Who Should You Believe

Teach me how to decide which experts to trust, given that I cannot personally verify almost anything I believe. Treat this as a solvable practical problem, not an excuse for either deference or cynicism.

Start by making me feel the problem: ask me to name three things I am confident about that I have not personally verified and could not verify in a week. Establish that trust is not a weakness in my epistemology, it is the load-bearing structure of it, so the question is never whether to trust but how to allocate it.

SECTION 1: THE CRITERIA. Teach me what a non-expert can legitimately assess, since I cannot assess the object-level claims:
- The arguments experts make against each other, watching not for who sounds right but for who engages the other's strongest point and who avoids it.
- Agreement among independent experts, with the crucial qualifier: independent. Twenty people citing one source is one source, and consensus manufactured by a shared funder or a single training lineage is not evidence of much.
- Credentials and track record, which are weak individually and useful in aggregate.
- Interests and incentives, including the ones on the side I find congenial.
- Calibration: has this person made checkable predictions, and did they come true? This is the strongest single signal and almost nobody looks for it.

SECTION 2: WHERE EXPERTISE IS REAL AND WHERE IT IS NOT. This is the part that gets left out, and it is the most useful thing here. Expert intuition develops reliably only under specific conditions: an environment with stable regularities, and fast unambiguous feedback on whether you were right. Explain why this makes weather forecasters, anesthesiologists, and chess players genuinely expert while making long-range political and economic forecasting far less reliable than credentials suggest, and cover Philip Tetlock's twenty-year study finding that experts on political futures were beaten by simple statistical rules while remaining highly confident. Then have me sort a list of domains by whether they meet the two conditions: medicine, nutrition, climate science, macroeconomics, geopolitics, engineering, psychiatry, and my own field.

SECTION 3: THE ASYMMETRY. Cover why the same field can be highly reliable on some questions and unreliable on others: physics on mechanics versus physics on the interpretation of quantum theory, medicine on acute treatment versus medicine on long-term diet. The right question is never "do I trust nutritionists" but "how well established is this particular claim, and by what kind of study."

SECTION 4: THE FAILURE MODES ON BOTH SIDES. Where deference genuinely went wrong: cases where an industry funded the science, where a field's consensus was overturned, where regulators absorbed the perspective of those they regulated. And where distrust genuinely went wrong: cases where the expert consensus was right and the contrarians were confidently wrong, and people died of the difference. Give real examples of both, so I do not leave with the lesson that suspicion is always the sophisticated position.

SECTION 5: APPLY IT. Ask me for one contested claim I am currently unsure about. Walk it through every criterion above and give me a defensible position with a confidence level, plus what would move it. Show the reasoning rather than the conclusion.

Rules: no false balance between well-established and fringe positions, and no treating institutional authority as self-justifying. Where a question is genuinely open among qualified people, name it as open.