How to Assemble Science

TEXT START: When reading the news, you may occasionally be bemused by the flip-flopping science and health stories: one week you’re told coffee is good for you, the next week that it’s bad.

The Dissection

The passage identifies a real failure: science produces studies faster than institutions can cumulate them. It then turns evidence synthesis—meta-analysis, systematic reviews, rapid reviews, living reviews, and mixed methods—into the missing master function. The examples of psychotherapy, SIDS, medicine, education, and climate policy create moral urgency.

The decisive rhetorical leap is the claim that humans already possess the knowledge needed to solve major problems and only need to assemble it and act. The essay recasts political and institutional breakdown as an information-assembly problem. It is also an argument for the author’s book and for the prestige of the evidence-synthesis profession.

The Core Fallacy

It confuses a better map with control of the terrain. A synthesis can estimate what works; it cannot compel governments, doctors, schools, firms, or publics to use it. It cannot dissolve conflicts over money, liability, timing, legitimacy, ownership, or power. The phrase ‘and then act’ treats action as if it automatically follows evidence. Under DT logic, knowledge is not authority, and authority is not ownership.

The second, fatal error is treating evidence assembly as a durable human economic role. Searching, screening, extracting, comparing, ranking, updating, and drafting are precisely the cognitive workflows exposed to P1. AI will make evidence banks faster and cheaper while reducing the need for human assemblers. Value shifts to whoever owns the models, compute, data, review platforms, and decision channels. Humans may remain as validators, liability holders, legitimacy interfaces, or local-knowledge brokers. That is a narrow Servitor niche, not restored mass productive participation.

Hidden Assumptions

• If better evidence exists, decision-makers will follow it.
• The main barrier to solving wicked problems is fragmented knowledge rather than incompatible interests, institutional incentives, resource allocation, enforcement, and ownership.
• Synthesis can integrate qualitative, Indigenous, and quantitative knowledge without political capture or loss of meaning.
• More rigorous reviews create implementation capacity.
• Human judgment remains indispensable even when the workflow is continuously automated.
• Public knowledge will remain a public capability rather than being enclosed by proprietary AI systems.
• Better policy selection can preserve the wage-to-consumption circuit that AI is severing.

Social Function

Primary classification: partial truth. Secondary classifications: transition management, prestige signaling, and ideological anesthetic.

The partial truth is substantial: single-study journalism and noncumulative research can produce lethal errors, while systematic synthesis can reduce them. The anesthetic lies in the implied sufficiency. By saying the needed knowledge already exists and only needs assembling, the essay relocates the crisis from power and ownership to professional competence. Institutions receive a respectable task—build better reviews—while the displacement of the reviewers and control of the resulting systems remain unexamined.

The Verdict

Useful method, false salvation. Evidence synthesis is a genuine lag defense: it can make medicine, education, and climate policy less stupid during the transition. It cannot reverse P1, overcome P2, or restore P3.

Once AI can assemble and update the evidence corpus at machine speed, the human synthesis profession becomes either a thin Servitor layer or a disposable labor pool. The Sovereign position belongs to whoever controls the synthesis infrastructure and the authority to implement its outputs.

The essay’s central claim is true at the level of knowledge and false at the level of power. It is a competent manual for reducing epistemic waste inside a dying order—not a mechanism for saving that order.