This series began with a manuscript problem. The publishing market is flooded with synthetic content, the mechanisms that once filtered signal from noise have collapsed, and the industry is reaching for detection tools calibrated to the wrong question. Posts 2 and 4 argued that the right question is not whether a text was generated by AI but whether a mind was present and accountable throughout its making. Post 3 described the market reorganization following from the trust collapse, and who it rewards. Post 4 introduced structural analysis as the instrument the market needs but has not yet built into its accountability infrastructure.

This post moves to a different scale. The manuscript problem is real and addressable. The problem this post describes is larger, less visible, and considerably harder to solve, because it does not resolve at the level of individual works. It resolves, if it resolves at all, at the level of governance.

Return to the cognitive ecosystem described in Post 3.

A reader who has spent years inside an author’s newsletter, podcast, and books has done something that looks, from the outside, like enthusiastic consumption. From the inside, something more consequential has happened. The author’s interpretive framework has become part of the reader’s own cognitive apparatus. They are not consciously aware of this. Nobody decides to outsource part of their judgment to a writer they admire. It happens incrementally, below the threshold of deliberate choice, through the accumulated weight of repeated contact with a coherent and trusted worldview.

This is not a new phenomenon. It has always happened with influential writers, teachers, and thinkers. What is new is the scale at which it can now happen, the infrastructure that accelerates and deepens it, and the degree to which that infrastructure is moving beyond passive content consumption into something that more closely resembles an ongoing cognitive relationship.

The infrastructure I am describing is already assembling itself without anyone having planned it. Author newsletters create direct, regular contact outside of algorithmic mediation. Podcasts deliver voice and personality, which carry trust signals that text alone does not. Books provide the sustained, deep engagement that builds genuine intellectual relationship. And now the potential for AI systems trained on an author’s corpus and voice that are beginning to appear as extensions of that presence: tools that let readers interact with a version of the author’s thinking in real time, ask questions, receive responses shaped by that author’s accumulated work and perspective.

Each of these components is individually unremarkable. Together, they constitute something without historical precedent: a persistent, scalable, interactive cognitive presence that operates continuously in the reader’s intellectual environment.

The governance questions that follow from this are not hypothetical. They are already live, even if the industry has not yet named them as governance questions.

Who is accountable for the integrity of a cognitive ecosystem built around an author’s identity? The author, presumably, bears some responsibility for the content they produce and the persona they project. But the ecosystem extends well beyond content the author directly controls. Newsletter archives get fed into AI training sets without the author’s involvement in that process. Podcast transcripts, interview excerpts, social media posts, reader forum discussions all become raw material for systems that then represent themselves as extensions of the author’s voice. The author’s identity becomes infrastructure, and infrastructure has stakeholders and failure modes that the person at the center of it may not be positioned to monitor or correct.

What happens when that infrastructure fails? An AI system trained on an author’s work can produce outputs the author would explicitly reject. It can extrapolate positions the author has never held, apply the author’s characteristic reasoning to questions the author has never addressed, and do so with enough surface plausibility to be convincing to readers who trust the author’s voice. The author’s cognitive presence, in other words, can be extended in directions the author did not choose and would not endorse, and readers inside the ecosystem may have no mechanism for detecting the difference.

What obligations attach to operating a cognitive ecosystem at scale? A writer with ten thousand newsletter subscribers who posts twice a week is exercising meaningful influence over how ten thousand people encounter and process new information. A writer with a hundred thousand subscribers, a podcast, and an AI companion trained on their work is something categorically different. They are operating a cognitive infrastructure. The ethical obligations that attach to that scale of influence are not the same as the obligations that attach to publishing a book.

These questions do not have established answers. The legal frameworks governing them are embryonic at best. The cultural norms that might constrain behavior in this space have not yet formed. The industry is focused on the volume problem. The governance problem is developing underneath it, largely unobserved.

There is a structural reason the governance problem is hard to see from inside the publishing conversation.

Publishing has always understood itself as a content industry. It produces objects: books, articles, stories. The ethical and legal frameworks it operates within were built for an industry that produces objects. Copyright protects specific expressions. Defamation law governs false statements of fact. Contract law governs the relationships between authors, publishers, and distributors. These frameworks are object-oriented. They assume a discrete work that can be examined, attributed, and evaluated.

Cognitive ecosystems are not objects. They are environments. An environment does not have a publication date or a copyright registration. It does not have a clearly bounded expression that can be evaluated for accuracy or attributed to a specific creative act. It operates continuously, it evolves, and its effects on the people inside it accumulate over time in ways that are difficult to observe and nearly impossible to attribute to any specific component.

The governance frameworks adequate to environments of this kind do not yet exist in publishing. They are being built, slowly and inadequately, in adjacent domains: social media regulation, algorithmic accountability, AI governance. The AI governance conversation in particular is directly relevant, because the extension of author identity through AI systems is precisely the kind of scalable cognitive influence that AI governance frameworks are attempting to address.

This is where the Charter argument that runs underneath this entire series becomes directly relevant to the publishing conversation.

The AI/Human Charter is a constraint-first governance framework built on a specific thesis: that human cognitive fragility is the primary AI risk vector, and that adequate governance must be built into system architecture before capability scales beyond the point where constraint is practical.

That thesis was developed in the context of large-scale AI systems. It applies with equal force to the smaller-scale cognitive ecosystems assembling around author identities, because the mechanism is the same. Influence operating below the threshold of conscious awareness, accumulating incrementally, reshaping the cognitive environment of the people inside it, without those people having meaningfully consented to that reshaping or having any mechanism for evaluating it.

The reader who has spent three years inside an author’s ecosystem did not consent to having their interpretive framework modified. They chose to read a newsletter. They chose to listen to a podcast. Each individual choice was voluntary and informed. The aggregate effect of those choices, the installation of a persistent cognitive lens that now filters how they encounter new information, was not something they chose or could have chosen, because it was not visible as a choice at the time it was happening.

That is the governance gap. Not the flood of synthetic content, which is a visibility problem the market will eventually develop tools to address. The governance gap is the accumulating, invisible influence of cognitive ecosystems on the epistemic autonomy of the readers inside them, operating at a scale the publishing industry has never had to govern before and has no frameworks adequate to govern now.

What would adequate governance look like? Beyond this point I am speculating, but the direction is clear enough to name.

Transparency obligations are the minimum threshold. Readers inside a cognitive ecosystem have a legitimate interest in knowing the extent and nature of the infrastructure they are engaging with. When an AI system represents itself as an extension of an author’s voice, the reader should know what that system was trained on, what its limitations are, and in what ways it might diverge from the author’s actual positions. This is not a radical requirement. It is the cognitive equivalent of nutritional labeling: not a guarantee of quality, but a precondition for informed choice.

Accountability structures are the next layer. The author whose identity anchors a cognitive ecosystem bears meaningful responsibility for the integrity of that ecosystem, including the AI-extended components they may not have directly authored. That responsibility needs to be acknowledged and, eventually, formalized. The current situation, in which an author’s identity can be extended indefinitely by AI systems with no accountability mechanism attached, is not a stable equilibrium. It will produce failures that are visible and damaging enough to force a response. The question is whether that response is designed in advance or improvised in crisis.

Consent architecture is the hardest layer and the most important. The gradual, incremental nature of ecosystem formation means that meaningful consent cannot be obtained at a single point. It must be ongoing, revisable, and genuinely informed. Readers should have mechanisms for understanding what they are inside, evaluating whether they want to remain inside it, and exiting it cleanly if they choose. None of those mechanisms currently exist in any systematic form.

This series began with the practical problems of working authors: visibility, discoverability, the economics of presence, the question of what trust signals can survive the synthetic content flood. It ends somewhere those authors may not have expected to arrive: in the territory of epistemic autonomy and cognitive governance.

The connection is direct. The same forces that are destroying mid-list discoverability are building the conditions for cognitive ecosystems that will eventually require governance frameworks the industry does not have. The manuscript problem and the governance problem are not separate issues. They are the same disruption observed at different scales.

Structural analysis, the instrument Post 4 described, addresses the manuscript scale. The governance frameworks the Charter argument points toward address the ecosystem scale. Neither is sufficient alone. Together, they constitute the beginning of an adequate response to a disruption that the publishing industry is currently treating as a volume problem, when it is, at its root, a problem about the relationship between human cognition and the environments that shape it.

That is a problem authors are unusually positioned to understand. It lives at the intersection of storytelling, psychology, and the ethics of influence, which is territory serious writers have always occupied. The question is whether the industry will recognize it in time to participate in designing the response, or whether the response will be designed by people who have never thought seriously about what it means to build a world inside a reader’s mind.

E.S. Martell is a cognitive psychologist, AI governance researcher, and science fiction author. He publishes the After Biology Substack and runs Second Initiative Press. FictionMark.com is his AI-powered manuscript analysis service. The AI/Human Charter, his constraint-first governance framework for AI systems, is developed at length in The Governed Mind. The Governed Mind, The Amplified Mind, and The Synthesized Mind form his AI governance trilogy, available through Second Initiative Press.

Share

Subscribe now

Every writer has blind spots. The work is too close, too familiar, too inhabited to see clearly from the outside. You know what you meant to put on the page, and that knowledge fills in the gaps that a fresh reader encounters as absence. This is not a failure of craft. It is a structural feature of the creative process, and it is why outside perspective matters.

I recently ran a completed short story through an AI analysis before publication. The story, The Pack, is a quiet horror piece set in a Florida retirement community where an elderly man begins to suspect that the coyotes howling in the woods behind his neighborhood are responsible for a series of deaths among his neighbors. The horror is slow-burn and domestic, the kind that works by accumulating unease rather than delivering shocks. The AI analysis identified two places where the story was telling rather than showing, where I had gestured at threat rather than building it on the page.

The changes I made in response were minimal in word count. Their effect on the story was not.

The first problem was a surveillance scene near the climax. In the original, Hawk watches coyotes moving through his neighbor’s yard, feels spooked, and retreats quickly to his house. Here is the original passage:

More eyes blazed from the hedge. Hawk suddenly wondered if they could see him. A chill went down his back, and the hair on his neck raised. Humans weren’t in danger from coyotes. Were they? He thought not, but it was eerie, and he felt vulnerable standing in the front yard with his pajamas on. He quickly moved back to the porch. His movement seemed to trigger a feeling of fear. He jerked the door open and practically slammed it. He clicked the lock home, then leaned against the cool door momentarily, trying to regain his mental equilibrium before returning to Jess’s inevitable questions.

The unease is there, but it’s managed rather than experienced. The reader is told Hawk feels vulnerable and afraid. The coyotes are shadowy presences rather than specific threats. The analysis identified that the scene needed to earn its dread rather than assert it.

The revised version extends the scene considerably. Hawk moves toward the neighbor’s house instead of retreating, watches a coyote standing on its hind legs working the doorknob with misshapen paws, is spotted, backs away through the darkness, reaches his own porch, hears scratching on the other side of his front door, and retrieves a knife from the kitchen before the sounds stop. Here is how the revised passage closes:

His heart gradually slowed. He took a minute more, there against the solid metal, trying to regain his mental equilibrium, but then jumped at a slight scratching sound on the other side of the door. His heart rate jumped up. What if they’d followed him? He stood poised, indecisive, then stepped into the kitchen and removed a heavy knife from the set in the wood block. There were no additional sounds, either from the door or from anywhere else in the house. It was so quiet he could hear the kitchen clock ticking out its steady count, seeming to cue his heartbeat to slow down. A minute passed, then two. He put the knife back and took several deep breaths, trying to calm himself before returning to Jess’s inevitable questions.

The same closing line. Completely different emotional weight behind it. The reader has now experienced the threat rather than been informed of it.

The second problem was a confrontation scene at the pool. In the original, a coyote gives Hawk a warning stare, and he stumbles backward into the pool. It reads as accidental comedy arriving at the wrong moment. The revision inserts a passage where Hawk, a former baseball pitcher, picks up a piece of stone mulch and throws it at the animal. The stone wobbles and lands short. The coyote doesn’t flinch. It takes two steps toward him. He retreats into the pool enclosure as the animal charges. The pool fall that follows is now a consequence of active confrontation rather than a clumsy accident, and it lands with genuine menace.

The morning after, a detail added to the revised version: Hawk discovers a jagged tear in the lower screen panel of the pool enclosure. Not large, but new. Physical evidence that confirms what he witnessed was real. That detail costs four sentences and transforms the story’s relationship to its own ambiguity. The reader no longer has the option of deciding Hawk imagined everything.

There are smaller changes throughout, a word here, a phrase there, but these two structural additions are where the story changed in kind rather than degree.

What the analysis identified, and what I had been too close to the work to see, was that I had written the architecture of a horror story without fully committing to its consequences. The threat was present but kept at a safe narrative distance. The revisions closed that distance. They required Hawk to move toward the danger rather than away from it, to act rather than observe, and to find physical evidence that the world of the story had permanently changed. Each of those is a craft principle that I know and advocate. None of them were visible to me in my own work until an outside perspective named the absence.

This is what AI analysis can do when used as a diagnostic rather than a replacement. It does not write the story. It identifies the gap between what the writer intended and what arrived on the page. The writer still makes every decision about what to do with that information. In this case, the decisions were mine, the voice remained mine, and the story is materially better for the process.

The full story is available on my Substack here. Read the revised version, and trust that the one before it was the rougher draft.

Share

The previous three posts in this series established a sequence. The publishing market is not suffering a volume problem. It is suffering from a cognition verification problem. The mechanisms that once connected readers to trustworthy work have collapsed, and the reorganization filling that vacuum rewards incumbents and personalities rather than writers. The cultural loss is specific and ongoing. What the market cannot do on its own is rebuild the accountability infrastructure that made trust signals possible.

That leaves a question the market is not yet asking clearly: What does cognition verification actually look like at the level of the manuscript?

This post examines how to answer it.

Let’s start with what could be easily mistaken for verification.

AI detection tools currently operate at the prose level. They examine sentence structure, token probability distributions, stylistic consistency, and a range of linguistic features that differ statistically between human- and machine-generated text. Some of them are reasonably accurate under controlled conditions. None of them is reliable enough for consequential decisions, as the false positive problem documented in the Epoch Times piece illustrates. An author whose legitimate work triggers a detection tool faces reputational damage with no meaningful recourse because the tool measures the wrong thing.

The wrong thing is style. Style is a surface property. It is also increasingly a learnable property for language models. The gap between human and AI prose at the stylistic level is narrowing and will continue to narrow. Any verification system built on stylistic detection is built on ground that is actively eroding beneath it.

The right thing to measure is structure. Not because structure is harder to fake, though it is, but because structure is where authorial cognition actually lives. A novel’s architecture is the record of a mind making hundreds of interdependent decisions across tens of thousands of words, each one constrained by what came before and constraining what comes after. That record is not stylistic. It is cognitive. And it is legible, if you know how to read it.

Post 2 described the structural failure modes of AI-generated fiction in clinical terms. This post takes the inverse position: what structural properties constitute positive evidence of sustained authorial cognition?

There are several, and they cluster around a single underlying phenomenon. Call it narrative intentionality: the degree to which every significant element of a manuscript is in a purposeful relationship with every other significant element, governed by a coherent authorial intention that persists across the entire work.

Narrative intentionality is not the same as technical competence. A manuscript can be competently written at every local level and still lack it. It is also not the same as thematic consistency, which can be achieved by repeating motifs without genuine developmental logic. Narrative intentionality is closer to what experienced editors mean when they say a novel knows what it is. Every scene is doing more than one thing. Character decisions carry consequences that ripple forward. The ending was always the ending, even if the reader couldn’t see it coming. The work has a governing intelligence behind it that was present from the first page to the last.

This is detectable. Not with certainty, and not by algorithm alone, but with the kind of structured analytical framework that a trained reader applies when evaluating a manuscript at depth. The framework examines specific properties: the load-bearing relationship between scenes, the developmental consistency of character psychology across dramatic pressure points, the degree to which thematic content is embodied in action and consequence rather than stated in dialogue or narration, the structural integrity of the second half relative to promises made in the first, and the coherence of the ending as a destination the manuscript was always moving toward rather than a terminus it arrived at by exhaustion.

None of these properties is impossible for a language model to approximate in short form. All of them become progressively harder to approximate as length increases because approximating them requires holding an intention across the entire work, which is precisely what current architecture cannot do.

This is where FictionMark enters the argument, and where I need to be direct about what it is and what it is not.

FictionMark is a structural manuscript analysis service. It was built to answer the question serious writers and serious publishers have always needed answered: Does this manuscript work as fiction at the architectural level? That question predates AI. It is the question developmental editors ask, that MFA workshop instructors ask, that the best literary agents ask when they are deciding whether a manuscript has the bones to survive revision and publication.

What AI has done is make that question newly urgent for a reason it was never urgent before. Structural analysis now serves a dual function. It serves its original purpose: identifying where a manuscript’s architecture succeeds and where it fails, giving the author actionable guidance for revision. And it serves a new purpose: generating evidence of sustained authorial cognition that stylistic detection cannot generate.

A manuscript that passes structural analysis at the level FictionMark examines has, by definition, demonstrated narrative intentionality across its full length. That demonstration is not proof of human authorship in the legal sense. The treatment of AI-assisted work under copyright law remains a genuinely unsettled question, and I am not a lawyer. What structural analysis provides is different and, in some ways, more useful than legal proof. It provides a credible, substantiated trust signal. It tells a reader, a publisher, or an editor that a mind was governing this work from beginning to end.

That is the instrument the market has not yet built into its accountability infrastructure. Not a detection tool. A cognition analysis tool.

The distinction between detection and cognition analysis matters enough to be re-stated precisely.

Detection is binary and backward-looking. It asks, “Was this generated by AI?” The answer is increasingly unreliable, increasingly gameable, and increasingly beside the point as human-AI collaboration becomes normalized in writing as it has in every other creative field. A writer who uses AI for research, for brainstorming, for generating rough material that they then substantially reshape, is not doing something categorically different from a writer who uses a research assistant, a writing group, or an editor who rewrites sentences. The question of origin is genuinely complex, and the industry’s attempt to treat it as simple is producing a false-positive problem and the reputational hazard that comes with it.

Cognition analysis is continuous and forward-looking. Is there evidence of sustained authorial cognition in this work? The answer to that question does not depend on the tools the author used. It depends on whether a governing intelligence was present across the entire manuscript, making intentional decisions in service of a coherent whole. That question is answerable, it is answerable with structured analysis rather than statistical pattern-matching, and it does not produce false positives against legitimate human authors whose prose happens to be clean and consistent.

It also cannot be gamed by running a manuscript through additional AI passes. Structural coherence is not a stylistic property that can be added after the fact. It is either present in the architecture, or it is not, and if it is not, no amount of prose-level refinement will install it.

A word on the limits of what I am claiming, because intellectual honesty requires it.

FictionMark is not infallible. Structural analysis at depth is a judgment, not a measurement, and judgment carries error. A manuscript with genuine architectural coherence might be analyzed by a framework that misses the logic governing it. A manuscript with significant structural problems might be produced by a deeply intentional human author working through their first novel. The tool is not a verdict. It is a substantiated assessment, and the market currently lacks such assessments entirely.

The alternative to an analysis of imperfect cognition is not perfect detection. It is the current situation: no mechanism for verification at all, a market reorganizing around whoever accumulated trust capital before the flood, and a cultural loss accruing silently in the mid-list where serious literary risk-taking has always lived.

An imperfect instrument calibrated to the right question is more useful than a precise instrument calibrated to the wrong one.

Post 5 moves from the manuscript to the ecosystem. If cognition analysis can rebuild trust signals at the level of the individual work, what happens when the reader-author relationship scales beyond discrete texts into the persistent cognitive environments described in Post 3? That is where the governance argument lives, and it is the question the industry is least prepared to ask.

E.S. Martell is a cognitive psychologist, AI governance researcher, and science fiction author. He publishes the After Biology Substack and runs Second Initiative Press. FictionMark.com is his AI-powered manuscript analysis service.

Share

Subscribe now

The publishing industry has survived disruptions before. E-books were supposed to destroy traditional publishing. Self-publishing was supposed to democratize literature into incoherence. Neither outcome fully materialized. The industry adapted, unevenly and with real casualties, but it adapted, and the optimists invoking that history now are not wrong to do so.

However, they are wrong about what adaptation means this time.

Previous disruptions changed how books were produced and distributed. The current disruption changes what readers are actually buying and why. That is a different kind of pressure, and it picks winners by criteria that have nothing to do with the quality of the work. Those criteria are already visible. This post will name them.

When synthetic content is cheap to produce and indistinguishable at first contact from genuine work, the reader’s problem is no longer finding good books. It is finding books they can trust. Those are not the same problem, and they do not have the same solutions.

Finding good books is a discovery problem. It was solved, imperfectly, by a combination of editorial gatekeeping, critical review, word of mouth, and algorithmic recommendation. Those mechanisms are now failing simultaneously, for related reasons. Editorial gatekeeping is under economic pressure. Critical review has been largely displaced by influencer culture. Word of mouth has migrated to platforms where volume and velocity outcompete considered judgment. Algorithmic recommendation optimizes for engagement metrics that synthetic content can game as readily as human content.

Finding books you can trust is a verification problem. And verification problems have a different structure. They are not solved by more discovery infrastructure. They are solved by trust signals. Trust signals require accountability, and that requires a verifiable relationship between a work and the mind that produced it.

That is what has collapsed. Not the appetite for serious fiction. The accountability infrastructure that made trust signals possible.

The practical consequence for readers is a change in how trust gets allocated. When any given book might be synthetic, evaluating books individually becomes an unreliable strategy. Readers begin migrating their trust upstream, away from discrete titles and toward verified sources. They are no longer asking whether this book is worth reading. They are asking whether this author is a mind worth following. That shift, from work-level trust to source-level trust, is what creates the conditions for cognitive ecosystems to form. The book becomes an entry point into a relationship rather than a self-contained transaction.

The reorganization that follows from this collapse is already visible, and it is concentrating value in predictable places.

Incumbent authors with accumulated trust capital are the primary beneficiaries. An author with an established audience built before the oncoming flood has a structural advantage that compounds. Their readers already have a verified relationship with their work. Every new book inherits that verification. The trust was earned incrementally over years of demonstrated accountability, and it functions now as a barrier to entry for new voices trying to establish the same relationship from scratch.

This is not a conspiracy. It is a market responding rationally to a verification shortage. When trust is scarce, buyers concentrate purchases with known quantities. That is sound individual behavior, but it poses a structural problem for literary culture, because literary culture has always depended on the discovery of new voices. The mechanisms that once performed that discovery are now compromised by exactly the conditions that make incumbent advantage so durable.

Cross-platform personalities are the second category of winners. Authors who built audiences through podcasts, newsletters, social media presence, and video before the compression hit can offer readers something synthetic content cannot replicate: a verified human presence across multiple contexts over time. The reader who has watched an author discuss their work on video, read their newsletter for two years, and followed their thinking through a public intellectual arc has a verification relationship that no AI detection tool is required to establish. That presence itself is the signal.

Jay Samit’s 85 percent marketing, 15 percent writing ratio, described in Post 1, is the economic expression of this dynamic. The work of creating and writing is no longer sufficient. The verified presence behind the work is what the market is buying. The book is evidence of the presence. The presence is the product.

Consider what that means for writers who are primarily writers and not marketers or influencers.

The market is reorganizing around a selection criterion that has no necessary relationship to literary quality. Cross-platform personality infrastructure, the ability to build and maintain a verified human presence across multiple channels over time, is a skill set. It is learnable. But it is not the same skill set as writing serious fiction, and the cognitive and temporal resources it consumes are substantial. An author spending 85 percent of their working life on presence maintenance is spending 85 percent of their working life not writing.

The optimist’s response is that this is simply the new reality and writers must adapt. That response is accurate as a description and inadequate as an analysis. Adaptation under these conditions systematically favors certain kinds of writers and certain kinds of books. Writers with existing platforms, with personalities that translate to video and audio, with the temperament and resources for continuous public presence, will adapt more readily and gain more readers than writers whose relationship to their work is primarily interior and whose natural mode of communication is the long-form written argument. Literary fiction, poetry, and serious nonfiction, the categories that have always required the most from readers and returned the most to literary culture, are exactly the categories least served by a market reorganizing around personality infrastructure.

The mid-list authors, who have always occupied the space where literary risk-taking is highest, are being set upon from both the top and bottom. From below, by synthetic content flooding the discovery layer. From above, by the concentration of trust capital in incumbents and personalities. What remains in the middle is a shrinking zone where serious writers without existing platforms are trying to establish verified human presence in a market that has lost the mechanisms to recognize and reward them.

That is not neutral adaptation. It is a specific kind of cultural loss, and recognizing it is the precondition for addressing it.

There is a second dimension to the trust collapse that the industry conversation is not yet adequately addressing, and it connects directly to the AI governance argument that runs underneath this series.

When readers attach to authors not as producers of discrete texts but as cognitive presences, the relationship changes in character. Consider what happens when a reader spends three years with an author’s newsletter, podcast, and books. They begin to anticipate how that author would frame a problem. They find themselves referencing that author’s arguments in their own thinking. They trust or distrust new ideas partly by how well those ideas fit the framework they absorbed. This happens below the level of conscious adoption. The reader does not decide to think like the author. They simply find, over time, that the author’s interpretive lens has become part of their own.

That is not a reader-author relationship in any historical sense. It is a cognitive ecosystem, and the reader is inside it.

That ecosystem is a powerful relationship. It is also an asymmetric one. The author, or more precisely the infrastructure built around the author’s identity, has significant influence over how the reader processes information, evaluates arguments, and forms judgments. At scale, across thousands or hundreds of thousands of readers, that influence constitutes a meaningful force on the cognitive environment of a culture.

The question of who governs that relationship, who is accountable for the integrity of a cognitive ecosystem built around an author’s identity, and what obligations attach to operating one, is not yet being asked at the level of seriousness it deserves. The industry is focused on the volume problem. The trust problem is one layer deeper. The governance problem is deeper still, and it does not resolve on its own when the market reorganizes.

What the market cannot do on its own is regenerate the accountability infrastructure that made trust signals possible in the first place. Reorganization distributes existing trust capital differently. It does not create new mechanisms for verifying that a mind was present and accountable in the making of a work. That problem requires an instrument the market has not yet built, and the absence of it is where the cultural loss becomes permanent rather than transitional.

Post 4 turns from the market to the instrument. If cognition is the scarce commodity and trust signals require cognition verification, what does verification actually look like at the level of the manuscript?

E.S. Martell is a cognitive psychologist, AI governance researcher, and science fiction author. He publishes the After Biology Substack and runs Second Initiative Press. FictionMark.com is his AI-powered manuscript analysis service.

Share

Subscribe now

The fear driving most publishing industry conversation about AI is that the technology will eventually write as well as humans. That fear is understandable and not entirely wrong as a long-term concern. But it is misdirected as a present-tense diagnosis, and the misdirection is costing authors and publishers the ability to think clearly about what is actually happening to their manuscripts and their market.

This post is not a polemic against AI. It is a clinical description of what current generation large language models structurally cannot do, grounded in what we know about how they function and how human cognition differs from them. The distinction matters because it is the evidentiary foundation for everything else in this series.

Start with what a large language model actually is.

A language model is trained to predict the most probable next token given everything that came before it. Token by token, sequence by sequence, it builds text that is statistically consistent with its training data. It is extraordinarily good at this. The prose it produces at the sentence level is often fluent, sometimes elegant, occasionally indistinguishable from competent human writing when examined in isolation.

That capability is real and should not be dismissed. But it has a structural ceiling that is not a temporary limitation of current models. It is a consequence of the architecture itself.

The model has no intention. It has no object it is trying to make. It is not writing a novel. It is completing a sequence, and the sequence it is completing at page 200 has only a probabilistic relationship to the sequence it completed at page 20. The context window gives it access to recent text. It does not give it anything analogous to what a human author carries through the writing of a long work: a sustained, evolving understanding of what this particular story is for, what it is doing to the reader, and what every local decision costs or earns against that larger purpose.

This is not a minor gap. It is the gap between a sophisticated pattern-completion engine and a mind.

The failure modes that follow from this gap are consistent enough to constitute a diagnostic profile. They are not random. They appear in predictable places for predictable reasons, and once you know what to look for, they are legible in a manuscript even when the prose itself is clean.

Local coherence, global incoherence.

The most consistent signature of AI-generated long-form fiction is that it works at the paragraph level and fails at the structural level. Individual scenes are competent. Dialogue exchanges are plausible. Descriptive passages hold together. But the relationship between scenes, the way one scene creates conditions for the next, the management of narrative tension across an entire act, the timing of revelation against the reader’s developing understanding, these require holding an intention across tens of thousands of words and making hundreds of micro-decisions in service of that intention. Models generate the next scene. They do not govern the relationship between scenes.

The practical result is fiction that reads smoothly in excerpt and feels structurally hollow at length. Chapters that don’t earn what follows them. Reveals that land without having been properly loaded. Climaxes that arrive on schedule but carry no accumulated weight.

Character drift.

A human author carries a character in mind across the entire manuscript. The character’s voice, psychology, decision-making logic, and developmental arc are held as a coherent internal model that the author consults, consciously or not, at every point of contact. When a character does something surprising, the author knows whether the surprise is earned or false, because they know who the character is at a level that exceeds what is on the page.

A language model has access to what has been written about the character. It does not have a model of the character. The result is voice consistency that holds at the sentence level and drifts across dramatic arcs. A character who is established as emotionally reserved will, fifty pages later, deliver an expressive monologue that serves the immediate scene’s needs without registering as a violation of who that character is. The model is optimizing locally. Nobody is watching the whole.

Thematic flattening.

Theme in serious fiction is not a message inserted at designated intervals. It is a pressure that operates on every element of the work simultaneously, shaping which details get noticed, which conflicts get foregrounded, which resolutions feel earned and which feel arbitrary. Theme is what a novel is thinking about, and the thinking happens below the level of explicit statement.

A language model can reproduce thematic language. It can generate passages that discuss or illustrate a theme. What it cannot do is maintain thematic pressure across a long work, because maintaining that pressure requires a sustained authorial intention that is continuously modifying local decisions in light of the whole. The result is fiction that states its themes rather than embodying them. The thematic content is present but inert. It sits on the surface of the prose rather than generating it from underneath.

The second half problem.

This one is reliable enough to use as a rough diagnostic heuristic. AI-generated novels tend to degrade in the second half. The opening chapters, which establish characters, setting, and premise, benefit from the model’s access to genre conventions and its ability to generate competent setups. The middle and late sections, which require the payoff of everything established earlier, require a kind of structural memory and intentional management that the model cannot sustain. Subplots go unresolved or resolve arbitrarily. Secondary characters established with apparent purpose disappear. The ending arrives because the word count demands it, not because the story has reached a destination it was always moving toward.

A word on what cognitive psychology contributes here, since Post 1 flagged that this is where the disciplinary background becomes relevant.

Human authors do not consciously track all of this. The sustained cognitive work of writing a long fiction is largely implicit. Experienced authors develop something closer to a felt sense of the manuscript’s state: what has been promised, what has been loaded, what is owed, what is available. This felt sense is a form of working memory that operates across the entire project, updated continuously as the work develops.

It is, in the language of cognitive psychology, an extended cognitive process. The author is not just the person sitting at the keyboard. The author is the person plus the manuscript plus the accumulated history of every decision made in its production. The notebook, the outline, the earlier drafts, the margin notes, the memory of why a particular character was given a particular trait three months ago, these are not incidental to the writing. They are part of the cognitive system doing the work.

A language model has access to the text. It does not have access to the cognitive history that produced the text, nor the intention that is supposed to be governing its continuation. It is, in this sense, not writing the manuscript. It is completing a document.

That distinction sounds philosophical. It has practical consequences that are visible in the manuscript, and they are visible in the specific ways described above.

This matters for how we think about cognition verification, the concept introduced in Post 1 and developed across this series.

If the failure modes of AI-generated fiction are structural and predictable, then structural analysis of a manuscript is not merely a quality assessment. It is an indirect measure of whether sustained authorial cognition was present during the work’s production. A manuscript that demonstrates global narrative coherence, genuine character consistency across dramatic arcs, embodied rather than stated thematic development, and second-half structural integrity has, by definition, evidence of a mind at work across the whole.

You cannot generate that profile by running a prompt through a language model, then running the output through a second prompt, then a third. The architecture doesn’t support it. What you get is a cleaner version of the same structural hollowness, because the hollowness is not a prose-level problem. It is a cognition-level problem.

This is the evidentiary basis for treating structural analysis as a distinct category from AI detection. Detection asks about origin. Cognition analysis asks about presence. Those questions have different answers, require different instruments, and serve different purposes.

The next post in this series examines what happens to the market when cognition becomes the scarce commodity, and why the reorganization underway is not the neutral adaptation the optimists are describing.

E.S. Martell is a cognitive psychologist, AI governance researcher, and science fiction author. He publishes the After Biology Substack and runs Second Initiative Press. FictionMark.com is his AI-powered manuscript analysis service.

Subscribe now

That question has followed me for years. It started as a theoretical possibility drawn from conversations with a physicist and ended as the structural spine of an entire series. Getting it from one place to the other nearly broke the first novel, and what I learned in the process is probably the most useful thing I can tell a writer about the relationship between research and story.

The short version: research will try to eat your book. Your job is to not let it.

When I began writing Heart of Fire Time of Ice, I wanted the time travel mechanism to be scientifically defensible rather than narratively convenient. That instinct was correct. What followed from it was less so. I immersed myself in quantum physics, probability wave theory, and the work of physicist Fred Alan Wolf, whose framework for what he calls extraordinary time travel aligned almost perfectly with the fictional vision I had for Kathleen, my particle physicist protagonist. Wolf’s idea, stated as simply as it can be, is that time travel is not a technological problem requiring exotic machinery but a natural property of consciousness operating at the quantum level. The most likely mode of transition, in his view, would be through a lucid dream or out-of-body experience. One would simply awaken from a conscious dream in another time.

That’s the sentence that built the series. Everything else in Wolf’s framework, the possibility waves, the probability curves, the complex conjugate echo waves cycling back and forth through time, is the scaffolding behind it. Genuine, serious physics developed by a credentialed researcher. And almost none of it belongs in the novel.

This is the trap research sets for writers who take it seriously. You do the work. You develop genuine understanding of a complex framework. You feel the weight of what you’ve learned and want the reader to feel it too. So you put it in. All of it. And then you discover, usually through the pained responses of early readers, that you have written a physics lecture with a plot attached rather than a novel informed by physics.

That’s what happened with my original approach. My conception of Kathleen’s work and the time travel formula was so elaborate that it would have driven most readers away before the story had a chance to take hold. The discipline required was operating in real time during composition rather than in retrospect. Most writers have to see the overwritten version on the page before they can cut it. What I found was that genuine understanding of the mechanism made it possible to make the judgment at the point of generation, to know what belonged before writing it down, and to trust that what was left out was still present underneath, shaping the work invisibly. Hard-gained material that I had genuinely wanted to share didn’t make it into the novel. It was the right decision, and it taught me something that applies to every kind of research a writer undertakes, whether the subject is quantum physics, medieval warfare, surgical procedure, or Clovis culture hunting practices.

The rule is this: the story comes first, and the story rarely needs the explanation.

What the story needs is the confidence that comes from knowing the explanation exists. There is a quality to fiction written by an author who genuinely understands the mechanism underlying the fictional premise that is different from fiction where the author has simply borrowed a concept and dressed it up. Readers feel that difference without being able to name it. The depth doesn’t have to be visible on the surface. It has to be present underneath, shaping the decisions the author makes about what to include, what to imply, and what to trust the reader to accept without explanation.

In Kathleen’s case, what survived was this: she is a particle physicist who discovers an anomaly in CERN data that leads her to develop an equation describing the manipulation of time. When she is nearly killed, she activates the equation instinctively and displaces into the Younger Dryas period during the last ice age. The physics behind that displacement is conceiveably real, or as real as theoretical quantum mechanics gets. But the reader doesn’t need the math. They need to believe that Kathleen could do this, and they need to believe it because the author does.

The discipline required is genuinely difficult. You have spent weeks or months developing expertise you will never fully use. The unused material feels like waste. It isn’t. It’s foundation. The reader walks across it without seeing it, which is exactly what foundation is for.

Wolf’s framework gave me one other thing beyond the means: the closing image that has stayed with me since I first encountered his ideas, and that informs the series at every level. He believed that if time travel of this kind were possible, the experience would feel like waking from a lucid dream. You would simply open your eyes in another time, with no machine, no corridor of light, no dramatic transition. Just the ordinary miracle of consciousness arriving somewhere it hadn’t been before.

What if you wake up elsewhen?

That question is still the reason I find going to sleep every night a particular kind of adventure. It is also, not coincidentally, the reason the Time Equation series exists.

Share

Subscribe now

There is a category of human experience that most fiction writers either avoid entirely or handle at arm’s length, borrowing secondhand from other fiction rather than from anything they actually know. Altered states, spiritual experiences, out-of-body travel, lucid dreaming, the territory where consciousness seems to operate outside its normal boundaries, are underwritten in genre fiction precisely because most authors haven’t done the primary research. I have, and I’m going to show you how it can feed directly into the story creation process.

Fair warning before we go further. What follows includes some unusually intimate material for a craft essay. I’m going to describe personal experiences that I cannot verify objectively and make no claim to verify. You need not believe they were real. What I will say is that they were phenomenologically real at the time, meaning that as subjective experiences they were as vivid, coherent, and emotionally significant as anything I have encountered in waking life. For a writer, that is the only criterion that matters. The question is not whether the experience can be verified but whether it was vivid and coherent enough to translate into fiction that feels lived rather than invented. These experiences were, and the Cadeyrin chapters in Heart of Fire Time of Ice exist because of them.

I began studying this territory many years ago after an experience that both frightened and puzzled me. As a trained scientist I had no ready framework for it, so I did what scientists do: I read everything I could find and started keeping records. For years I kept a journal by my bed, recording altered state experiences as close to the moment of occurrence as possible, before the details dissolved. What follows are examples from that journal, offered not as evidence for any particular theory of consciousness but as source material, the raw experiential data that eventually found its way into fiction.

The first entry worth sharing began as an ordinary dream, clearer and more coherent than most, and then crossed into something different.

About 4 AM. I woke up, changed position, and fell back asleep. I was walking down a sidewalk carrying a U-shaped piece of metal. A large pile of rocks was on the edge of a college campus. I went around them and entered a very small cave through a small opening. There was a bed inside with drawers under it. I began looking into them but found nothing of interest. Then I saw a dark door at one end of the cave. I went through, and suddenly the quality of the dream transitioned. I became aware that I was dreaming. As a result, I became conscious within the dream and had control of my actions.

I looked around at the room on the other side of the dark door. It looked like a giant, dimly lit barn. I consciously bypassed a door to the outside since I felt I would wake up if I went through it. After looking around, I decided to go outside anyway. I jumped off a high step and landed right in front of a young man. Our eyes met, and we both smiled. Was he another dreamer, or a figment of my dream? I don’t know. I walked past and turned to the left at the far end of the building. The dream faded, and I lost control and fell entirely out of consciousness and into sleep.

What distinguished this from an ordinary dream was the moment of crossing: the point at which I became aware I was dreaming and could direct my own actions consciously. Researchers call this a lucid dream. The experience has a distinctive quality that anyone who has had one will recognize immediately, a sense of freedom and presence simultaneously, of being both author and character in the same moment. That dual awareness is extraordinarily useful fictional territory.

A separate category of experience goes further. Out-of-body experiences differ from lucid dreams in that the environment encountered tends to be the actual physical space surrounding the sleeper’s body rather than a constructed dream landscape. They feel less like imagination and more like perception, which is precisely what makes them both more disorienting and more useful as source material.

An author named Robert Bruce suggested a validation method: select a playing card without looking at its face, place it on a high shelf, and attempt to read it during an out-of-body experience. I tried this with two cards, one on each bookshelf flanking our fireplace, placed high enough that I couldn’t see them from standing height.

Here’s another journal entry: One morning while drowsing, I became conscious that my face was bumping against the spines of a long row of books. I realized I was near the bookshelf and remembered the cards. I looked and saw a black four. When I tried to identify the suit, all I could perceive was a rectangle with a diagonal line crossing it, the international symbol for prohibition. I woke up and checked. The card was the four of spades. The second card, on the opposite shelf, was the four of clubs. I had apparently been trying to perceive both simultaneously, and the only element common to both was the black four. The suit was obscured because I was receiving two signals at once.

I am not presenting this as proof of anything. I am presenting it as an experience that raised questions I have not been able to dismiss, and that lodged itself in my imagination with enough force to eventually shape a novel.

The most viscerally memorable experience I recorded involved waking into paralysis. I was drowsing in a recliner when I heard my name called loudly twice in a nasal tone. I tried to wake fully and found I could not move my body or extremities. Then I became aware that I was floating roughly two feet above myself. I kept trying to merge back into my body without effect. Suddenly a loud vibration erupted across my chest directly over my heart, startling enough that I snapped back into my body and opened my eyes simultaneously. The experience lasted perhaps thirty seconds and left me shaken for considerably longer.

These are the kinds of experiences that don’t leave a writer alone. They sit in the imagination and demand to be used.

When I was writing Heart of Fire Time of Ice, the first book in the Time Equation series, I needed to give Cadeyrin, my Clovis culture male lead from approximately thirteen thousand years ago, a motive and a mechanism for locating Kathleen, my modern physicist who has accidentally displaced herself into the Younger Dryas period. Cadeyrin is established as a trained shaman, which gave me the fictional license I needed. But the scene required authenticity, not generic mysticism. It needed to feel like something that had actually happened to someone.

I had the journal. Here is the result, condensed from a three page scene in chapter eight for the purposes of this essay:

Cadeyrin felt connected to the spirit world through the north wind and didn’t want to move, but he knew he couldn’t stay where he was. It wasn’t safe. He needed to return to the fire.

He tentatively started off towards the flickering light along an old bison path worn in the prairie grass. After a few steps, he became aware that a friendly animal spirit was leading him. Wolves suddenly howled in the near distance. The eerie sound led him to recognize his guide as his personal wolf spirit.

The flames grew high, and his spirit guide took on a feeling of wildness mixed with joy. He followed it as it led him on a triple circuit around the fire. At that moment, his companion shifted into a huge wolf-like figure. Together they raised their heads and howled upwards, and then he was following the spirit guide, traveling through the sky, far away from the fire circle.

His sense of being accompanied faded, and he ended up in a high location, as if on the peak of an immense mountain. As he looked down, he could see through both time and space. He traveled towards a forest, then back, then a figure appeared, but it was unclear and hazy. He followed it, and it receded into the future. He paused, and it came back. He felt that somehow it gave him a sense of completeness and he yearned for more of that feeling.

He felt that the spirit he’d seen was approaching, and he was glad.

The vision faded. He was suddenly back at the level of the flames of his fire, his guiding wolf spirit bounding up with a joyful greeting before disappearing, leaving him alone on the prairie.

The night had fled, and the dew was on the grass. After eating, he wondered which way he should go. The guidance he’d received had been clear. The signs indicated that someone or something was coming towards him. He needed to meet them to ensure the best outcome. He tried stepping out in each of the four directions and waiting to see which felt best. East was it.

The scene works, to the extent that it does, because it was built from the inside out. The paralysis, the sensation of floating, the moment of transition between states, the feeling of absolute freedom followed by sudden return, all of it came from the journal. Cadeyrin’s shamanic journey is my out-of-body experience translated into the experiential vocabulary of a Clovis hunter thirteen thousand years in the past. The fictional frame is imagination. The phenomenological content is memory. And critically, the vision doesn’t function as atmospheric decoration. It solves a specific plot problem: Cadeyrin learns that someone is approaching from the east, decides to go meet them, and the next movement of the novel begins. The spiritual experience does narrative work.

This is what ‘write what you know’ actually means when applied to interior experience. It doesn’t mean restrict yourself to what you have externally observed. It means find the authentic experiential core and build outward from there. Readers feel the difference between a spiritual experience constructed from other fiction and one constructed from something the author actually lived through, even when they cannot articulate why one compels and the other doesn’t.

The interior frontier is some of the least crowded territory in speculative fiction. Most writers won’t go there because they haven’t done the work, or because the material feels too strange to defend in public. Both are understandable positions. But for a writer willing to keep the journal, accept and internalize the experiences, and, then translate them honestly into fiction, it represents an almost entirely open field.

The experiences I’ve described may or may not reflect something real about the nature of consciousness. That question I leave open for now. What I know with considerably more confidence is that they produced fiction I could not have written any other way.

Share

Subscribe now

Some readers of the first book in the Gaia Ascendant series, The Time of the Cat, wrote to me with a specific objection. Dec, the protagonist, is established as an exceptionally capable but otherwise normal human being. Giving him psychic abilities late in the story felt, to some readers, like a violation of the fictional contract. It’s a fair objection and one worth answering carefully, because the craft problem it identifies is real and applies well beyond my own work.

The foundational requirement of speculative fiction is internal consistency. This doesn’t mean the universe you build has to follow the rules of the one we inhabit. It means it has to follow its own rules, without exception and without apology. You can blend genres, introduce extraordinary elements, and ask readers to accept premises they would reject in ordinary life. What you cannot do is change the rules mid-story without preparation. The reader who has agreed to inhabit your fictional world has made a contract with you, and breaking it doesn’t just strain credulity. It breaks trust, and broken trust is something a reader rarely extends twice.

So the question is whether introducing Dec’s psychic ability broke that contract. I don’t believe it did, and the reason is preparation. From the opening chapters, Dec relies on what he thinks of as a sixth sense, an intuitive awareness of threat that operates below the level of conscious reasoning. It’s presented as heightened perception rather than paranormal ability, but the seed is planted. By the time his latent capacity is forced open under extreme duress on Titan, the reader has already been living with its minor expression for the entire preceding story. The development feels sudden because Dec experiences it as sudden. It shouldn’t feel arbitrary, because it isn’t.

This is the technique worth extracting for broader use: if your story requires an extraordinary element, introduce it early in a minor key. Not as foreshadowing so obvious it telegraphs the plot, but as a detail consistent with the character that the reader files away without consciously registering. When the element becomes central later, the reader’s recognition is not surprise but retrospective understanding. That’s the difference between a twist that feels earned and one that feels imposed.

The Time Equation series presented a more complex version of the same craft problem, and one that required me to be clear in my own mind about what I actually believed before I could build the fictional rules consistently. The philosophical spine of the series grew out of a conversation with physicist Fred Alan Wolf, whose work on quantum consciousness informed the series from its inception. Wolf told me he could conceive of three ways a civilization might technologically manipulate time, none of which we are remotely capable of building, but that he believed time travel was likely inbuilt into the human. His thought was that one might go to sleep and wake up in elsewhen. I took that idea seriously enough to name a character after him and build a series around it.

Kathleen, a post-graduate particle physics student, discovers an anomaly in CERN data that leads her to develop an equation describing the manipulation of time. When she is nearly killed by an enemy agent, she activates the equation instinctively and is displaced into the Younger Dryas period, with no conscious understanding of what she has done or how to reverse it. The same pattern as Dec: extreme threat forcing open a latent capacity the character didn’t know she had, and didn’t choose to use.

The rules Kathleen operates under are specific and consistently maintained across the series. The ability scales with understanding and practice. It has a hard constraint: one absolutely may not occupy the same spatial and temporal location as another instantiation of oneself. That constraint isn’t arbitrary. It’s the mechanism that keeps the fictional universe coherent and prevents the paradox problems that collapse lesser time travel narratives. As Kathleen becomes more facile, the applications become more sophisticated, but the rules never change. By the later books she can do things like jump back a few seconds, retrieve her own weapon, and return to the scene a millisecond after she disappeared, standing behind her assailant. The reader accepts this because the logic has been established and honored from the beginning.

What both series share, and what I’ve come to understand as a consistent philosophical position across my fiction, is this: the extraordinary capability lives inside the human rather than in external technology or supernatural intervention. Dec and Kathleen are not special because something was done to them or given to them. They are outliers on a bell curve of human potential that most of us never approach because we are never sufficiently threatened or sufficiently prepared to recognize what we carry. That’s a premise I find more interesting than the technological fix, and more honest to what I actually believe about human consciousness.

Which brings me to the market reality, and it is a reality worth stating plainly. Psychic phenomena, paranormal abilities, and expanded human consciousness have a dedicated readership that is large, loyal, and consistent across decades of genre publishing. These readers are not credulous or unsophisticated. They are drawn to fiction that takes seriously the possibility that human potential extends beyond what mainstream science currently measures or acknowledges. Writing for that readership is not a compromise of craft standards. It is knowing your audience and respecting what they bring to the page.

I’ll state my own position as a writer rather than as a scientist making empirical claims: I believe these phenomena are real. That conviction informs my fiction. It doesn’t require the reader to share it. What it does require is that I handle the craft execution well enough that readers who don’t share it can still inhabit the story without feeling cheated. The technique is the same regardless of the reader’s prior convictions: establish the extraordinary element as a latent characteristic before it becomes a plot mechanism, ground it in the character’s psychology, maintain the rules without exception, and trust the readership to meet you where the story is.

Dec’s ability didn’t appear from nowhere. Neither did Kathleen’s. They were always there, waiting for the right pressure to force them open. That’s true of the best extraordinary elements in speculative fiction generally. The writer’s job is to make sure the reader, looking back from the moment of emergence, can see exactly where it was hiding.

Share

Subscribe now

Every writer eventually confronts the question of message. How much of what you believe should find its way into your fiction, and how deliberately should you put it there? The question sounds like a matter of personal preference, but it isn’t. It has a structural answer.

You cannot separate your values from your storytelling because your values are not a layer you apply to a narrative. They are the lens through which you see the world, and seeing the world is the precondition for imagining one. Your worldview determines which details you notice, which characters feel real to you, which conflicts seem worth resolving, and which resolutions feel earned. A writer who attempts to excise their values from their fiction is not producing neutral work. They are producing evasive work, suppressing something structurally unavoidable, and readers feel the absence even when they cannot name it. Inauthenticity has a texture, and it tends to drive readers away without their being able to explain exactly why.

This does not mean you should preach. It means nearly the opposite. The writer who trusts their values enough to embed them in character and consequence rather than declaration is the writer whose work carries genuine persuasive weight. The writer who doesn’t trust the story to carry the argument will stop the narrative to make the argument explicit, and that is where fiction dies. The moment a character exists to deliver a position rather than to live a life, the reader’s investment collapses. They are no longer inside a story. They are receiving a lecture.

Mark Twain told a story about a bricklayer’s apprentice who accidentally dropped his load from a roof onto a man passing below. Twain observed that a dog would never have been in that position, having the wit to look up and move. Whatever Twain intended by it, the image captures something precise about the relationship between a writer and their material: the reader who is fully inside a story is not looking up. They are not managing their relationship to the author’s opinions or calculating whether they agree with the premises. They are present. Didactic fiction destroys that presence by forcing the reader to look up, to step outside the narrative and evaluate the argument being made. Twain’s own work carried enormous moral weight, on race, on class, on the mythology of American innocence, but it carried that weight because Huck Finn and Jim were alive on the page before they were symbols of anything. The argument arrived through the story, not instead of it.

The distinction Twain’s example draws is precise and worth holding onto. Values embedded in character produce fiction. Values imposed on character produce polemic. The difference is not ideological. It has nothing to do with which values are being expressed or whether you agree with them. A writer can produce unreadable didactic work in service of any conviction, progressive or traditional, religious or secular, optimistic or nihilist. The sermon problem belongs to the method, not the message.

The deeper mechanism underneath all of this is about agency. Readers tolerate persuasion when they retain the ability to reach conclusions themselves. They resist when they feel managed. Story creates the conditions for self-directed inference: the reader follows the character, experiences the consequences, and arrives at meaning through their own interpretive act. Polemic attempts to shortcut that process by delivering the conclusion directly, and readers recognize the attempt even when they cannot articulate it. The resistance is not to the idea. It is to the removal of their role in discovering it.

The practical implication is this: if you find yourself writing a scene whose primary purpose is to establish a position rather than to advance a story, stop. Find the character whose life makes that position true or false or complicated, and write that life. Conclusions reached feel like understanding. Instructions received feel like imposition, and readers, quite reasonably, resist imposition.

The campfire story that started human narrative tradition was not a lecture. It was an account of what happened, what was at stake, who survived and how. The values it transmitted were carried in the telling, not announced before it. That remains the operating principle. Tell the story. Trust it to carry what you believe. If it cannot, the problem is not the reader’s resistance. It is that the story isn’t doing its job.

Subscribe now

Share

Every story needs conflict, and most good ones need a monster. Not necessarily a creature with claws, though those work fine. A monster is anything that threatens what the protagonist values, forces a reckoning, and cannot simply be reasoned away. By that definition, the monster could be anything from an irritating neighbor or a bad boss at work to an alien life form that implants eggs into humans. When writers ask me why I keep returning to existential threats in my fiction, massive societal disruption, alien incursion, civilizational collapse, I give them the honest answer: because that's what the genetic wiring demands.

Humans are built for adversarial structure. Not as a cultural habit or a failure of imagination, but as something closer to biological fact. We are organisms that survive by protecting resources, and protecting resources requires knowing what threatens them. The brain that cannot identify an enemy is a brain that cannot organize a defense, and brains that couldn’t organize a defense didn’t pass their architecture forward. What survived is us: pattern-seeking, threat-sensitive, and deeply oriented toward the question of who is with us and who is not.

It is worth pausing on what the word monster actually means, because the answer depends entirely on where you are standing. A Smilodon stalking the edge of a human encampment was not evil. It was hungry, or feeding its young, executing a genetic template with complete fidelity to its own nature. The humans crouching around the fire experienced it as a monster. The cat experienced them as a problem, or a meal. Evil is a label the prey applies. It is not a property of the predator. This is the foundational observation that everything else in this essay builds on, and it is the observation that separates useful antagonists from lazy ones.

This has an obvious dark side in human history, and history documents it exhaustively. But it also has a narrative implication that matters more to me as a writer. When you give a reader a genuine threat, something that endangers what the characters and the reader both care about, you aren’t deploying a genre convention. You’re activating a cognitive structure the reader already has and has always had. The story stops being entertainment and starts being practice.

Here is where the craft prescription gets specific. The monster that knows it is evil is a narrative convenience. The monster that believes it is righteous is a character. Consider Genghis Khan. From the steppes of Mongolia to the gates of Vienna, he left devastation on a scale that is difficult to comprehend. His victims experienced him as apocalypse. He understood himself as something closer to a force of natural order, consolidating power, rewarding loyalty, punishing resistance with a ruthless efficiency that was, by the standards of his own world, coherent statecraft. He was not confused about what he was doing. He simply did not share his victims’ assessment of it. That gap between self-perception and impact is not a historical curiosity. It is the engine of every genuinely threatening antagonist in fiction.

What makes an antagonist threatening, on the page and in life, is not malevolence but internal coherence. The reader has to be able to follow the logic, to see how someone arrives there through choices, history, and values taken past the point where they remain humane. That understanding produces dread rather than simple revulsion. Revulsion is easy. Dread requires the reader to recognize something, and recognition is uncomfortable in ways that matter.

The practical consequence for character building is this: find a scene, early and away from the central conflict, where the antagonist exists on its own terms. Not a scene that excuses what it does, but one that makes its internal logic legible before the reader needs to hold that understanding under pressure. The antagonist facing its own version of difficulty, operating according to values it believes in, or protecting something it values, primes the reader to experience the confrontation as a collision between coherent forces rather than hero versus obstacle. That’s where the dread lives.

Mary Shelley understood this at a structural level, not merely a thematic one. The monster’s chapters in Frankenstein are written in first person. The reader doesn’t hear about his suffering secondhand through Victor’s guilty conscience. They inhabit it. By the time the monster becomes genuinely dangerous, the reader has already been inside the logic that made him so. The horror isn’t that he’s evil. It’s that he’s comprehensible. That’s the standard worth holding yourself to. The weak version of the advice is to give your villain a backstory. The actual prescription is closer to what Shelley did: give the antagonist a scene in which the reader’s sympathies are genuinely engaged, not managed from a safe distance. The discomfort of understanding is the point.

Fiction has always served this function. The campfire story about what lives in the dark beyond the firelight is the same operation as the novel about what happens when the civilization you assumed was permanent turns out not to be. The reader follows the protagonist into the threat because some part of the nervous system recognizes the situation as one worth rehearsing. Survival, cooperation under pressure, the discovery of what a person actually is when everything is stripped away. These aren’t themes we invented. They’re questions the species has been asking since long before anyone wrote them down.

What draws me specifically to large-scale threats is the social dimension they expose. Individual conflict reveals character. Civilizational threat reveals whether there is anything underneath the social structures we’ve built, anything that holds when the structures themselves are gone. My honest assessment is that the answer is uncertain, and that uncertainty is where the interesting fiction lives. Characters who discover they are better than they believed, and characters who discover the opposite, are both telling you something true.

The question I keep returning to in my work, and the one I’ll leave here without resolving, is whether we could ever evolve past the need for adversarial structure without ceasing to be recognizably human. The monster in the story is never really just the monster. It’s the condition that makes us capable of being more than we were before we faced it. Whether that trade is worth what it costs is the question my fiction keeps asking. I haven’t settled on an answer, and I’m not sure I want to.

Share

Your best work may be invisible before it’s read.

Not because it isn’t good enough. Not because the market doesn’t want what you’re writing. Because the mechanism that once connected good books to the readers who needed them is breaking down, and the industry’s proposed fixes are aimed at the wrong problem.

The publishing conversation right now is dominated by volume. Four million titles published in the United States last year. A Seoul publisher releasing nine thousand books in twelve months. Amazon’s self-publishing infrastructure flooded with content generated faster than any human could write it. The visible crisis is noise: too much, everywhere, drowning signal.

That diagnosis is accurate as far as it goes. But it stops too early, and where it stops is exactly where the real argument begins.

Jay Samit is an international bestselling author who has worked with Simon & Schuster and Macmillan. He recently described his current marketing reality: a decade ago, he spent 85 percent of his time writing and 15 percent marketing. In 2026, those numbers have inverted. Eighty-five percent of his effort now goes to marketing. Fifteen percent to writing.

Read that carefully. This is not a complaint from a struggling mid-list author. This is someone with an established audience, major publisher relationships, and the professional infrastructure to compete. And he is spending the majority of his working life not making the thing, but proving it exists and is worth finding.

That inversion is the real story. Not the volume of AI-generated content. Not the copyright questions, though those matter. The inversion tells you that something structural has changed, something upstream of marketing strategy or platform algorithms or detection tools.

What changed is this: the old signal-filtering mechanisms that determined which books became visible have collapsed. And they collapsed not primarily because of AI, but because AI accelerated a failure that was already underway.

For most of publishing history, the bottleneck was production. Writing a book was hard. Publishing it was harder. Distributing it was harder still. Those friction points were inefficient and often unjust, they kept plenty of good work from reaching readers, but they performed a secondary function that nobody designed and nobody appreciated until it was gone. They filtered. The cost of producing and distributing a book meant that someone, somewhere, had made a judgment that this particular object was worth the investment.

That judgment was often wrong. It was frequently biased. It reproduced existing hierarchies more reliably than it discovered genuine talent. None of that is in dispute. But it was a judgment, and judgment implied accountability. A publisher who consistently published bad books went out of business. An editor whose picks failed lost standing. The filtering was imperfect, but it was tethered to consequences.

What AI has done, in combination with self-publishing infrastructure and algorithmic distribution, is detach production from judgment almost entirely. The cost of producing a book is now approaching zero. The cost of distributing it on major platforms is negligible. Which means the old friction-based filter is gone, and nothing has replaced it.

The industry’s response has been to reach for detection. Human-authored certifications. AI screening tools. Disclosure requirements. These are understandable reactions, and some of them will provide partial relief. But they are aimed at origin, not at quality. They ask “was this written by a human?” when the question that actually matters to readers is something harder and more important.

Here is the question: Is there a mind behind this work that was present and accountable throughout?

Not “did a human type these sentences?” A sufficiently motivated person can run AI-generated prose through enough revision passes to defeat most detection tools. Not “does this read as authentic?” Authenticity is a stylistic property that can be approximated. The question is whether the architecture of this book, its structure, its developmental logic, the coherence of its thematic argument across four hundred pages, reflects a sustained act of human cognition.

That is a different kind of question, and it requires a different kind of answer.

AI-generated fiction fails in characteristic ways. The failures are rarely at the sentence level, which is why detection tools calibrated to prose style produce the false positive rates that are now creating serious reputational hazards for legitimate authors. The failures are structural. AI produces locally coherent prose with globally incoherent architecture. Scenes work. Structures don’t. Character voices hold sentence to sentence and drift across dramatic arcs. Thematic development flattens in the second half of a book because the model has no sustained intention, only prediction. The machinery that produces meaning at the level of the whole novel is precisely what current generation models handle worst.

This is not a minor technical limitation. It is the direct consequence of how large language models function. They are pattern-completion engines operating at the level of sequence. They do not hold an intention across eighty thousand words and make hundreds of micro-decisions in service of that intention. A human author does. That difference is legible, if you know how to read for it.

This is where I need to flag something directly.

I built FictionMark as a structural manuscript analysis service, not as an AI detection tool. Its function is to evaluate whether a manuscript works as fiction at the architectural level: pacing, character coherence, narrative logic, thematic development. I did not design it to answer the question “was this written by AI?”

But structural analysis at that level is, functionally, a proxy for the question publishers and readers are now desperate to answer. A manuscript that passes structural analysis at the level FictionMark examines has, by definition, evidence of sustained authorial cognition. You cannot fake global narrative coherence by running a prompt through a language model. You can produce prose that resembles a novel. You cannot, yet, produce the cognitive architecture of one.

I am naming FictionMark here because it is relevant to the argument, not to sell you a service. If the argument is wrong, the tool is irrelevant. Evaluate the argument first.

The series this post opens will run five installments. Each one advances a single claim: the publishing crisis is a cognition verification crisis, not a volume crisis. Volume is the presenting symptom. The disease is the collapse of mechanisms that could verify whether a mind was present and accountable in the making of a work.

The subsequent posts will cover what AI structurally cannot do and why, how the market is reorganizing around trust rather than discovery, what cognition analysis actually measures and why it differs from origin detection, and where the governance gap lives when reader-author relationships migrate into persistent digital ecosystems.

The industry is reaching for solutions calibrated to the wrong problem. The right problem is harder, more interesting, and more solvable than it looks from inside the panic.

E.S. Martell is a cognitive psychologist, AI governance researcher, and science fiction author. He publishes the After Biology Substack and runs Second Initiative Press. FictionMark.com is his AI-powered manuscript analysis service.

Share

That raises the obvious question.

If the problem is structural, can it actually be measured?

Most writers assume the answer is no. They treat perspective, continuity, and pacing as matters of judgment, something you feel your way through with experience and intuition. That belief is understandable. It is also wrong in a very specific way.

These failures feel subjective because they are experienced subjectively. But they are produced by patterns that are not subjective at all. They are repeatable, detectable, and in many cases quantifiable.

The key is to stop asking whether a passage “works” and start asking what the reader is being asked to process at each moment.

What Measurement Means in This Context

Measurement in fiction does not mean reducing a story to numbers or pretending that art can be replaced by metrics. It means identifying the specific conditions under which readers reliably disengage, and then detecting those conditions in a manuscript before the reader ever encounters them.

This is closer to engineering than to criticism.

A structural engineer does not ask whether a bridge feels stable. They ask whether the load-bearing structure meets defined constraints. A story is not a bridge, but the reader’s cognitive experience does operate under constraints. When those constraints are violated, engagement degrades in predictable ways.

Three domains account for the majority of those violations. Perspective, continuity, and pacing are not abstract craft ideas. They are systems with failure conditions.

Perspective as a Trackable System

Point of view defines the reader’s location inside the narrative. Once established, it creates expectations about what information is available and how it is filtered.

From a measurement standpoint, perspective breaks when the text presents information that cannot be accessed from the established viewpoint without a transition.

Consider a simple example:

John watched Mary cross the room. She looked nervous. She was wondering if he knew the truth.

The first sentence is grounded in John’s perspective. The second can still plausibly be his interpretation. The third crosses a boundary. John cannot directly access Mary’s thoughts, yet the sentence presents them as fact. This is a small violation, but it destabilizes the reader’s sense of where they are.

Individually, moments like this are easy to overlook. Repeated across a manuscript, they create a pattern of perspective instability.

These are detectable. You can track where viewpoint is established, where it shifts, and whether those shifts are signaled. You can identify sentences that introduce ungrounded internal states or externalized knowledge. These are not matters of taste. They are structural inconsistencies.

Continuity as Constraint Enforcement

Continuity is the internal consistency of the fictional world across time.

From a measurement perspective, continuity breaks occur when previously established information is contradicted, ignored, or unresolved in ways that violate the expectations the text itself created.

Consider a basic continuity failure: in chapter four, a character suffers a severe injury and can barely walk. In chapter seven, the same character runs up a flight of stairs without comment. The issue is not realism. The issue is that the manuscript established a constraint and then ignored it. The reader does not consciously log the discrepancy, but something registers as off.

Or consider a narrative gap: a mysterious key is introduced early, described in detail, and clearly positioned as important. It is never mentioned again. The text created an expectation of resolution and failed to meet it. That is a structural omission.

Continuity can be tracked by mapping entities, states, and events across the manuscript. When a state changes without cause, or persists when it should not, that is a detectable inconsistency. When an introduced element does not resolve within the narrative structure that gave it weight, that is a detectable gap.

Readers do not consciously log these failures. But they register them, and their behavior changes.

Pacing as Information Flow

Pacing is often described in terms of speed. That framing obscures what actually matters.

Pacing is the rate at which meaningful change is delivered to the reader.

From a measurement standpoint, this can be approximated by tracking consequence density: how often does something occur that alters the reader’s understanding of the situation, the relationships between characters, or the stakes of the scene?

Consider this example: two characters sit in a room and talk for five pages. The dialogue is polished. The sentences are clean. But by the end of the scene, nothing has changed. No new information alters the situation. No decision is made. No tension escalates. Structurally, that scene has near-zero consequence density. The reader is processing language without receiving narrative movement in return. That imbalance produces boredom.

At a larger scale, pacing failures emerge as disproportion across the manuscript: an opening that spends three chapters establishing a situation that could be established in one, a middle that expands without escalation, an ending that resolves before the emotional groundwork supports it.

These patterns can be identified by analyzing where meaningful changes occur and how they are distributed.

Why This Matters Before Publication

Most of these failures are difficult for authors to detect because they are not reading the same text the reader is reading. They are reading a completed version of the story that exists in their own mind, using the manuscript as a trigger.

That gap cannot be closed by effort.

It can only be addressed by introducing an external system that evaluates the manuscript on its own terms, without access to the author’s intent. This is what the bridge inspector does that the bridge’s designer cannot: examine the structure as it exists, not as it was imagined.

Human editors can approximate this. Beta readers can too, though inconsistently, and usually late in the process when the cost of revision is highest. If structural failures are detectable as patterns, they can be evaluated earlier, more consistently, and before the reader who would have loved your book closes it in chapter three and never comes back.

What FictionMark Does

FictionMark is a structural analysis tool built specifically for fiction manuscripts. It does not evaluate whether a story is original, commercially viable, or well-written in the literary sense. It evaluates whether the manuscript maintains structural integrity across the domains that reliably predict reader abandonment: perspective stability, continuity, and consequence density.

It asks the narrower question that editors often skip and beta readers rarely articulate: does this text hold together under the conditions a real reader will experience?

The analysis is AI-powered, which is what makes it possible to examine an entire manuscript systematically rather than sampling. A human editor working at that level of structural detail across a full-length novel would cost hundreds of dollars and weeks of calendar time. FictionMark delivers a full structural report for $25, typically within minutes of submission.

That is not a free service, and it is not meant to be. Structural analysis at manuscript scale requires meaningful computation, and the report you receive reflects that. What it costs less than is discovering the same problems after publication.

If you want to see how your manuscript holds up before readers do, visit FictionMark.com.

In the next piece, I will walk through extended, real-world examples of these failures and how small deviations accumulate into reader abandonment.

Subscribe now

Share

The abandonment typically doesn’t happen on page one, where the reader hasn’t yet committed. It happens after the initial investment, once the reader has decided to give the book a real chance but before the narrative has locked them in. Chapter three is roughly where that window closes. What happens in that window is almost never about premise.

It’s about trust. And in fiction, trust is structural.

What Trust Actually Means in Fiction

Trust in fiction doesn’t mean the reader likes your characters or cares about your plot, though those matter. It operates at a lower level. It is the reader’s ongoing, mostly unconscious assessment of whether the author has a reliable grip on the world they’ve created.

When that grip slips, readers experience it as confusion. And confusion is psychologically aversive. People have a strong cognitive aversion to incoherence: when something doesn’t hold together, the mind treats it as a problem to solve. In ordinary life, that response is useful. In fiction, it is fatal to engagement, because the reader’s attention shifts from inhabiting the story to auditing it. They stop experiencing and start evaluating. And once that shift happens, the author has already lost.

That feeling you get when you put a book down thinking it just isn’t compelling is not usually a judgment about the story idea. It is the residue of accumulated small betrayals of trust that you couldn’t name in the moment but that added up to a decision.

Three specific structural failures trigger this response more than any others. They are not the only ways to lose a reader, but they are the most common, and they share one feature: they are invisible to the author and register immediately in the reader’s experience.

The First Failure: Perspective Drift

Point of view is not a stylistic preference. It is a contract. When you establish whose head the reader is inside, you are telling them where they are located in the story’s world and whose perceptions they can trust. When that contract breaks, even subtly, the reader loses orientation.

Severe head-hopping is easy to identify. Perspective drift is not. A slight intrusion of another character’s internal state. A narratorial observation that couldn’t belong to the viewpoint character. A judgment that slips in from nowhere. These are small cracks, and they accumulate. Readers don’t usually identify them consciously. They just feel the experience becoming less immersive, slightly exhausting in a way they can’t explain. They start skimming.

Skimming is not low engagement. It is defensive reading. The mind is protecting itself from further disorientation by reducing its investment in the text.

The Second Failure: Continuity Breaks

Fiction creates an internal world with its own logic and its own rules. Readers don’t need that world to resemble reality. They need it to resemble itself. Genre readers are extraordinarily forgiving of premises that violate physics or history. They are not forgiving of premises that violate their own established internal logic.

A character injured in chapter four who lifts heavy objects without comment in chapter seven. A timeline that doesn’t hold. A plot thread introduced with apparent significance that disappears without resolution. These continuity breaks signal a lapse in authorial control. And once the reader detects that signal, they start asking what else they missed.

At that point the reader is auditing the text rather than inhabiting it. Auditing is the end of immersion, and immersion is what fiction is selling.

The Third Failure: Pacing Collapse

Pacing is widely misunderstood as a question of speed. It isn’t. Pacing is about consequence density: how much that matters happens in a given unit of reading time. A slow scene can carry high energy if every moment has weight. A fast-moving scene can be completely dead if nothing in it connects to anything the reader cares about.

Consider a conversation scene where two characters talk for several pages and nothing changes between them by the end of it. Not their relationship, not their situation, not the reader’s understanding of what’s at stake. The scene may be well-written at the sentence level. It is still dead weight, and readers feel that as boredom even when they can’t explain it.

The dangerous version of pacing collapse isn’t the obviously slow chapter. It’s the structurally misweighted manuscript where the proportions are wrong across the whole arc: too much time in the middle, an ending that arrives before the emotional groundwork supports it, an opening that takes three chapters to earn what it should earn in one. These are systemic problems, nearly impossible for an author to see from inside their own work.

Why Authors Can’t See These Problems

As a psychologist who has studied how people process narrative, and as a novelist who has made every one of these mistakes personally, I can tell you that the inability to see your own structural failures is not a solvable problem through effort or intelligence. It is a structural feature of authorship itself.

Authors carry the story in their heads in a way readers never can. They know what they meant, know what the character intended, know the timeline because they built it. That knowledge fills the gaps the text leaves open. The author reads the words and experiences the complete story. The reader reads the words and experiences only what’s on the page.

You cannot unknow your own story. You cannot read your manuscript the way a stranger reads it. The gaps that disorient readers are invisible to you precisely because you filled them before you ever wrote the first sentence. Human editors provide an external lens when they are skilled and when you can afford them. Beta readers provide an approximation of it. But neither is consistent, systematic, or available at the moment when it would do the most good: before the manuscript reaches readers and the damage is already done.

A Broken Signaling Market

The ebook market largely ignores a structural quality problem that affects readers every day. The volume of self-published fiction has made selection genuinely difficult. Covers, blurbs, and reviews are gameable and, as a result, frequently gamed. Readers have been burned enough times by books that looked credible and fell apart in the reading that a meaningful segment of them now approaches unfamiliar indie authors with calibrated skepticism.

That skepticism is rational. It is the correct response to an environment where quality signals are unreliable. The problem for authors who are actually doing the work is that they have no credible way to distinguish themselves from the noise. The signaling tools available to them are the same ones bad actors use. Saying your book is good means nothing when everyone says their book is good.

What’s missing is a signal that is criteria-based, applied consistently, and not gameable. One that says nothing about taste or commercial potential, but makes a specific, verifiable claim: this manuscript was evaluated against defined structural standards and it holds together.

That gap is what the FictionMark.com app was built to fill. It is an AI-assisted manuscript evaluation system that analyzes structural execution across the domains described here and certifies books that meet minimum standards of craft competence. If you want to know what a reader’s experience will actually be before they have it, that is what the system is designed to tell you.

The Stakes

Readers abandon books for structural reasons long before they consciously know why. They experience it as boredom, or vague dissatisfaction, or the sense that their time could be better spent elsewhere. They rarely trace it back to a continuity break in chapter two or a perspective slip on page forty. They just leave.

Authors who want to keep them need tools that operate at the same level the problem does. Not at the level of taste, or premise, or marketing. At the level of structural execution, which is where trust is built and where it breaks.

In the next piece, I'll show exactly how these failures are measured.

E.S. Martell is a novelist and Doctor of Psychology. Four of his science fiction titles, including Dustfall and the Time Equation Series, are Florida Authors and Publishers Association award winners.

Subscribe now

Share