What is Vibe Science?

Vibe Science is the applied study of how weak human signals and machine inference become validated decisions under uncertainty. It formalizes the discipline between intuition and metrics using the VSV (Vibe → Signal → Validation) loop.

What is the VSV Loop?

The VSV Loop is a three-phase methodology: Vibe (human perception, intuition, discomfort, excitement), Signal (weak data amplified by machines), and Validation (structured experiments, often co-designed with AI). The novelty is that machines enter before certainty, not after.

How is a Vibe Scientist different from a Vibe Coder?

Vibe Coding is tactical prototyping. Vibe Science is systems-level inquiry. A Vibe Scientist designs and validates hypotheses at the boundary between human intuition and machine inference — creating probes, instruments, simulations, and falsifiable narratives.

Who is already doing vibe science work?

Many existing roles already practice aspects of vibe science: Chief Product Officers synthesizing weak signals, CTOs making architectural bets under uncertainty, Heads of Design holding ambiguity, Chiefs of Staff detecting organizational drift, UX Researchers clustering qualitative data, and early-stage founders operating on pattern recognition daily.

Field Notes

2026-07-10

Not every problem is agent-shaped

In mid-2026, 'add an agent' has become what 'add a chatbot' was in 2023: the reflexive answer to any product question that involves an LLM. The pitch decks are indistinguishable. The demos are impressive. The retention curves — where we can see them — are not.

An agent-shaped problem has three properties. First, the work is multi-step and the steps are not knowable up front. Second, the user's cost of specifying the steps exceeds the model's cost of choosing them. Third, the environment tolerates the failure modes — cost overruns, wrong tools called, silent partial completions — that autonomy introduces.

A form-shaped problem has none of those. It has a known sequence, a small input surface, and a low tolerance for surprise. Wrapping it in an agent adds latency, cost, and a new class of bugs, in exchange for the aesthetic of intelligence.

The honest question, before shipping any agent, is: if the model were a highly literal intern doing this exact task for the first time, would I trust it to run unsupervised? If not, the product is not an agent. It is a workflow with an expensive front door.

agents2026