The Verdict Travels With The Claim

The compendium has 1 rule and the rule is 1 sentence: nothing goes on screen that is not already in the files. It is not a style guide. It is a gate, and the gate is the only reason the rest of the file is usable, because a distillation with no gate is just the most recent memory of the work wearing the clothes of a finding. Underneath the rule sits a ledger of 127 numbered validation rows, running from row 1 to row 127. Of those 127 rows, 8 were never read at primary source, and every 1 of the 8 is named rather than absorbed: row 107 is Kosfeld 2005 and it survives only as the target of the replication that failed it, row 108 is Johnson and Goldstein 2003 and its headline figure is barred from the screen, and 6 more rows are recorded with the same discipline in the validation file. 161 claims sit in 4 families, and every 1 of them carries 1 of 2 verdicts. Verdict 1 is evidenced by the primary source. Verdict 2 is framing or inference, and it is usable as a shape or as a mechanism but never as a citation. 2 classes, no third class. That is the design decision that removes the grey zone where a good-sounding inference quietly becomes a finding, and the grey zone is where almost all confident screen writing is born. 1 probability sits at the front of the file. The probability that a knowledge base with no verdict field drifts from evidence into recollection across 11 iterations is high, because recollection is cheap and a citation is not. The probability that anyone watching the finished film could tell which sentence carried which verdict is close to zero, which is why the verdict has to travel with the claim rather than wait in a separate document that nobody opens twice. 11 seams are construct-complete and the 8 open questions are all answered. Novelty read 0 for the 3rd consecutive iteration, and 0 is the number that closes a file rather than the number that opens it. A knowledge base that keeps producing new material has not finished. 1 that stops producing it has.

The spine of the whole knowledge base is 1 word. Motivation comes from the Latin movere, to move, so the study of motivation is the study of action (Eccles and Wigfield 2002, Annual Review of Psychology 53:109-132). That single etymology does more work than a definition, because it moves the question away from a feeling and toward a behaviour that can be counted. The drivers sit at 3 places. Expectancies are the can I, values are the why, and goals are the where. Both expectancies and subjective task value drive choice, persistence and effort, and neither 1 alone is sufficient. A production that spends its whole second act on the why and never touches the can I in act 1 has built half a lever and called it the machine. Quality is a separate dimension from amount. Intrinsic and instrumental motivation behave differently rather than merely more or less, and the difference runs through 3 innate needs: competence, autonomy and relatedness (Ryan and Deci 2000, American Psychologist 55:68-78). 3 needs is a small number, and it is the reason a reward can be large and still fail to move anyone. Perceived self-efficacy governs goal setting, choice, effort and persistence, and efficacy expectations are distinct from outcome expectations (Bandura 1997, read through Eccles and Wigfield 2002). Those 2 expectations get flattened into 1 in most production meetings, and the flattening is expensive: a person can believe the outcome is desirable and still be certain they cannot produce it. 1 probability closes the section. The probability that a motivational film shot without the can I, the why and the where separated ends up appealing to the people who already agree is high. The probability that it moves anyone sitting on the fence is low, and the fence is where the audience is.

Frustration, loss and suffering take 1 shape: the gap between what was expected and what arrived. Satisfaction is a function of expectation and disconfirmation (Oliver 1980, Journal of Marketing Research 17:460-469), and that 1979 finding has held up under far larger samples than the 1 it came from. A meta-analysis of 150 records and N = 58,597 confirms that disconfirmation drives satisfaction, and it also finds expectations positively correlated with satisfaction at r = 0.29, with no contrast effects (Schiebler, Lee and Brodbeck 2025). People who expect more are not systematically more disappointed. They are, on average, slightly more satisfied. Prospect theory is reference-dependent: outcomes are gains or losses relative to a reference point, and the value function is steeper for losses (Kahneman and Tversky 1979, Econometrica 47:263-291). The magnitude is contested. 1 camp puts the loss coefficient at lambda = 1.955 across 607 estimates (Brown et al. 2024). Another puts it at lambda = 1.31 and calls it weak loss aversion (Walasek et al. 2024). 2 peer-reviewed camps, 3 literature families, 1 shape, 2 numbers that do not agree. So the file bans every multiplier. Losses bite harder than equivalent gains, and that is usable. Losses hurt twice as much is not, because the 1st number is 1.955 and the 2nd is 1.31 and neither is 2. A magnitude that is contested in the literature does not become safe by being rounded inside a voice-over, and the rounding is exactly the step that turns a hedge into a false claim about other people's pain. 1 probability sits over the whole section. The probability that a chosen multiplier survives contact with the review is low. The probability that the same sentence survives without its number is high, which is the trade the file makes on purpose.

A verdict kills things, and the kills are the reason to keep it. Wanting and liking are dissociable: rats depleted of dopamine lost all motivation to eat and still showed normal liking reactions to sugar, while stimulating mesolimbic dopamine quadrupled wanting without enhancing liking (Berridge and Robinson 2016, American Psychologist 71:670-679). Liking rides small, fragile opioid and endocannabinoid hotspots, not dopamine. Dopamine neurons emit a reward prediction error: a burst for better than expected, a dip for worse or omitted, and then they fire to the earliest predictive cue rather than to the reward itself (Schultz, Dayan and Montague 1997, Science 275:1593-1599). Probability and uncertainty are coded phasically and peak at P = 0.5 (Fiorillo, Tobler and Schultz 2003, Science 299:1898-1902). The cue moves. The reward does not. 4 prohibitions ride on top of that chemistry. There is no oxytocin claim of any human kind, because the flagship trust finding failed a pre-registered replication at above 95 per cent power, the assays are self-acknowledged as flawed, the genotype studies are null, and almost none of an intranasal dose reaches cerebrospinal fluid. There is no organ-donation default figure, because the 2003 comparison of 42 per cent against 82 per cent was never read at primary source and the 2024 five-country longitudinal test found opt-out defaults raised nothing. Frankl is not the meaning evidence, because the integrative review of meaning cites him nowhere and logotherapy as a package stays interpretive. And there is no dopamine detox, because no primary human evidence supports a reset. The binding part is the part that costs. A banned line stays banned even when the direction of the claim is right. Direction is cheap; a claim can point the right way and still be unusable, and the file treats that as a reason to cut rather than as a reason to soften the wording until it passes. 1 probability applies. The probability that a banned line removed at the writing stage is cheaper than a banned line removed after the edit is close to certain. The probability that the same line, left in, becomes the 1 quote a hostile reader remembers is higher than most teams price.

The rule is 1 sentence: nothing goes on screen that is not already in the files.

The practitioner layer is where the verdicts do their loudest work, because it is the layer that arrives pre-loaded with confidence. The working rule is that the film is name-free science: the verdicts keep unsupported folk claims off the screen, and where a mechanism survives it appears without attribution. The 6-need map partially survives. Needs are plural and largely additive, with each need's satisfaction uniquely predicting well-being across 4 cultures, and a separate study of 123 countries finds the same plural, additive pattern (Chen et al. 2015; Tay and Diener 2011). Exactly 6 needs, in a stated sequence, with unmet needs blocking the others, does not survive. Needs behave like vitamins, not like rungs on a staircase, and the staircase is the part that sells. Burning desire does not survive as stated, and the reason is measurable. Positive fantasy de-energised the people who used it: systolic blood pressure fell, t(82) = 2.33, p below 0.02, and accomplishment a week later was lower (Kappes and Oettingen 2011). What works instead is mental contrasting plus if-then plans, g = 0.34 over 24 effect sizes and N = 15,907, and it is best taught live (Wang et al. 2021). The losing arm of that research was the visualisation everyone was already doing. The fixed 40 per cent does not survive either, and neither does its self-punitive form. The nearest support is the psychobiological model, where the limit is perceived effort rather than muscle failure, and self-compassion rather than self-attack predicts re-engagement after failure (Breines and Chen 2012). Dosed, controllable challenge builds capacity. Punishment does not build anything that lasts past the session. Identity framing survives with a boundary: noun framing moved turnout by 10.9 percentage points, and it works where the person already believes they should act, while it would backfire on behaviour seen as shameful (Bryan et al. 2011). The standing conclusion is 1 line: no unsupported practitioner claim remains available for the screen. That is not a statement about the practitioners. It is a statement about the ledger, and the ledger is checkable row by row. 1 probability closes it. The probability that a screen line written to the evidence reads flatter in the first draft is high. The probability that it survives a 2nd and 3rd review without a rewrite is also high, and the 2nd number is the return.

Handed over as a production order, the file reduces to 2 verdict classes, 127 rows, 161 claims, 11 construct-complete seams, 1 rule, and every screen line written to the evidence rather than written to sound good. The 1 thing the file does not contain is the 1 thing a motivational film needs. There is no study that manufactures motivation in a person with genuinely zero want and then measures sustained behaviour. So the turn in act 2 has to be a want-forming beat: raise the expectancy of success, or open a gap that pulls. The execution levers can only be spent after that beat has done its work. The plan multiplies a want. It has never once created one. The same order shows up in the work the file sits beside. Companies built across 12 countries. A EUR 75 million industrial group restructured. 210 energy systems deployed across Africa and Asia. 200 wood-gasification machines built across the UK and Europe. 16 years of field work from rural Nigeria to post-conflict Serbia, Kazan and Fukushima. A book of 393 pages published on 8 April 2026. 25,000 applications arrive for 25 places in a cohort, and 210 alumni across 19 countries now run the same order in their own work. In every 1 of those the sequence was the same as it is here: write the claim, check it against the source, mark the verdict, and cut what the verdict cannot carry. Nothing travels on screen that cannot travel with its evidence. 1 last probability. The probability that a claim shipped with its verdict intact is harder to attack than a claim shipped alone is high. The probability that the verdict is what makes the claim reusable in 2 years is higher still. So the next step is small and it is yours. Take 1 claim you repeat often enough that it has stopped feeling like a claim, and write down the verdict beside it. Is it evidenced, or is it a shape you have been borrowing?