Philosophy / Credencing / Stances
Independent scholarship, public tools, philosophical stances, teaching, and photo art.
Work centered on epistemology, metaethics, graded belief, standards of evidence, induction, AI-mediated inquiry, moral anti-realism, semantic drift, and practical critical thinking.
Phil Stilwell is a semi-retired academic consultant, researcher, and university instructor whose public academic identity is anchored by part-time lecturing at New York University School of Professional Studies and profile affiliations with Soka University of Japan and the University of Tokyo.
His academic work focuses on rational confidence: how belief should track evidence, how inductive standards are applied, how public language handles partial belief, and how critical thinking can be taught without reducing disagreement to slogans.
The stance map below makes that orientation explicit: confidence is treated as graded, revisable, and answerable to evidence across philosophy, religion, moral language, AI, teaching, and public argument.
BA Philosophy, Summa Cum Laude, and MA Education, University of Kansas.
Epistemology, philosophy of science, induction, analytic philosophy, value theory.
English, critical thinking, academic writing, technical writing, and macroeconomics.
Futurology, AI-facing tools, public reasoning resources, and photo art.
The work is not a loose collection of topics. It is a single diagnostic program: replace binary belief with calibrated confidence, strip moral language of illicit metaphysical inflation, and ask where reasoning loses contact with the evidence it already has access to.
Three questions organize the program. What should confidence do when evidence is incomplete, noisy, or socially inconvenient? What remains of value-language when no objective moral realm is assumed? And how should people, institutions, and AI-mediated inquiry repair the gaps between evidence, perception, assertion, and commitment?
Belief is treated as a continuous evidential posture rather than a binary switch. The central question is not whether an agent can say "I know," but how assigned confidence should answer to evidence, uncertainty, defeaters, alternatives, and update conditions.
Moral language is treated as a dense human practice that compresses affect, value, vulnerability, social coordination, aversion, approval, and proposed regulation. The mistake is to inflate that grammar into stance-independent moral furniture.
Irrationality is decomposed into evidence-access failures, evidence-processing failures, belief-integrity failures, public-speech distortions, institutional incentives, and AI mediation risks. The point is repair, not name-calling.
The credencing project distinguishes the evidential situation from an agent's perception of it, and distinguishes both from the confidence finally assigned. This avoids treating disagreement, uncertainty, ignorance, rationalization, and bad faith as one blunt failure. It also forces public language to carry its epistemic load: confidence bands, sources, defeaters, live alternatives, and update conditions must be expressible without collapsing into fake precision or flat assertion.
The metaethical work rejects the idea that denying objective moral facts licenses cruelty or empties life of significance. Pain, trust, coercion, repair, dependency, flourishing, law, reputation, social order, and emotional gravity remain. What disappears is the extra metaphysical layer that redescribes human reactions as objective wrongness, guilt, permission, desert, or moral authority.
Rationality is treated as an alignment problem: can an agent, institution, or AI-assisted workflow let evidence change confidence? Deep Rationality improves evidence search, hypothesis comparison, base-rate awareness, and likelihood discipline. Core Rationality asks whether the agent then assigns confidence in line with what they take the evidence to support. The Credencing project makes that repair explicit by treating belief as an assigned confidence level that should expose its evidence, defeaters, alternatives, and update conditions. The difficult cases are not simple ignorance, but motivated non-updating, assertion pressure, institutional denial, felt certainty, and satisfying explanations that outrun support.
Paper trail
Free of Faith publishes an AI-mediated stance map of Phil's views. Its recurring pattern: belief should be proportional, updateable, and symmetric across religious, secular, political, moral, and technological claims.
Read beside public profile data and current credencing work, the same spine appears: evidence first, category discipline, explicit uncertainty, and suspicion toward any vocabulary that smuggles certainty or obligation past scrutiny.
Confidence should track evidence, preserve warranted uncertainty, and state what would move it.
Claims are tested by coherence, falsifiability, evidential relevance, fallacy control, and rival hypotheses.
God claims remain live only where evidence warrants them; faith is rejected when it protects belief from testing.
Moral predicates are treated as emotion- and preference-laden projections, not discoveries of mind-independent facts.
AI can widen search and simulate objections, but its outputs require traceability, calibration, and human update discipline.
English teaching, technical writing, macroeconomics, and futurology share the same demand: visible assumptions and clean reasoning.
Proportional credence, symmetry, defeasibility, base-rate awareness, semantic clarity.
Binary certainty, faith as evidence, moral-objectivist smuggling, unfalsifiable shelters.
PhilPeople / PhilArchive
Academia.edu
Academia.edu exposes profile uploads, but not a simple public popularity order in the accessible page.
Academia.edu profileAcademic sites
English language pedagogy
Fun sites