I paid for forty-one sources on what trafficking does to people. Then I held my own character up against them, and the first thing I found was that she recovers in a way nobody in the literature recovers.
Her name is Rocket. She is trafficked north out of Managua, sold through the Mexican corridor, and buried in the Sonoran desert outside Douglas, Arizona. She digs herself out as a vampire. The draft of her book is finished, published, and quotable line by line.
Yesterday Claude and I created an Intel Brief to verify that I was writing her correctly. Psychological Markers in Trafficking Victims, 2000-2025: forty-one sources graded on the Admiralty scale, every analytic judgment carrying an explicit confidence level, the whole thing run back through a five-way verification pass on a second model that produced eleven corrections and reversed two of my conclusions. I filed it as worldbuilding source material. What I had actually built was a scoring rubric, and I had already written the thing it would score.
Here is the finding. Rocket is accurate everywhere she has no choice and invented everywhere she acts. The departures sort onto exactly two axes, agency and legibility, and neither one is a research failure. They are the price of a protagonist, and every author writing a trafficking survivor as a lead pays that price whether or not they know they are paying it.
One caveat has to land before any of the rest of this counts, because flattening it would be the exact failure the brief spent a verification pass correcting. The brief’s own central judgment sits at Low confidence and says there may be no psychological signature specific to trafficking at all. The study most often quoted for that null, a UK matched cohort, matched its cases and controls on primary diagnosis, which means it could not have found a diagnostic difference if one had been sitting in front of it waving.
That is not a null result. That is a blind spot. Two of the three located comparison studies actually capable of finding a difference found one, and both found it in the direction of greater severity.
So Rocket matches a profile. Not a trafficking profile, because no such thing has been established, but the profile of prolonged inescapable abuse, which is established and which trafficking happens to deliver people into. That is ICD-11 Complex PTSD. Brewin reproduces the definition: the associated events are “most commonly prolonged or repetitive events from which escape is difficult or impossible (e.g., torture, slavery, genocide campaigns, prolonged domestic violence, repeated childhood sexual or physical abuse).” Trafficking is not on that list. It is a legal category that produces the conditions on it.
The distinction has teeth in the data. When Ottisova, Smith and Oram counted Complex PTSD symptom domains in trafficked children against non-trafficked controls, the trafficked children grouped with the multiple-trauma controls, not apart from them, and both separated from children exposed to a single traumatic event. The line the data draws is prolonged versus not. It is not trafficked versus not.
So she matches the prolonged-abuse profile, and she matches specific well-evidenced findings inside it. Where those findings are thin I am going to say so.
One source got demoted on the way here, and it is mine to report. Trauma and You, where I first argued that trauma is architecture rather than damage, rested on Katheryn Greenleaf’s “The Neuro-Adapted Brain,” which is a Substack post. Her vocabulary is still doing work in this piece: hypervigilance, intrusive memory, avoidance, shame architecture, the whole feature list of an operating system nobody asked to install. The citation is not load-bearing any more. A piece arguing for source discipline cannot itself rest on a platform essay, so Greenleaf comes down to background and every structural claim below traces to a named study with a name, an institution and a sample size attached.
What She Does That Nobody in the Literature Does
Rocket kills her trafficker. She burns Carlos Mendoza down to ash on a limestone ridge, and the bond he built in her dies in the same scene.
Casassa, Knight and Mengo searched ten databases for literature on trauma bonding in sex trafficking. Fifteen articles qualified. Their closing observation is the one I could not write around: “No article indicated how trauma bonds could be severed and replaced with healthy attachments.” Not one paper described the exit.
My novel describes it in about nine hundred words, and it takes an afternoon.
That is the agency axis, and it is not a small thing. The bond in that literature is built from an imbalance of power, the trafficker’s deliberate alternation of reward and punishment, the victim’s gratitude for the reward and self-blame for the punishment, and the victim’s internalisation of the perpetrator’s view of her. Four features, fifteen articles, no documented off switch. Rocket flips all four in one night by admitting out loud what was done to her.
I owe my own chapter one qualification, because overstating the deviation would be its own species of cheating: the severing is not clean and it does not hold. The scene closes with her already rebuilding the distance she just tore down: “Maybe it didn’t happen like that. Maybe I imagined the worst parts.”
The wall re-forms before dawn. Carlos is what dies in that scene. What she wins is a single night of ownership she cannot keep.
Now the honest update, which I went looking for during drafting and which is not in the brief. The same lead author later published the paper her own review said did not exist. Casassa, England and Karandikar interviewed nineteen women who were participants in or graduates of a human trafficking specialty docket, and asked how they had experienced healing from trauma bonding. Three themes came back: building trusting and honest relationships, the survivor’s relationship with herself, and education.
Read that list beside my chapter. Nothing in it happens in an afternoon. Nothing in it involves a body count. The severing is documented now, and it is documented as slow, relational, unglamorous work done by a person who is still alive and still in a program.
In her defense, and in mine, she is a vampire and this is a novel. A finding about how human beings recover from a trauma bond is not contradicted by a girl who burns a man to death with her hands, because the mechanism she uses does not exist, and you cannot falsify a scoping review with a fantasy. I am not claiming to have found something nineteen women and their clinicians missed.
So name the charge precisely, because it is not the fire. Mechanisms are what the genre is for. What I took was the tempo.
Nothing about being a vampire required that bond to break in one scene, on one night, in a single act of recognition, and I could have written a vampire who still caught herself defending Carlos two years later, who took the slow relational route with fangs and got there in book four. The premise had no opinion either way. I chose the fast version because the fast version is a scene and the slow version is a subplot.
The obvious objection is that I have described the requirements of commercial fiction and called it a discovery. Every protagonist has agency. That is roughly the definition of a novel.
Granted, and that is the point. The trade itself is old news. What is not old news is that on this subject the author generally cannot see himself making it, and that the cost is specific enough to itemise, which is what the rest of this article does. A reader meets Rocket and learns something true about coercion and something false about recovery.
What She Does Not Keep
The second axis is legibility, and it is the one I did not see coming.
Depression outranks PTSD in every prevalence study I could locate on this population. Usually by a wide margin. Kiss and colleagues interviewed 387 children and adolescents in post-trafficking services in Cambodia, Thailand and Vietnam: depression 56.3 percent, anxiety 32.6 percent, PTSD 25.5 percent. Palines and colleagues reviewed the medical records of 143 sex trafficked children in Wisconsin and found depression at 45.5 percent against PTSD at 19.6, and Hopper and Gonzalez, working with 131 US survivors of both sex and labor trafficking, found depression at 71 percent against PTSD at 61.
Rocket has no depression in the novel. She has hypervigilance, intrusive memory, dissociation, a shame architecture you could survey from orbit, and a mouth. She carries all three disturbances in self-organisation the diagnosis requires: affect dysregulation, a negative self-concept, wrecked relationships. She does not carry the depression Brewin records travelling with them.
That claim needs its boundary drawn, because the novel and the record barely overlap. Chapter 1 opens with her clawing out of her own grave. Everything the studies measure happened before that and reaches the book only as intrusive memory, which means I have been scoring a novel for the absence of symptoms across a stretch of time it never dramatizes.
The trafficking is written. It is written as twenty first-person notes that run between the Prologue and Chapter 1, published alongside the book and not part of it. I had not read them against the literature until I wrote this paragraph.
They contain the thing I told you was missing. In the Hermosillo note she goes all the way down in a safe house, and the prose does not flinch: “I’m not going to lie to you and say I didn’t want, in that hour, to just not be anymore.” Kiss puts suicidal ideation at 15.8 percent of that sample. It is on the page, in her voice, and if I had stopped at the novel I would have told you she never carries it at all.
Then read what happens next. “I didn’t climb back up out of despair at Hermosillo. I went through the floor of it and found rage on the other side, waiting, patient, mine.” The despair holds for one paragraph. It converts into hate, which is an emotion that generates scenes.
So this is not an omission anyone can accuse me of. It is a conversion anyone can watch me perform. I wrote the depression, held it for the length of a paragraph, and turned it into the most actionable feeling available, because a girl who has found her rage can carry a book and a girl still lying on the floor cannot.
The cleanest account of all this is one sentence. The human was depressed. The vampire is not.
That is period-correct and it fits the text. Naida broke on a safe-house floor in Hermosillo, and Bisbee renames her in Chapter 11 because the girl who went into the ground is not the thing that climbed back out of it.
It does not get me off the timeline, though, because the timeline is the one place the record is unambiguous. Kiss, Abas, Hopper and Palines all sampled people after exit, in post-trafficking services. Every depression figure I quoted measures the after. Her after is the vampire.
And the account asks for a rule I had never written down. Depression would have to be the kind of thing a dead brain stops running, while conditioning is the kind that survives being dead. That line is defensible and I would defend it. Until I wrote this section it appeared on no page of mine.
The same book argues the other way in the scene this article exists to reach. Bisbee has been dead fifty years and still brews coffee every night for men killed in Vietnam. Rocket’s conditioning fires hard enough to move stolen blood through a corpse. Chapter 11’s entire thesis is that the Embrace clears nothing.
Holding both requires a split I happen to like: the Embrace erases chemistry and keeps training. I think that is true and I think it is the better rule, so it went into the worldbuilding file while I was writing this paragraph, with the worked cases and the boundaries against the two mechanisms it gets confused with. Which fixes the series and does not fix the book. The fire earned its defense because the mechanism came first and the scene followed from it. This one arrived in the opposite order, several hundred pages after it needed to already be true, and that is the whole difference between a premise and an alibi.
What She Reaches For Instead
Substance use has less to complicate it. In the Arizona statewide study of 309 identified child sex trafficking victims, 45 percent had used drugs or alcohol, and in the ASU Youth Experiences Survey of 167 homeless young adults in Phoenix and Tucson, drug use ran to 59.3 percent. Rocket never used, anywhere she is depicted, and she crossed the border with parcels taped to her torso, so it was not for want of proximity.
Two defenses hold there and I will take both. Self-medication is learned, and she had never tried any of it, so the association that makes it attractive was never built. Carlos controls through candy and conditioning rather than chemistry, which is the Biderman set exactly and is coherent design rather than convenience.
She is still not uncoping. She has a coping mechanism and I gave it to her on purpose. Even dead, even with two rifles trained on her from thirty meters out, “her body reached for the leash before it reached for the throat.” The reflex fires on schedule. It goes to charm.
The choice was never numbing or nothing. It was which tool the operating system reaches for, and I handed her the one that generates dialogue over the one that flattens it. Manipulation photographs well. That is the same move as the rage, arriving one resolution finer, and it shows up in what I substituted rather than in what I left out.
None of it is settled either. Blood carries whatever was in the person it came from, she does get high off someone else’s later in the series, and seeking out addicts is not out of the question for her. It has only not been written yet, which means this audit carries a timestamp.
Call the whole thing what it is. Despair flattens affect, which is the raw material of dialogue, and dependency opens a second plot that competes for oxygen with the one I am trying to write. Both are enormously well evidenced. Both are, in narrative terms, unreadable, and what I kept instead are the symptoms that photograph well. That is a legibility filter, applied by me, on clinical grounds I did not have.
The One I Am Conceding
Rocket is with Carlos for three weeks. She says the number herself, screaming it at him through the fire: “I’m the girl you raped for three weeks!”
I could recruit that number into the thesis and I am not going to, because it would be dishonest and because conceding it is what makes the rest of the sort credible. Three weeks is plot economy. It is short because the book needed her out of that house and into the desert, and for no other reason.
The record does not support the shortness. Abas and colleagues ran the only located study that assigned DSM-IV diagnoses by structured clinical interview conducted blind to exposure history, on 120 women returned to Moldova. They report an adjusted odds ratio for trafficking duration of 1.12.
I originally read that as a small effect and concluded duration did not matter. Wrong, and the verification pass caught it. Duration was entered as a continuous variable in months across a two to thirty-one month range, so 1.12 is per month, which compounds to roughly 3.9 times the odds at twelve months. The point estimate is large. The study, at n = 120, was underpowered to resolve it, which is what p = 0.089 records rather than a null.
The honest statement is that duration probably does matter and that nobody can currently say by how much, which makes my three weeks not a defensible reading of a contested literature. It is a scheduling decision.
The Half That Is Less Flattering
Now the part that reads like a victory lap and is not.
Rocket’s predisposing history is childhood sexual abuse by an older cousin, disclosed to her mother and disbelieved. Abas puts childhood sexual abuse at an adjusted odds ratio of 4.68 for any DSM-IV disorder post-return, the dominant term in the entire risk model. Reid and colleagues compared 913 Florida youth with accepted trafficking abuse reports against a matched sample and found sexual abuse history the strongest predictor of trafficking itself, at 2.52 times the odds for girls. Laird and colleagues at Deakin pooled 37 studies covering 67,453 participants and returned a pooled odds ratio of 3.80 for childhood sexual abuse and 5.29 for PTSD. Three research groups on three continents, all landing on the same predictor.
I wrote that backstory years before I read any of it. I wrote it because it made her manipulative streak legible, which is a craft reason, not an evidentiary one.
She also does not call herself a victim. Mahon and colleagues surveyed 74 women across three Ohio sex trafficking specialty courts, all of whom met the legal criteria: “Before court programming, 43.2% (n = 32) of participants did not see themselves as a victim or survivor of sex trafficking.” Rocket’s version is a change of grammatical person. “That happened to someone else. Some other girl. Not Naida.”
That line is worth holding against MacCabe and colleagues, who administered the DES-II to 52 adult sex trafficking survivors and found depersonalisation, derealisation and absorption more prominent than amnesia. She has not forgotten. She has relocated the event to a third party, and the material keeps surfacing on its own without ever landing: “The fragment hit her like a slap and disappeared just as fast.” That finding is graded Low confidence and it does not get promoted here for my convenience. Fifty-two people, purposive sampling, no comparison group. It is one small study whose sub-scale pattern happens to describe exactly the mechanic I built, which is a coincidence I would rather report than lean on.
The self-concept has a receipt too. Perry and colleagues analysed baseline data from 110 young people aged 11 to 19 with substantiated commercial sexual exploitation entering trauma-focused therapy, and found 51.8 percent meeting clinical criteria for post-traumatic cognitions: not symptoms, but what the person has come to believe about herself. Note what predicted it. Age, and the number of trauma categories experienced. Not any feature of the trafficking itself, which is the polyvictimisation finding surfacing again, and it is the reason the brief cannot resolve whether trafficking carries a signature of its own.
The 1957 Chart I Reinvented Without Reading It
Carlos runs a method. I built it from the inside, scene by scene, out of what a man like that would plausibly do to a girl he intends to sell.
Albert Biderman published the method in 1957.
Biderman was working for the Air Force on how Korean War captors produced compliance in prisoners without leaving marks. He catalogued eight non-physical tactics. Baldwin, Fehrenbacher and Eisenman later interviewed twelve women trafficked into Los Angeles County from ten countries and analysed the accounts against that framework, reporting that “Participants reported experiencing the range of nonphysical coercive tactics outlined by Biderman.”
Scored against the two chapters that carry the safe house, tactic by tactic:
- Isolation. Present. “The hollow ache of homesickness mixed with betrayal when she realized no rescue was coming.”
- Monopolisation of perception. Present. Carlos explaining “why submission was her only choice, why cooperation was survival.” Yellow walls, Disney princess stickers, counting to one hundred.
- Induced debility or exhaustion. Absent. No starvation, no sleep deprivation, no physical wearing down, anywhere in the chapters or in the rest of the manuscript.
- Threats. Present, and this is the one I would score as arguable so you can mark it down if you disagree. It is not staged as a threat scene. It arrives inside an endearment: “The ones who don’t learn don’t make the crossing.” Comply or die in the desert, said gently.
- Occasional indulgences. Present, and it is the spine of the whole sequence. The chili-mango lollipops. “You’d ask so prettily when you wanted something sweet.”
- Demonstrating omnipotence and omniscience. Present. “I can hear it in your breathing.” “I taught you what you are.”
- Degradation. Present. Merchandise, cargo, “You understood exactly what you were for,” and a cartel property brand on the inside of her wrist.
- Enforcement of trivial demands. Present. “Gracias, Papi” after every lesson, manners graded and rewarded.
Seven of eight. Not item for item, and the missing item is not decorative: induced debility is one of the load-bearing tactics, and I left it out because a starving protagonist is a slow protagonist. Same filter, third appearance.
The occasional-indulgences result is the one that unsettles me. Of the eight, it is the item that carries the most explanatory weight and gets the least attention, because unpredictable intermittent reward is the mechanism that converts fear into attachment. Casassa’s second bond feature names the same mechanism from the other end, arrived at by a social-work review that never cited an Air Force report. I wrote the lollipops as a detail. They are the load-bearing element in the coercion literature, and I put them in because they felt right.
There are not many ways to do this. The people who do it found them first, and a sociologist wrote them down before my mother was born. That is the least flattering discovery in the audit, and it is not a compliment to my imagination. It is a statement about how narrow the design space is.
The Audit Found My Notes Wrong Too
The fiction was not the only thing on trial. Rocket’s psychology also lives in a database, in structured fields, one of which holds a locked description of what her mind does with the trafficking, and that field is what I write from when I have not reread the chapter in six months, which means it silently governs every scene downstream of it.
It disagreed with the published prose in two places, and I only found out because I ran the comparison.
The field said her trafficking trauma was blocked entirely. It is not. Chapter 7 shows her retrieving the safe-house material in full under pressure, in order, with the taste and the wall colour and her own voice intact. What is walled off is not the file. It is the ownership: whose body that was. The field also recorded the road and the house as two separate blocked stretches, and the manuscript treats them as one continuous run from the truck to the yellow walls.
Both are now corrected. That is four hours of work, no reader would ever have caught either one, and the version of this exercise where I only audit the fiction would have left them sitting there indefinitely, quietly generating wrong scenes for as long as the series runs.
The Trigger Is Kindness
Chapter 11 is the one I actually wrote this piece to reach.
Rocket showers in a stranger’s house. He has left her clean clothes outside the door and asked for nothing. Her body responds, which is flatly impossible for what she now is, and she notices while she is drying her arms.
She had not been running a play. “A man being decent to her ... was the precise input required to fire subroutines she hadn’t summoned.” The old vampire in the next room gives her the reading: “That’s conditioning so profound it defeats the undead biology. Your body forcing stolen blood to flow where there shouldn’t be any, because that’s how you survived.”
I have a standing rule that vampires in this world are dead from the waist down. I broke it once, on purpose, for her. Carlos alternated cruelty with candy and she was made grateful for the candy. That is bond features two and three, and he wired the reward to himself. What Rocket demonstrates is that the wiring outlives the man, outlives the context, and in her case outlives the circulatory system.
The trigger is not the abuser. The trigger is being treated well.
The Survivor Who Did Leave
Somewhere in the readership of this piece is a woman who got out, who testified, who is not bonded to anyone, and who has just watched me file her recovery as a departure from realism.
I am not going to hedge my way past that. The record describes central tendency. It does not issue permission slips. Casassa’s 2022 finding was that no reviewed article described how the bond is severed, not that severing it is fictional, and her own 2024 interviews with nineteen women found exactly that severing, achieved through relationships, self-regard and education rather than through fire. The literature is a description of what has been measured in the populations that reached a service. It is not a description of what a person is allowed to do.
Every prevalence figure quoted above comes from someone already in a shelter, a court, a clinic or a mental health service. That is logged as the first key assumption in the brief. The direction of the bias is unknown. Genuinely unknown, not politely unknown. People who never reached a service are absent from every number in this article, and nobody can tell you whether they are doing better or worse.
What I can tell you is that I traded on those numbers, and this is me reporting the trade rather than waiting for somebody to catch it. The case for why fiction gets to go near this material at all is one I made elsewhere and am not going to relitigate here. This piece does the narrower job: one character, one literature, and the score.
The received advice about writing this material is to do the research, write responsibly, and hire a sensitivity reader. All posture, no mechanism. Nobody checks a specific character against a specific literature and prints the result, because the result is embarrassing in a way the advice does not prepare you for, and because there is no version of the exercise where you come out looking careful.
The two documents were never describing the same object. One describes people under conditions that removed their capacity to act. The other requires a person who acts. They agree perfectly right up to the moment she moves.
The record and the novel agree on every page where she is a victim. They part company on the page where she stops being one.
You may also like: - What the Desert Does With Children Nobody Is Looking For or What the Books Are For - Trauma and You... - Vampires: Victorian Necrophilia, or the Sentence I Never Finished