“Data is Just an Added Sense” — @CedricChin
Why this is in the vault
This article gives RDCO a clean, memorable positioning metaphor for what MAC and the agent-deployer role actually do: they add a sense the operator didn’t previously have. It reframes the tired “data-driven vs gut-driven” debate as a category error, which is exactly the frame needed when explaining MAC to skeptical operators.
The core argument (paraphrased)
Data is not better or worse than intuition or qualitative sensemaking — it’s simply a third sense that most operators fail to use.
Chin’s framing: humans have five senses; good operators typically use two — intuition (gut built from years of decisions) and qualitative sensemaking (“get out of the building”). Data is a third sense. “Data is just an added sense” — not superior, not inferior, just different tradeoffs. Arguing sight is better than hearing is the same category error as arguing data beats intuition.
Key moves:
- The misconception is symmetric. Both the “demand hard numbers for everything” crowd and the “data can never capture what could have been” crowd commit the same error — they rank the senses instead of using them all.
- Data has distinctive reach. It captures what interviews can’t (e.g. actual engagement vs stated love for the app) and speeds process-improvement feedback loops — “you can get to certainty much faster” — when effects are measurable.
- Stories beat data when they conflict. From Amazon’s WBR: “when stories contradict the data, always investigate the stories, not the data.” Data is always lossy compression of reality.
- Sophisticated operators reach laterally. The best he’s seen move fluidly across all three senses — “let’s go talk to some customers” is as likely a response as “let’s instrument this.”
- The role of data is to build and verify intuition. Not to replace it. Intuition updates from both qualitative and quantitative input. “Your use of data should always stem from a deep qualitative understanding of your customer” (Colin Bryar).
- “Data is an Added Sense” works as an operating principle. A CEO Chin was talking to adopted it on the spot, specifically as a hedge against data being used to browbeat (because it’s so legible it tends to overtake other forms of sensemaking).
The practical warning: sensemaking is a skill. Most people don’t reach for data because nobody taught them how to read a wiggling chart — but customer interviews are no easier. No excuse to skip the training.
Mapping against Ray Data Co
Medium-to-strong mapping. This is primarily a positioning asset — a rhetorical frame — rather than a structural one like the First Principles essay. Four mappings:
1. MAC as “adds a sense the operator didn’t have.” The cleanest positioning sentence RDCO has been searching for. When a prospect asks what MAC does, the answer isn’t “validates your data quality” — it’s “it gives the operator a sense they didn’t have before: whether the model’s outputs are trustworthy today, at the cell-level resolution they care about.” This frames MAC not as governance overhead but as sensory augmentation. Ship to the landing-page copy and the ../01-projects/data-quality-framework/testing-matrix-template explainer.
2. The agent-deployer role is “the operator who uses all three senses on AI outputs.” Per 2026-04-14-levie-agent-deployer-role-jd, the agent-deployer instruments AI workflows and manages evals. Chin’s frame says: the job is to make sure the business operator using the agent has data-sense about the agent — otherwise they’re flying with two senses on a system that produces output at a rate no human can qualitatively audit. MAC is the data-sense for agent outputs.
3. Defense against “AI will replace data quality work.” Chin’s symmetric-misconception point maps directly onto the harness-thesis debate. When 2026-04-13-moura-entangled-software-agent-harnesses-dead-style critics argue data quality work is unnecessary (“the model will figure it out”), Chin’s reply: that’s the same category error as “data beats gut.” You need all three senses; removing one because another is loud is self-inflicted impairment. Operators who celebrate being data-impaired are doing the business equivalent of hiking blindfolded.
4. phData/MG differentiation. phData/Monstera Group-style consulting tends to sell data platforms as the primary sense (“be data-driven!”). RDCO’s posture — per ../04-tooling/rdco-state-ownership-architecture — is that data is one sense of three and the operator owns the causal model. This is philosophically closer to Chin than to platform vendors, and it’s a wedge: we’re not selling data maximalism, we’re selling sensory completeness. The operator still trusts their gut and still talks to customers; MAC just makes sure the data sense isn’t lying to them.
One caution: “Data is an added sense” is softer than the harder claims in the First Principles piece (predictive validity, SPC rigor, etc.). Use this as the opening frame — the memetic catch phrase that gets the prospect nodding — and follow with the First Principles material once they’re bought in. Don’t let the softness become a reason to underinvest in MAC rigor.
Related
- 2026-04-15-commoncog-becoming-data-driven-first-principles — the harder, structural companion to this positioning piece
- ../04-tooling/rdco-state-ownership-architecture — “operator owns the causal model” maps to Chin’s three-senses frame
- ../01-projects/data-quality-framework/testing-matrix-template — MAC as the “added sense” for agent outputs
- 2026-04-14-levie-agent-deployer-role-jd — the agent-deployer is the three-sense operator applied to AI workflows
- 2026-04-13-moura-entangled-software-agent-harnesses-dead — harness-skeptic dissent that Chin’s symmetric-misconception frame refutes