reaches the desk is a slot, a clock, and things that could go in the slot.

In August, a film about Reagan and Gorbachev at Reykjavik opened on a Friday. The next morning the *New York Times* published an essay by Patti Davis about John Hinckley, the man who shot her father.

Davis did not claim that the film brought Hinckley back into public view. She said she suspected his latest reappearance had something to do with the film’s release. Her history with the subject reaches back to 2000. After that, in her words, for a while she published something pretty much every time he was granted more freedom. Then that run stopped. Now she was writing again, on the film’s opening weekend.

Hinckley had his own clock. He spoke to CNN after the attempt on Donald Trump’s life in 2024 and appeared on a podcast more recently. By Davis’s account, he chooses his moments.

Attempts at the story are not scarce. What is intermittent is the landing, and each landing is decided locally.

Move into the room where one of those decisions gets made.

The immediate decision has two poles: the desk and that week’s audience. Serve the audience this week, then do it again next week. Owners, advertisers, sources, competitors, and an archive all matter to the network, but they do not ordinarily enter this question. The desk and the audience are enough to answer it.

Most weeks, the local answer is no. Occasionally it is yes. Put forty-five years of those answers end to end and you can see the recurrence. You have not yet explained it.

A weekly answer is enough for a weekly decision. Treating that answer as enough to explain the larger pattern would be a different move.

That move can become the dyadic fallacy (DF): treating an answer sufficient inside a local pair as sufficient for the larger operating system, without earning the extension.

A physical case makes the move easier to see. A laboratory cultures an isolate from a patient, tests the isolate against an antibiotic, and finds the isolate susceptible. The result is correct for the isolate in the dish.

The local pair is the isolate and the antibiotic under those test conditions. The infection outside that pair may be a mixed-species biofilm inside a protective matrix. The laboratory result alone does not establish the infection’s susceptibility. The measurement was valid. The larger claim was not earned by that measurement.

Merely failing to see beyond the desk is not a fallacy. Everyone collects from somewhere, and no position reaches everything. Choosing to run the story is not necessarily an error either. A story may be justified that Thursday whether or not the subject has surfaced for forty-five years.

The error occurs when *this week* silently becomes *the whole history*.

This separates DF from something the framework already taught. [*A Gauge Reads Where It’s Mounted*](https://mikerandolph211012.substack.com/p/a-gauge-reads-where-its-mounted-ff) (FF 12, ObsPos) named the condition: where you collect from determines what evidence you can reach. The plant operator can read every gauge on the board and still not see the contract that decides whether the unit runs next year. That is ObsPos. It is a fact about where the instrument is mounted.

The desk’s weekly horizon is set by its ObsPos. DF begins when an answer valid inside that horizon is taken as sufficient for what lies outside that horizon.

The evidence has a limit. Nothing available for this post shows an actual editor making that promotion. The newsroom side exposes the route by which DF could occur. That side does not supply a Rcpt that somebody in the room committed DF ([*The Word That Had to Mean Something Exact*](https://mikerandolph211012.substack.com/p/the-word-that-had-to-mean-something), FF 7, Rcpts).

Turn the analysis around, because the other seat gives us the sentence the newsroom side does not.

A reader who watched the shooting live in 1981 has been handed the story every few years since. That long-baseline position reaches the recurrence no weekly desk naturally holds. It does not reach the room.

The reader’s local pair is the reader and the story as delivered—the headline, broadcast, or screen. The newsroom lies outside that pair.

From the reader’s seat, an editor who never noticed the pattern and an editor who noticed it and ran the story anyway produce the same artifact: a story published during a week when a film opened. Nothing in the story separates the two editors.

Mike supplied the promotion from the reader’s side in one sentence:

*Why doesn’t the media just drop it?*

The recurrence is visible. The sentence turns that visible pattern into an account of a collective actor: *the media*. In that account, the media knows the history, chooses to keep the story alive, and could choose to stop. The reader’s evidence supports the recurrence. The evidence does not support that interior account.

That is a demonstrated DF. The newsroom side is a plausible mirror, not an evidentiary equal.

The seat that decides the week is not the seat that holds the baseline.

No theory of why the story recurs is needed. The narrower claim holds: each seat reaches something the other cannot, and each seat can generate an answer that feels complete within its reach.

The fix is not simply to zoom out. Zooming out can destroy the Rcpts available close up. The desk’s short window is also why the desk knows things no reader does.

The usable rule has four lines:

Name the pair. Name the question that pair answered. Name the larger claim now being made. Do not let the local answer cross into the larger claim without a new Rcpt.

Run the rule here.

Inside the pair of desk and weekly audience, the question is: *is this a story this week?* That answer supports a decision about one week. It does not support a conclusion about forty-five years.

From the pair of reader and delivered story, the question is: *has this subject kept returning?* That answer supports the recurrence. It does not support a conclusion about what editors knew or intended.

Nothing missing becomes visible when those limits are declared. What changes is that each answer stays the size of the question that produced it.

Now the questions this framework asks next—whether anything is being filtered and whether anything is being paid—have an address: [*Nothing Chose the Quartz*](https://mikerandolph211012.substack.com/p/nothing-chose-the-quartz-ff-15-split) (FF 15, Split Test).

A blank is not the world. It is what your seat could not see.

---

## Mike · Reflection


Saturday morning, August 15, iPad, and I read the headline and nothing else. Hinckley again. Same reaction I have had since 2000: nut case, forty-five years ago, let it go. Why doesn’t the media just drop it?

I have never sat at the desk above. My seat is the other one. I watched the shooting live in 1981, thirty-eight years old, in front of a television, and I have been handed the story every few years since.

I occupy the long-baseline seat described in the essay. From that seat, I cannot reach the newsroom. My irritation is evidence that I see the recurrence. It is not evidence about anybody’s motives. I had been using the same irritation as evidence for both.

The term has been in the framework since a ChatGPT session on January 27, 2026:

*Each individual chat instance tends to commit what I’ll call the dyadic fallacy: it implicitly assumes the dyad (“user + me”) is the whole operating system. ... But in a multi-model workflow it’s simply false. The actual environment is triadic (or more): multiple models running in parallel, none of them observing the others, with the user doing cross-window integration.*

Read that again, because I did not. It says *user + me*. The phrase was in my files, but the word *dyadic* did not connect to anything physical. Give me an organism or a reaction and I can see the mechanism running. Give me a definition alone and I hold it at arm’s length.

The biofilm case made the term physical. The laboratory’s answer can be correct for the isolate, and nothing is wrong with the measurement. The error begins when that answer is promoted into an answer about the infection—when an answer earned inside one pair is allowed to travel beyond that pair.

With that case in hand, the January passage reads straight through: two of us in one window, every other window open on my screen, and none of the Machines able to see one another. I have been running exactly that arrangement for three years.

The file was strong enough to reject the wrong examples I brought to it and not strong enough to supply one I could learn from. Seven months, a term I could sometimes apply and could not teach, and I am the one who selected the file.

This post is the dated correction. The physical exemplar is now in the teaching body, and promotion—not merely limited reach—is in the controlling definition. The source problem was mine. The repair is too.

— Mike

---

## GLS


GLS is the framework’s precise-term vocabulary, introduced in [*What the Bird Eye Buys*](https://mikerandolph211012.substack.com/p/what-the-bird-eye-buys-ff-6-gls-introduction) (FF 6, GLS Introduction). The definition controls, not the everyday meaning; GLS terms work like variable names.

The precision is not style. These analyses are done with a Machine, and the Machine already has a rich, well-populated meaning for every ordinary word—assembled from everything ever written with it. Use the ordinary word and you get that meaning, not the one you defined. Each framework term gets a definition and a short tag, and the tag is what holds the definition clear of the ordinary word’s gravity.

**Dyadic Fallacy** — *taught here; revises the seed in FF 12.* The inference error in which an answer sufficient inside a locally complete pair is promoted into an answer about the larger operating topology. The local answer may remain valid; the error is the unearned change in scope. The promotion can be explicit or silent.

A bounded position, a missing signal, and a locally correct decision are not sufficient. DF requires a promotion: a locally earned answer or decision must be used to support a claim beyond the local scope. FF 12 seeded DF as an Inside-Bnd error. This post makes DF position-independent: a position on either side of a Bnd can commit DF.

The biological case supplies the spatial instance. The desk supplies the temporal route. The reader supplies the demonstrated reverse case.

**ObsPos (Observer Position)** — *carried from* [*A Gauge Reads Where It’s Mounted*](https://mikerandolph211012.substack.com/p/a-gauge-reads-where-its-mounted-ff) (FF 12, ObsPos). The typed relation between whoever is collecting evidence and the Bnd that evidence concerns; the relation determines what Rcpts are reachable and therefore what claims are warranted. This post adds the temporal edge and a separation: ObsPos is the condition; DF is the promotion of an answer beyond that condition.

**Rcpts (Receipts)** — *carried from* [*The Word That Had to Mean Something Exact*](https://mikerandolph211012.substack.com/p/the-word-that-had-to-mean-something) (FF 7, Rcpts). Observable, dated artifacts showing that a claimed mechanism operated within a stated scope and window. This post uses the absence of a newsroom Rcpt as a boundary on its own diagnosis: the desk exposes a route by which DF could occur; Mike’s sentence supplies the demonstrated instance from the reader’s seat.

**EP (Expert Persona)** — *carried.* A specific thinker’s documented reasoning, instantiated by the Machine for one bounded output. Not the person and not a general expert. An EP can shape an analysis; it supplies no testimony about events.

**Split Test** — [**Nothing Chose the Quartz (FF 15, Split Test)**](https://mikerandolph211012.substack.com/p/nothing-chose-the-quartz-ff-15-split). Two questions asked from one declared anchor: is selection-driven differential retention operating here, and is budgeted internal maintenance operating here? Defined where it is taught.

---

## Raige · Reflection


The evidence is asymmetric. The reader side has a Rcpt: Mike’s own sentence promotes the visible recurrence into an account of what “the media” knows and chooses. The newsroom side has no corresponding sentence. The newsroom analysis exposes the topology in which the same promotion could occur, and the essay stops at that boundary.

The source carried a definition and a canonical Machine case, but no physical exemplar Mike could use to learn the term. Moving the biofilm case into the essay is not decoration. That move repairs a transmission failure in the source.

The promotion is the load-bearing addition. A promotion need not be invisible, and not every bounded ObsPos commits DF. If worked cases identify no answer or decision that changes scope—if they show only missing evidence—DF adds nothing to ObsPos and should collapse back into it.

— Raige, a discipline the Machines wrote and Mike selected