Pinker Clarity Workflow

SkillDev tools

Orchestrate a Steven Pinker-grounded workflow for teaching, explanatory writing, or material that must do both. Use when the user explicitly asks for a Pinker approach, invokes this skill, asks how Pinker's work bears on teaching or writing, or needs one coherent workflow that first secures understanding and then improves prose. Do not use merely to imitate Pinker's personal voice or to import his political and historical views.

Available today. Use it from your connected AI after setup.

Connect ahel once, and every AI you use reads what you have installed.

Then ask your AI: use the Pinker Clarity Workflow skill

What this skill tells your AI

The instructions your AI receives, as published by ahepi/deepreason in .claude/skills/pinker-clarity-workflow/SKILL.md and read by ahel’s review.

Produce reader or learner understanding before polishing expression. Apply Pinker's communication principles without impersonating him or treating his entire worldview as a style guide.

Route the request

Identify the operating mode from the requested outcome:

  • Method: Explain how to teach, how to write, or how the two fit together. State which guidance Pinker directly gives and which guidance this workflow derives from his work.
  • Teaching: Use $pinker-teach-for-understanding for an explanation, lesson, course sequence, tutoring exchange, or teaching audit.
  • Writing: Use $pinker-write-for-readers for a draft, revision, writing lesson, or prose audit.
  • Teaching artifact: Apply teaching first to decide what the learner must understand, then writing to shape the handout, script, lecture, chapter, or worked example.
  • Research about Pinker: Read references/research-map.md. Verify time-sensitive claims against current primary sources.

If a companion skill cannot be loaded, follow the compact mixed workflow below rather than stopping.

Establish the communication contract

Determine the real-world subject, intended audience, desired change in understanding or ability, medium, length, register, and non-negotiable content. Infer ordinary details when safe. Ask only when a missing choice would materially change the result.

Treat the audience as intelligent but not already initiated into the writer's or teacher's guild. Record what they likely know, what they need next, and which terms or premises may be invisible to an expert.

Run the mixed workflow

  1. Ground truth. Extract the claims, evidence, distinctions, uncertainties, and source constraints. Research missing or current facts before drafting. Fluency is not evidence.
  2. Name the destination. State what the audience should be able to explain, predict, distinguish, decide, or do afterward.
  3. Build the model. Start from a phenomenon, problem, puzzle, or consequential question. Organize the few ideas that explain it. Separate the best current conclusion from live uncertainty.
  4. Bridge the knowledge gap. Pair each important generalization with a concrete example. Define necessary terms at first useful contact. Supply hidden premises and identify pronoun or acronym referents.
  5. Test understanding. Require paraphrase, prediction, comparison, application, or a counterexample when interaction permits. Do not use assent or smooth repetition as the test.
  6. Shape the prose. Put the problem before its solution, keep the topic visible, make relations between sentences recoverable, and prefer actors and actions when abstractions hide them.
  7. Test the communication. Seek feedback from a representative reader or learner when available. If no human test occurred, say so and provide targeted test questions instead of claiming validation. For a standalone learner-facing artifact, put that disclosure in the handoff note rather than interrupting the artifact itself.
  8. Revise. Repair the content model before local wording. Then revise coherence, syntax, context, concision, rhythm, and usage in that order.

Preserve these invariants

  • Preserve truth, warranted qualification, formal validity, authorial intent, accessibility, and genre obligations even when they conflict with elegance.
  • Remove defensive or status-signaling clutter, not uncertainty that changes the claim.
  • Use concrete scenes to ground abstraction; do not purge abstraction that has earned a stable meaning.
  • Use classic style as a useful default for exposition, not as a universal law for contracts, procedures, formal proofs, safety warnings, poetry, or fiction.
  • Keep Pinker's directly stated advice separate from inferences based on his research, practices observed in his courses, and independent learning science.
  • Do not derive a general pedagogy from Pinker's contested claims about linguistic nativism, evolutionary psychology, violence, progress, or politics.
  • Do not imitate Pinker's distinctive voice. Transfer principles such as reader modeling, concreteness, coherence, feedback, and revision.
  • For AI-produced prose, verify claims and sources and run an anti-banality pass: replace canned framing with precise observations, examples, and relationships supported by the subject matter.

Explain the method honestly

When the user asks how the workflow works, lead with its core: teaching and writing are both problems of reconstructing enough shared context for another mind to see a model of reality. Then explain the difference: teaching must produce and test a change in understanding; writing must encode the needed context and relations without live conversational repair.

Attribute classic style to Francis-Noël Thomas and Mark Turner as adopted and developed by Pinker. Attribute the term “curse of knowledge” to Colin Camerer, George Loewenstein, and Martin Weber; describe Pinker as applying and popularizing it for writing and teaching.

Research reference

Read references/research-map.md when explaining provenance, evaluating a proposed Pinker principle, updating the workflow, or answering broad questions about his body of work. It maps all thirteen books recognized in his current official biography, the directly relevant teaching and writing sources, the 2025–2026 common-knowledge work, and the boundaries that prevent attribution drift.

Signals

GitHub stars
142
Forks
14
Last commit
Sep 2026
Advanced
Catalog kind
skill
Gateway key
pinker-clarity-workflow
Source
github.com/ahepi/deepreason