A bored agent stands in a doorway on Gorokhovaya ulitsa, writing down the time a fat peasant came home drunk and who he brought with him.1 He has no opinion about Rasputin. He is filing a report for a recipient — the tsar — who will glance at it, shrug, and do nothing. That agent's notebook is the single most trustworthy thing ever written about Grigory Rasputin. Every glittering memoir written by someone who actually knew the man — his daughter, his secretary, his murderer, his enemy — is worth less than the doorway log kept by a man who didn't care.
That inversion is the engine of Brian Moynahan's whole book. He never states it as a rule, but he runs it on every page: sort the sources into two tiers, and when they fight, let the bored watcher win. Tier A is the external record — the Okhrana surveillance logs that tracked Rasputin's movements, his visitors, his benders, his women, and the 1917 Extraordinary Commission depositions taken under oath while the participants were still rattled and the regime that protected them had just fallen.2 Tier B is everybody's memoir. The two tiers are not equal, and Moynahan refuses to pretend they are.
The reason the watch outlives the confession is motive. The Okhrana agent had no story to sell. He gained nothing whether Rasputin was a saint or a fraud; he was paid to record arrivals and departures, and the report went to a reader who never acted on it.1 A memoir is the opposite of that doorway. Every memoirist who knew Rasputin had a reason to shape him — an alibi to build, a friendship to defend, a book to sell, a hatred to feed. Moynahan's two-tier model is the discipline of trusting the indifferent record over the invested one, and reporting the gap when they diverge instead of smoothing it shut.
The model runs on a peculiar feedstock — a man at the center of a court that lied reflexively, surrounded by people who either worshipped him or wanted him dead, in an empire that censored its own press and then collapsed.3 Almost nobody who left a record was neutral. That is the problem the two tiers are built to solve.
Tier A ingests two kinds of cold material. First, the surveillance archive — "Rasputin in the Eyes of the Okhrana," the agents' day-to-day reports of who entered the apartment, when, of what rank, and what time the lights went out.4 The agents noted the social standing of visitors where they could, ran informants inside the building (a maid, the concierge, the gatekeeper), and in Moscow even followed Rasputin into the streets outside his nightclubs.5 Second, the depositions — the Extraordinary Commission of 1917, where village witnesses to his childhood, ministers, priests, and Vyrubova herself testified under oath after the autocracy fell.2 Both are records produced for a purpose other than burnishing or burying the man.
Tier B ingests the people who knew him and could not help themselves. Felix Yusupov, who killed him and then wrote two versions of the killing.6 Aron Simanovich, his secretary, who portrayed his friend kindly and inflated his own importance at every turn.7 Maria Rasputin, the daughter, who gave her father visions in the Garden of Gethsemane and prophetic alarms on a riverbank.8 Iliodor, the defrocked monk who tried to have him murdered and wrote a book called The Holy Devil.9 Anna Vyrubova, the bridge to the throne, whose memoir reads like devotion in print. Each one is a primary witness. Each one is also a motivated one. Moynahan never throws the memoir away — he reads it, then turns to the doorway log and checks.
The mechanism has three moves, and Moynahan runs them in order every time.
Move one: classify by distance from motive. The further a source sits from a personal stake in the story, the higher it rises. The Okhrana agent sits furthest — paid, indifferent, reporting to an idle reader. The sworn deposition sits close behind, taken under oath in the wreckage of the regime. The memoir sits at the bottom, written for money, vindication, or revenge, often years later in exile. This is not a ranking by who was closest to Rasputin. Maria, Simanovich, and Yusupov were far closer to the man than any agent in a doorway. Moynahan inverts the obvious hierarchy: proximity to the subject is a liability, because proximity is where motive lives.
Move two: cross-check Tier B against Tier A. A memoir claim only earns trust when the cold record confirms it. When Simanovich describes Rasputin's humanitarian streak toward petitioners, Moynahan notes the Okhrana agents' own comments on Rasputin's treatment of petitioners corroborate it — so the claim survives, even though Simanovich "had an interest in portraying his friend in a kindly light."10 When Djanumova describes a visit, the secret police records confirm it happened.11 The memoir is not believed because the memoirist swore to it; it is believed because the watcher in the doorway saw the same thing.
Move three: when they contradict, privilege the surveillance and report the gap. This is the move that defines the model. Moynahan does not resolve the contradiction by picking a winner and erasing the loser. He privileges the external record, then leaves the wound open on the page. He tells you Maria's account of watching her father leave the apartment the night of the murder is contradicted by her own deposition interview, and he says so plainly rather than choosing one.12 He tells you Vyrubova's and Spiridovich's accounts of the Siberian trip "differ in almost every respect other than that the visit took place," and he refuses to manufacture a single reconciled version out of two that share only the bare fact of occurrence.13 The gap is data. He keeps it.
This page is the epistemic spine the rest of the vault's Rasputin corpus hangs on. Every concept page drawn from Moynahan inherits its confidence rating from this model: a claim resting on the Okhrana log or a sworn deposition carries weight; a claim resting on Yusupov, Simanovich, Maria, or Iliodor alone carries a [SINGLE SOURCE] flag because Moynahan himself flagged the chronicler as motivated.
To the vault's surveillance pages — The Okhrana Surveillance Logs, Okhrana Surveillance Architecture — this page supplies the why it matters for truth: those logs are not just an instrument of the police state, they are the load-bearing floor of everything we can say about Rasputin's daily life with any confidence. To The Extraordinary Commission of 1917 it supplies the same: the depositions are Tier A not because the Provisional Government was virtuous but because oath plus fresh memory plus a fallen protector strips away the motive to flatter.
And it hands the vault's multi-warrant architecture a concrete instance. Radzinsky works the inside — the File, the testimony, the man's interior. Moynahan works the outside — the watcher's log, the indifferent record. The two warrants triangulate the same Rasputin from opposite vantages, and where they converge, the vault holds a fact rather than a story.
Take the single sharpest test of the model — the women at the murder. The two killers, Yusupov and Purishkevich, both wrote accounts of the night they killed Rasputin in the Yusupov palace cellar.14 Moynahan notes their versions "correspond broadly with each other and with the autopsy." On the mechanics of the killing, Tier B agrees with itself and with the forensic record. By the model's own logic, that convergence earns trust. Two motivated chroniclers and a corpse tell the same story about how the man died.
Then comes the gap. Neither Yusupov nor Purishkevich mentions that there were women in the palace that night. The Tier A record does. Constable Yefimov, one of the policemen first aware of the disturbance, gave a deposition — recorded verbatim by the secret police chief Vassilyev from the original statements — describing a cry "as if uttered by a woman."15 The killers' polished narratives are silent exactly where the bored constable's deposition speaks. Moynahan does not decide who the women were or invent a scene to cover them. He sets Yefimov's "as if uttered by a woman" beside the killers' silence and lets the contradiction stand as evidence that the confessors curated their confession.
Why does this matter beyond a true-crime footnote? Because it shows the model catching a lie of omission — the hardest kind. Yusupov and Purishkevich did not contradict the surveillance record; they edited around it. The two-tier method exposes the edit precisely because the doorway witness had no narrative to protect and therefore no reason to leave the woman out. Moynahan sharpens the point on Yusupov's alibi: Yusupov claimed Rasputin telephoned "insisting that I go with him to the Gypsies," a transparent attempt to build cover, which Vassilyev's record undercuts.16 And the excuse that Purishkevich was "very drunk" is, Moynahan says flatly, "feeble; he was known not to drink."16 The killers' story is a constructed thing with seams, and the cold record is where the seams show.
The case study is the model in miniature: trust Tier B where it converges with Tier A and the autopsy (the mechanics of the killing), distrust it where it diverges from the deposition (the erased women), and never paper over the divergence. The watch outlived the confession even on the night the watch's subject was murdered.
You are at the desk with three documents open. On the left, the Okhrana report: 15 December, returned 2 a.m., drunk, accompanied by woman, rank below nobility. No adjectives, no drama, no Rasputin-the-saint, no Rasputin-the-devil — just a time and a rank, filed by a man who would rather be home.4 In the middle, Maria Rasputin's memoir, where the same father is dreaming of God since childhood and weeping over the moral filth of Mount Athos.8 On the right, Yusupov's Lost Splendor, where the killer narrates himself into a hero of the realm.6
You do not start with the memoirs, however vivid they are. You start on the left. The agent's log fixes the floor: where he was, when, with whom, how drunk, what rank his visitor held. That is the skeleton you can stand on. Then you pick up the middle document and ask one question of every sentence — does the doorway log confirm this, or is the daughter the only one saying it? Where the log nods, the claim stays and you build on it. Where the log is silent and Maria alone insists, you write the claim down with a tag on it: Maria says, and no one else does, and Maria gave her father visions. You do not delete her. You quarantine her.
Then the document on the right, the killer's confession. You read it against the autopsy and against Purishkevich, and where all three agree on the mechanics, you let it stand. Then you reach for Yefimov's deposition and ask what the confession left out. The woman's cry. The phantom phone call to the Gypsies. The non-drinker who was "very drunk." Each silence in the confession that the cold record fills is a place you write the killers edited here.
What you never do is the thing that feels most like history-writing — reach into two contradicting accounts and braid them into one smooth narrative. When Vyrubova and Spiridovich differ "in almost every respect" about the Siberian trip, you do not average them.13 You report that two witnesses to the same journey agree only that it happened, and you let the reader feel the size of that gap. The discipline is to keep your hands off the seam.
The model fails the moment a historian forgets which tier he is standing on. The first warning sign is the smooth narrative — a Rasputin biography that reads like a novel, every scene fully furnished, every motive legible, no seams showing. Smoothness is the tell. It means the writer reached into Tier B for color, took the memoirist's most vivid detail, and welded it to the surveillance skeleton without flagging the weld. The reader cannot tell the doorway log from the daughter's vision because the prose has fused them.
The second sign is proximity-worship — treating the closest witness as the most reliable one. Maria Rasputin was his daughter; surely she knew him best. Simanovich was his secretary; surely he saw everything. This is exactly the instinct the model is built to resist. Proximity is where motive concentrates, not where truth does. A biography that leans on the intimates because they were intimate has inverted the tiers and will reproduce their flattery as fact.
The third sign is the silently resolved contradiction. When two sources clash and the page gives you one clean version with no acknowledgment that another existed, the historian has done the forbidden thing — picked a winner and buried the loser. The Siberian trip becomes a single confident paragraph. The murder night loses its women. The reader never learns there was a fight in the evidence, because the writer ended it off the page. Moynahan's whole method is the refusal to do this; a biography that does it has abandoned the model even if it cites the same archives.
The model is Moynahan's, and he applies it as a working historian rather than announcing it as a theory; reconstructing it as a "two-tier model" is the vault's framing of what he consistently does, not a phrase he uses.1 [POPULAR SOURCE] The reconstruction is well grounded in his Notes on Sources, where he repeatedly rates his witnesses by motive — Maria's accounts "clash with other evidence,"8 Simanovich "had an interest in portraying his friend in a kindly light,"10 Vyrubova and Spiridovich "differ in almost every respect,"13 the prophecy volume was at best "dictated" with no named drafter.17
A real tension inside the model: Tier A is not pristine. The Okhrana was a political police force, and the vault's The Shchyogolev Forgeries page is a standing reminder that documents from this period were sometimes manufactured. Moynahan's defense is implicit — the surveillance logs were internal working reports, not public instruments, and their indifference to Rasputin's reputation is what makes them hard to fake into a coherent slander. But the model would be stronger if it stated where Tier A itself can fail.
Open question: does the privileging rule ever cost truth? A memoirist sometimes knows a thing the watcher in the doorway could not see — an interior, a private word, a motive. By ranking the cold record above the intimate one, the model may systematically lose exactly the inner life that a biography of a "holy man" most needs. Moynahan accepts this trade; whether it is the right trade for this subject, whose entire claim was interior, is unresolved.
A Moynahan↔Radzinsky tension worth flagging: the vault's Radzinsky corpus works largely from the File and from testimony, the inside record. Where Radzinsky trusts a piece of testimony that Moynahan would tier down as motivated memoir, the two warrants will disagree about the same fact — and that disagreement is itself the multi-warrant architecture doing its job, not a flaw to resolve.
Moynahan is the author of this model, so the cleanest tension is internal — between his own narrative drive and his evidentiary discipline. He writes a propulsive, scene-rich book, and a propulsive book is hungry for exactly the vivid Tier-B detail the model tells him to distrust. You can feel him managing that hunger. He uses Maria's vision in the Garden of Gethsemane, then tells you in the notes it is Maria's alone and clashes with other evidence. He uses the killers' cellar drama, then turns and points at the women they erased. The convergence is that his storytelling and his skepticism end up serving the same end: the seams he refuses to hide are also the most interesting thing on the page.
Where the vault's Radzinsky warrant touches the same events, the split is illuminating rather than embarrassing. Radzinsky, working the inside, can give Rasputin an interior that Moynahan's outside-in method structurally denies. Moynahan, working the outside, can catch a curated confession that an inside-the-testimony reading might take at face value. Neither is more honest; they are aimed at different distances. The two together triangulate a man neither could fix alone — which is the entire argument for keeping more than one warrant in the vault and refusing to collapse them into a house consensus.
History asks what happened; behavioral science asks how a person hides what happened and why we believe the wrong witness. Moynahan's model lives exactly where those questions meet, because his whole craft is reading motivated testimony for the places it bends — and that is a deception-analysis problem dressed as a historiography problem.
The first handshake is to Ultimate Alibi Buster Method. Lieberman's distinction is movies versus photographs: a truthful memory plays like a movie the speaker can run forward and backward, while a fabricated alibi is built as a series of stills strung together to fake motion. That is precisely the difference between Moynahan's two tiers. The Okhrana log is a movie — continuous, time-stamped, running whether or not anyone watches, with no scene constructed for effect. Yusupov's alibi is photographs — the phantom phone call to the Gypsies, the stills he assembled to place himself elsewhere — and it breaks exactly where a constructed account breaks, at the seam the builder didn't think to fill (the woman's cry he left out). The insight neither domain produces alone: the interrogator's trick and the historian's source-criticism are the same operation run at different scales. Lieberman drops a fake detail to see if the alibi has a movie behind it; Moynahan drops the deposition next to the confession to see if the confession has continuous reality behind it or only curated stills. History is forensic deception-detection with a century of lag.
The second handshake is to Messenger Effect & Source Credibility — and here the handshake is an inversion, which is what makes it worth its place. The messenger effect says we judge a message by the status of its source: the same claim is more believed from Oppenheimer than from Pravda, from the doctor than from the politician. Human credibility-judgment tracks the prestige of the speaker. Moynahan's model does the opposite on purpose. His most trusted source is the lowest-status speaker in the entire cast — an anonymous police agent in a doorway, a man no court of public opinion would rank above a prince, a holy man's daughter, or the tsar's confidant. The intimates have all the prestige; the watcher has none. By the messenger effect, we should believe Yusupov the prince over Yefimov the constable. Moynahan flips it: rank the indifferent low-status witness above the prestigious invested one, because prestige and proximity are where motive concentrates. The insight neither domain generates alone: good historiography requires actively overriding the messenger-effect heuristic our minds run by default — the discipline is not a neutral sorting of evidence but a deliberate fight against the cognitive bias that makes us trust the famous mouth over the bored one.
A third, lighter handshake runs to Historiography as Ideology. That page tracks how the frame a historian brings determines which evidence becomes visible; this one shows the counter-move — a frame built specifically to resist the ideologies (devotion, vindication, revenge) baked into the sources themselves. The Bolsheviks needed Rasputin to be an impersonal symptom of class decay; the émigré memoirists needed him to be a devil or a saint.3 The two-tier model is the closest thing to an ideology-proof reading available: not because the police were neutral, but because their indifference to the story makes their record the one document none of the competing ideologies bothered to shape.
The Sharpest Implication. The people who loved a man, killed a man, or raised a man are the worst witnesses to him, and the bored functionary who logged his comings and goings is the best — and you run the opposite ranking every day of your life. You trust the friend's account over the stranger's, the participant over the observer, the passionate over the indifferent. Moynahan's model says that instinct is precisely backwards wherever motive is in play: closeness is contamination, and the most reliable record of your own life may be the one kept by someone who didn't care about you at all.
Generative Questions.