Grant Sanderson proposes “motivated explanations” as a new metric for mathematical contribution

Takeaway

Grant Sanderson (3Blue1Brown) proposes that the mathematics community formally define, celebrate, and reward “motivated explanations”—explanations that answer how and why a result is natural—so that human understanding, not just proof generation, remains the core value of the field.

What a “motivated explanation” is

Definition vs. discovery – In a traditional proof, definitions appear at the start and the argument proceeds rigorously from them. In a motivated explanation, the problem is presented first; new definitions are introduced only when they become necessary to address the problem.
Iterative reasoning – A motivated explanation may begin with an idea that is partially wrong, then correct it, showing the thought process that leads to the final insight.
Scope – While a proof explains why a theorem is true, a motivated explanation also explains why the theorem is the right question to ask and how it fits into a broader mathematical context.
Verifiability – The property is not binary like a proof, but it is still assessable: one can ask whether each new idea is clearly motivated by the preceding discussion.

“I want this to feel like you could have discovered it yourself.” – Grant Sanderson, describing his own guideline for motivated explanations.

Historical precedents

Princeton Companion to Mathematics (Part IV) – Essays by Andrew Granville, David Ben‑Zvi, and others provide the kind of expert‑level intuition that Sanderson calls a motivated explanation.
Bill Thurston’s essay “On Proof and Progress in Mathematics” – Thurston argued that the value of a proof lies in the human understanding it generates, not merely its logical correctness.
Timothy Chow’s “A beginner’s guide to forcing” – Introduces the notion of an “open exposition problem,” i.e., a problem whose resolution is a clear, fully motivated exposition.

Why the shift matters now

AI‑generated proofs – Large language models can produce technically correct proofs without conveying the underlying ideas, echoing concerns raised by Thurston about the 4‑color theorem proof.
Public perception – If the community continues to celebrate only proof‑generation, outsiders may conclude that mathematicians are replaceable by machines.
Career incentives – Current hiring, tenure, and prize structures heavily weight proof‑oriented results; motivated explanations receive little formal credit.

Community reactions (selected HN comments)

Historical bias – @ForgotMyUUID notes the 1900 Hilbert‑Poincaré debate that cemented proof over intuition, suggesting modern curricula still undervalue intuition.
Economic analogy – @sweezyjeezy compares the situation to software engineering, where AI automates routine tasks, forcing a redefinition of human contribution.
Subjectivity concern – @c7b points out that measuring “motivated” is inherently subjective, and questions whether this shift actually protects mathematicians from AI.
Goodhart’s Law – @kurthr warns that once “motivated explanations” become a target metric, they may lose their usefulness, echoing a classic pitfall of performance measures.
Teaching vs. research – @pcfwik wonders how the proposal would affect PhD career paths, potentially blurring the line between research‑intensive and teaching‑focused positions.
AI assistance – @contubernio shares a personal account of using AI to discover a proof, highlighting that the workflow is already changing and that clear exposition remains essential.

Concrete actions to elevate motivated explanations

Redefine PhD deliverables – Require students to present a solved (or AI‑assisted) problem as a talk that makes the solution understandable to peers, not just as a written proof.
Curate “open exposition problems” – Senior mathematicians could publish a list of important results that lack clear, motivated explanations, analogous to Hilbert’s or the Millennium Problems.
Develop a verification rubric – Journals and societies should draft criteria (e.g., logical flow, explicit motivation for each new concept) that make motivated explanations as assessable as proofs.
Create dedicated venues – Journals such as Mathematical Discourse can focus on publishing high‑quality expositions; new venues could be launched for this purpose.
Reward exposition in hiring – Extend awards like the AMS Steele Prize for Exposition to early‑career researchers and factor exposition quality into tenure and promotion decisions.

Cultural impact

Student recruitment – Emphasizing human understanding counters the narrative that AI will render mathematicians obsolete, making the field more attractive to new talent.
Public good – By producing explanations that scale from experts to lay audiences, mathematics can reinforce its role as a public intellectual resource.
Funding justification – Demonstrating that mathematicians add unique explanatory value provides a stronger case for public and private investment in the discipline.

Bottom line

Grant Sanderson’s call to recognize “motivated explanations” reframes mathematical contribution from a binary proof‑centric model to a richer, human‑understanding‑centric one. Implementing clear definitions, reward structures, and dedicated publication venues can preserve the unique role of mathematicians in an era where AI can already generate correct proofs.