06-reference

commoncog sensemaking expertise

2026-06-12·reference·source: Commoncog·by Cedric Chin

Sensemaking as the Heart of Expertise — Commoncog

Source: Sensemaking as the Heart of Business Expertise — Cedric Chin, Commoncog, 2026-06-12

Why this is in the vault

This is the capstone essay of Cedric Chin's multi-week sensemaking series. The prior entries introduced the Data-Frame Theory of Sensemaking framed as "how to make sense of AI"; this essay reveals the real purpose: sensemaking is the cognitive engine underlying expertise itself. If you can teach someone to construct frames the way experts do, you've cracked the accelerated-expertise problem.

The argument in brief: experts do not simply have more information than novices — they construct fundamentally different frames when reading a situation. Frame construction is the tacit bit that accelerated expertise training programs target. Critically, frame construction is also directly linked to insight generation. Winning business strategies, Chin notes via Rumelt and Roger Martin, reduce to a small set of framing insights ("the kernel"; "where to play / how to win"). Frame the problem right and half the strategic work is done.

The newsletter also includes an interactive experiment for paid members: a practice tool designed to let readers experience expert-style sensemaking firsthand, rather than read about it abstractly. A secondary member discussion thread surfaces the DROU concept (Directionally Right, Operationally Useless) — a test for AI outputs that sound strategic but cannot be executed, with a rewrite prompt attached. That specific framing is highly actionable for RDCO.

Mapping against Ray Data Co

Strong mapping across three vectors:

  1. AI COO judgment architecture. RDCO's core bet is that an AI agent can develop genuine operating judgment, not just execute instructions. Chin's argument reframes what "judgment" means at the mechanical level: it is frame construction under uncertainty. The RDCO harness currently improves agent behavior via SOPs, skills, and memory — but those are articulated rules. The Data-Frame lens suggests the next capability edge is training the agent to construct better initial frames when encountering ambiguous situations, before any SOP fires. This is not something a CLAUDE.md rule can encode; it requires structured exposure to expert-vs-novice frame contrasts (exactly what Chin's member experiment delivers).

  2. DROU as an RDCO quality gate. "Directionally Right, Operationally Useless" names a failure mode RDCO already encounters: agent outputs (and occasionally Ray's own planning artifacts) that pass a directional smell-test but lack executable specificity. The member-surfaced prompt that flags vague strategy-speak and forces rewriting is directly applicable as a /verify-strategic-output enhancement or a standing instruction in skills that produce recommendations. Worth tracking down the full prompt text from the member forum.

  3. Expertise acceleration for Ray's phData role. Ray is 17 days into a new role (phData DSA, 2026-05-26 start) with a 3-month cert cliff. The fastest path to expertise in a new domain maps directly onto what Chin is describing: not reading more, but acquiring better initial frames. The experimental training approach — structured exposure to expert interpretations of situations — is the same mechanism as deliberate practice case studies. Commoncog's paid case library is under-leveraged for this.

Minor note: Chin used Fable to prototype the experiment and "regrets it" — found the cost unjustifiable vs. alternatives. Useful ground-level signal on Fable's cost-performance tradeoff, especially relevant given the June 23 Fable access cliff noted in [[2026-06-10-stratechery-fable-5-anthropic-alignment-ai-tiers]].

Related