Chapter 36 — Further Reading

A note this chapter needs more than any other. A future-facing chapter is where invented citations are easiest to get away with, because nobody checks a reference for a trial that has not reported. So this list names categories of source and how to use them wherever a specific citation would require me to assert something I cannot verify. Where I name a work, I stand behind it. Where I do not name one, the absence is deliberate, not an oversight. Everything is dated to the chapter's frame: as of this writing, in 2026.

Tier 1 — Verified canonical

Philip E. Tetlock and Dan Gardner, Superforecasting: The Art and Science of Prediction (2015). The chapter's epigraph and the source of its central distinction. If you read one thing from this list, read this. Its argument — that forecasting skill is real, measurable, and produced by keeping score rather than by expertise alone — is what §36.10's three dated lists are built to imitate.

Karl Popper on falsifiability. The Logic of Scientific Discovery and Conjectures and Refutations are the canonical statements of the idea that a claim earns its status by specifying what would refute it. Rating rule 5 is Popper applied to a press release; the argument is available in the opening chapters of either book.

Trial registries. ClinicalTrials.gov (United States), the WHO International Clinical Trials Registry Platform, and the EU Clinical Trials Information System are the primary way to answer question 1 — what phase, actually? — for yourself, in about two minutes, for free. Search a compound name and read the study records directly: phase, enrollment, primary outcome measure, comparator, inclusion and exclusion criteria, and estimated completion date. This single habit will do more for your reading of this field than anything else on this page. It also answers question 3, because the eligibility criteria are listed in full and are usually startling.

Regulatory review documents. Drugs@FDA, FDA advisory committee briefing materials, and the European Medicines Agency's European public assessment reports are the rung on §36.2's ladder just below full publication. They are written by reviewers whose job is adversarial and who saw the complete dataset, including the analyses that did not appear in the sponsor's summary. They are long, unglamorous, and considerably more informative than any news article about the same drug.

The primary trial publications for the compounds named in this chapter. Tirzepatide, semaglutide (including the cardiovascular outcome trial discussed in Chapter 10), oral semaglutide, retatrutide's earlier-phase program, cagrilintide, CagriSema, orforglipron, bimagrumab, and lutetium-177 dotatate all have peer-reviewed literature behind them at various stages of completeness. Find them by searching the compound name in PubMed and filtering to randomized controlled trials, or by following the publication links in the registry record. I am not listing journal citations here: getting one wrong in a chapter about not overclaiming would be the worst possible place to be sloppy.

This book's own prior chapters, which supply the machinery this chapter reuses: Chapter 5 (mechanism is not evidence), Chapter 6 (source hierarchy and selection effects), Chapter 9 (phase 2 optimism), Chapter 10 (the attrition base rate; outcome trials), Chapter 16 (surrogate versus hard outcome; myostatin), Chapter 27 (PRRT), Chapter 28 (active comparators), Chapter 32 (solid-phase synthesis), Chapter 33 (engineering strategies and their limits), and Chapter 35 (captopril, ziconotide, computational design). Appendix C holds the blank dossier workbook.

Tier 2 — Attributed, specifics unverified

These are worth reading. They are also, without exception, written by someone with an interest in how you read them, which is question 5 in practice.

Company press releases and topline announcements. Read them for the fact that a trial reported and for the endpoint named. Do not read them for magnitude, safety, discontinuation, or subgroup behavior, none of which the format is obliged to contain. Note the gap to full publication, often a year or more.

Conference abstracts and presentations from the major diabetes, obesity, endocrine, and oncology meetings. More detail than a press release, not peer-reviewed, and a substantial fraction never appear as full publications at all.

Quarterly earnings calls and investor presentations. The audience is investors and the purpose is to shape revenue expectations. That does not make the content false — securities law applies and companies are generally careful — it makes it selected. Read for pipeline stage and timing guidance; discount the framing entirely. The CagriSema episode in §36.5 is the worked example of what happens when this material reaches a general audience unlabeled.

National patent registers and commercial patent databases for the §36.9 material. Expiry dates differ by jurisdiction, shift with litigation and administrative decisions, and are exactly the kind of false precision this chapter refuses to supply. For a specific country, consult a current legal or regulatory source, not a textbook — including this one.

Science and health journalism, read with the awareness that it sits one compression layer and one headline layer below whatever it summarizes.

Tier 3 — Illustrative / constructed

The press releases in exercises n, p, and q are written for this book. They describe no real compound, company, or trial, and are labeled [constructed teaching example] every time they appear.

The two worked dossier entries in the chapter's dossier section — oral peptide delivery and peptide-drug conjugates — are demonstrations of the four-line format, not recommendations about what to track.

§36.10's three lists are the author's judgment with no evidence behind it, and the chapter says so in its first sentence. Their only value is structural: they are specific, dated to 2026, and gradeable. Treat them as a template, not as information about the future.

If you only do one thing: take the most exciting "coming soon" claim you currently believe, look the compound up in a trial registry, and write down what phase it is actually in, what the primary endpoint actually is, and what the comparator actually is — then compare that to what you thought you knew, and notice which of the six questions your source had quietly answered for you.