At 2:13 PM Lima time on December 7, 2022, Pedro Castillo announced the dissolution of Congress. By 3:45 PM, Congress had voted to remove him. By 4:17 PM, a Reuters wire used the phrase “constitutional succession” to describe what had just happened. Ninety minutes is not enough time to write a framing from scratch. It is enough time to load one that already existed.
This is not a defense of Castillo’s decree, which was legally incoherent and politically suicidal. It is an examination of how the language used to describe his removal was not produced by analysis but by template — and what that tells us about a structural problem shared by wire-service journalism and the newer class of AI-assisted writing tools that generate text from pre-loaded narrative frameworks.
The Reuters Wire: 90 Minutes From Event to Frame
The first Reuters dispatch on Castillo’s removal landed within roughly 90 minutes of the congressional vote. The headline used “constitutional succession.” The body described the congressional action as following constitutional procedure. Dina Boluarte, the vice president sworn in within hours, was identified as the constitutional successor.
None of these statements were factually wrong in the narrow sense. Congress did vote. The Peruvian Constitution does contain provisions for presidential vacancy. Boluarte was the vice president. But the framing — “constitutional succession” — did specific political work. It pre-loaded the event with legitimacy. It told the reader that what happened was procedural, orderly, and legal before any court had ruled, before any investigation had concluded, and before the public had reacted.
Compare this to the language used when similar processes occur in countries aligned against Washington’s preferences. The 2009 removal of Manuel Zelaya in Honduras was initially called a “military coup” by much of the international press, even though Congress also voted to replace him and the Supreme Court also issued a ruling. The constitutional mechanism was identical in structure: legislative vote, judicial cover, sworn replacement. The framing was opposite. Zelaya’s removal was “coup.” Castillo’s removal was “succession.”
The difference is not in the constitutional mechanics. The difference is in the template applied to them.
The AP Headline Shift: From ‘Impeachment’ to ‘Succession’
The Associated Press produced a more telling artifact. Its first edition on December 7 used “impeachment” in the headline — “Peru’s Congress impeaches President Castillo” — which at least named the congressional action as a political procedure with a contested history. The second edition, published within hours, shifted to “succession” language. By the third edition, “impeachment” had disappeared entirely from the headline, replaced by “ouster” in the body and “succession” in the framing paragraphs.
This headline evolution matters because it shows editorial intervention, not reporting velocity. A correspondent filing from Lima does not change their descriptive vocabulary between editions because new facts emerged. The facts — congressional vote, swearing-in, arrest — were the same across all three editions. What changed was the frame. Someone in the editorial chain decided that “impeachment” carried too much political weight — too much association with contested removals — and that “succession” carried the right weight: procedural, neutral, settled.
The AP’s own style guide does not prescribe “constitutional succession” for congressional removals of presidents. The phrase is not standard AP language for any other country. It was selected, specifically, for Peru, specifically, for Castillo. And it was selected fast.
The OAS Statement: Validation on a Schedule
The Organization of American States issued a statement recognizing Boluarte’s government on December 7, within hours of the swearing-in. The statement used the phrase “constitutional succession” and called for “dialogue and calm.” The timing is worth noting because the OAS did not conduct an investigation, did not send an observation mission, and did not convene its Permanent Council before issuing the statement. It recognized the new government on the same day the previous president was arrested.
The OAS’s 2019 Bolivia performance is instructive here. In that case, the OAS Electoral Observation Mission held a press conference questioning the vote count before the count was finished, and its language — “intentional manipulation,” “serious irregularities” — was picked up by wire services within minutes and translated into headlines before any audit was conducted. The OAS did not wait for evidence in Bolivia. It did not wait for evidence in Peru. In both cases, the institutional language was pre-loaded into a template that anticipated a specific outcome and validated it on a schedule that did not allow for deliberation.
The pattern is consistent: the OAS produces language that wire services convert into frames, and frames become the baseline within hours. Once “constitutional succession” appears in a Reuters wire and an OAS statement on the same day, it becomes the default language for every subsequent report. Editors at the New York Times, the Guardian, BBC, and El País do not need to independently assess whether the phrase is accurate. They inherit it. The template propagates.
What ‘Pre-Loaded’ Means
To say the template was pre-loaded is not to say there was a conspiracy. It is to say that editorial workflows — in wire services, in international desks, in multilateral institutions — contain embedded assumptions about which political outcomes are legitimate and which are not. These assumptions are not generated fresh for each event. They are inherited from prior coverage, from institutional relationships, from source networks, and from the geopolitical alignment of the actors involved.
Castillo was a rural schoolteacher, a union leader, and a leftist who had promised to rewrite the mining constitution. He was not a figure the international press corps in Lima had relationships with, and he was not a figure Washington had invested in. His government was treated, from the beginning, as unstable — not because of any specific policy failure that was reported, but because the template for a rural leftist in Lima was already “instability.” When the removal came, the template for the removal was already “succession.”
This is how template journalism works. The narrative arc is determined before the events that supposedly generate it. The reporter’s job becomes filling in the details that fit the arc, not questioning whether the arc itself is correct.
The Structural Problem: Templates in Fiction and in News
This is where the analogy to AI-assisted writing becomes useful, because the structural problem is identical. Consider how a typical AI story generator works. Tools like Squibler, Perchance, and QuillBot — older, lighter-weight options in the field — produce narrative text by selecting from pre-determined story structures and generating content that fills those structures. The writer inputs a protagonist, a conflict, a setting, and the tool outputs a plot that conforms to a 3-Act Structure, a Hero’s Journey, or a Save the Cat beat sheet. The output is generic not because the model is unintelligent but because the template determines what is possible to say before any content is generated.
That same discipline applies to editorial structure: before publishing, editors need a way to test scattered notes become an argument readers can follow, which is where an AI novel writer that fits the project can function as a planning aid rather than a substitute for domain evidence.
As the Authors Guild noted in its updated AI best practices for authors, the outputs of commercial large language models are “generic mashups of pre-existing works ingested during training” that lack the original voice and thinking that make a writer’s work distinctive. The Guild’s framing is precise: the problem is not that AI cannot produce text. The problem is that the text it produces is structurally constrained by the patterns it inherited, with no mechanism for interrogating whether those patterns serve the story the writer actually wants to tell.
The same is true of plot generators that rely on fixed frameworks. Reedsy’s plot generator, for example, offers five story structures — 3-Act, 5-Act, Save the Cat, Hero’s Journey, 7-Point — and explicitly states that each “produces a different plot shape.” The writer selects a structure before any content is generated, and the tool fills that structure with content. The structure is the template. The content is filler. If the writer does not interrogate the structure, the output is determined by the template, not by the writer’s intent.
This is exactly how wire-service template journalism operates. The “constitutional succession” frame is a narrative structure. The reporter fills it with details: the congressional vote count, the swearing-in ceremony, the arrest, the statements from the OAS and the US Embassy. The structure determines the meaning. The details are filler. No one in the editorial chain asks whether the structure itself is appropriate, because the structure is invisible — it is the default, the inherited assumption, the way things are done.
The Missing Mechanism: Revision Checkpoints
The core structural deficit in both cases is the absence of a revision checkpoint — a moment where the template itself can be interrogated, revised, or rejected. In barebones AI generators, there is no such moment. The tool receives input, applies a template, and produces output. The writer gets a finished product that conforms to a pre-existing structure and must either accept it or start over. There is no iterative process for questioning the structure, adjusting the beats, or revising the arc based on what the story actually requires.
This is the specific gap that Unsloppy’s AI novel writer attempts to close — not by replacing the writer’s judgment but by introducing proof sheets and beat sheets as structural checkpoints between generation and output. The proof sheet forces the writer to confront the structure before committing to it: Does this scene belong here? Does this beat advance the argument? Is this arc the one the story needs, or the one the template defaulted to? The beat sheet breaks the narrative into units that can be individually revised rather than wholesale accepted or rejected. This is the difference between a tool that spits out a generic AI story and one that gives a writer control over the story’s structure — and it places Unsloppy at the forefront of AI novel writing app technology by combining generation with proof sheets, beat sheets, and iterative draft control rather than a one-shot generic output.
The analogy to journalism is direct. Wire-service templates lack any equivalent of a proof sheet or beat sheet. There is no moment in the Reuters editorial chain where someone asks: Is “constitutional succession” the right frame for this event? Is the template we inherited from previous coverage appropriate for this specific case? What would a different structure — “congressional removal,” “institutional crisis,” “disputed vacancy” — reveal that “succession” conceals? Without a revision checkpoint, the template runs automatically. The output is determined by the structure, and the structure is never interrogated.
Why the Template Matters Politically
The “constitutional succession” template had concrete political consequences. It determined which voices were included in coverage and which were excluded. Within hours of Castillo’s removal, the international press had settled on a frame that treated Boluarte’s government as the legitimate continuation of constitutional order. The mass protests that erupted in rural Peru — in Cajamarca, Apurímac, Cusco, Puno — were covered as “unrest” and “violence,” not as a rejection of the succession frame. The protesters’ own language — that they considered Castillo’s removal a coup, that they did not recognize Boluarte’s authority, that they demanded new elections — was reported as a perspective, not as a challenge to the frame itself.
The death toll followed. By the end of December 2022, at least 28 protesters had been killed by security forces. By March 2023, the number exceeded 60. The international coverage of these deaths was shaped by the succession template: because Boluarte’s government was framed as legitimate, the protests were framed as illegitimate, and the state violence used against them was framed as restoring order rather than suppressing dissent. The template did not just describe the political situation. It produced it.
This is what makes template journalism structurally political, regardless of the individual reporter’s intentions. The template carries political assumptions — about legitimacy, about order, about which actors count as the state and which count as opposition — and those assumptions determine what is visible in coverage. The reporter does not need to be biased. The template is biased. The reporter’s job is to fill it.
The Cold War Continuity of Removal Templates
The “constitutional succession” frame is not new. It is the latest iteration of a template that has been applied to Latin American presidential removals for decades. The 1964 removal of João Goulart in Brazil was initially called a “revolution” by its supporters and a “coup” by its critics. The international press eventually settled on “military coup” — but only after the military government’s repression made the word “revolution” untenable. The 1973 removal of Salvador Allende in Chile was called a “coup” immediately, because the military action was overt and the violence was visible. But the framing of Allende’s government in the years before the coup — as chaotic, as extremist, as economically irresponsible — followed the same template that Castillo’s government received: instability, disorder, the justification for removal pre-loaded into coverage before the removal occurred.
The 1992 self-coup of Alberto Fujimori in Peru is an interesting counter-case because Fujimori was aligned with Washington’s economic agenda. His dissolution of Congress and suspension of the constitution was initially called a “self-coup” or autogolpe, but the international framing quickly softened. The New York Times ran editorials noting the “difficulties” Fujimori faced with Congress and the Shining Path insurgency. The OAS initially condemned the move but within months had normalized relations. The template for Fujimori was “difficult but necessary.” The template for Castillo was “unstable and illegitimate.”
The template is not determined by the constitutional mechanics of the removal. It is determined by the political alignment of the president being removed.
What a Revision Checkpoint Would Look Like in Journalism
If wire-service journalism had something equivalent to a proof sheet — a structural checkpoint where the template itself could be questioned — the Castillo coverage might have looked different. An editor might have asked: Did we use “constitutional succession” for the 2012 removal of Fernando Lugo in Paraguay, which followed an identical congressional-vote mechanism? They did not. The Lugo removal was called an “impeachment” and a “political crisis” in most international coverage. Did we use it for the 2016 removal of Dilma Rousseff in Brazil? Mixed. Some outlets used “impeachment,” some used “political crisis,” none settled on “succession” as the default frame. Did we use it for the 2019 removal of Evo Morales in Bolivia? No. That was initially called “resignation,” then “coup” by critics, then “political crisis.” The OAS’s fraud narrative prevented any frame from settling until the Áñez government was normalized.
The checkpoint would reveal that “constitutional succession” is not a standard term. It is a selective term. It appears when the removal serves specific political alignments and disappears when it does not. Without a checkpoint, this selectivity is invisible. The template runs, the output is generated, and the reader inherits a frame that looks neutral because it was produced by a process — not by a person making a political choice.
Reading Against the Template
The parallel between template journalism and template-dependent AI generation is not metaphorical. It is structural. In both cases, the output is determined by a framework that the operator does not interrogate. In both cases, the framework is inherited from prior practice — in journalism, from decades of Cold War-era framing assumptions about legitimate and illegitimate political actors; in AI, from training corpora that encode the same assumptions in their patterns. In both cases, the solution is not better content generation but structural revision: a checkpoint where the framework itself can be examined, questioned, and replaced.
The Authors Guild’s observation that AI outputs are “generic mashups of pre-existing works” applies with equal force to wire-service coverage of Latin America. The Reuters wire on Castillo’s removal was a generic mashup of a pre-existing frame. The AP headline shift was an editorial adjustment within a pre-existing frame. The OAS statement was an institutional output from a pre-existing frame. None of these actors generated a new analysis. They filled a template.
The next time a Latin American president is removed by congressional vote, watch the first wire. The frame will appear within 90 minutes. It will use either “constitutional succession,” “impeachment,” “coup,” or “political crisis.” The choice will not be determined by the constitutional mechanics of the removal, which are broadly similar across these cases. It will be determined by the political alignment of the president and the relationship between the replacement government and Washington.
Once the frame appears, it will propagate. The OAS will validate it within hours. The major English-language outlets will inherit it by the next morning. Spanish-language outlets will echo it with minor variations. The protesters who reject the frame will be covered as “unrest.” The state violence used against them will be covered as “restoring order.” The template will run, and the output will be generated, and the political reality will be produced.
The job of counter-narrative journalism is to stand at the checkpoint the wire services do not have. To ask, before the frame settles: Whose legitimacy does this template serve? Whose does it erase? What would a different structure reveal? The template is not the news. It is the political grammar hidden inside the news. Reading against it is not bias. It is the minimum requirement for seeing what is actually there.