Vibe Science doesn't exist in a vacuum. These are the thinkers, frameworks, and tools that inform the discipline — or that operate in adjacent territory worth understanding.
The intellectual foundations this discipline builds on.
Bounded rationality and satisficing
The origin of the argument that real decisions are made by agents with limited information, limited time, and limited computation — and that 'good enough, now' is often the rational move rather than a compromise. Every claim on this site about acting before metrics exist rests on Simon's foundation.
Key work: Administrative Behavior; A Behavioral Model of Rational Choice (1955)
Base rates, the outside view, and separating skill from luck
The most practical body of work on asking 'how often does this kind of thing work in general' before telling yourself a story about why this case is different. His base-rate research is the corrective to inside-view forecasting, which is the default mode at the frontier.
Key work: The Success Equation; The Base Rate Book
Scout mindset and motivated reasoning
Her distinction between soldier mindset (defending a position) and scout mindset (mapping the terrain) is the clearest available language for what a null-case is actually for. Useful precisely because it treats accuracy as a habit rather than an intelligence trait.
Key work: The Scout Mindset
Fast strategic decision making in high-velocity environments
Her field research found that fast decision makers use more real-time information and more alternatives than slow ones — not less. The direct empirical rebuttal to the idea that speed under uncertainty requires cutting rigour.
Key work: Making Fast Strategic Decisions in High-Velocity Environments (1989)
Pace layering
The idea that a system's layers change at different speeds — fashion fast, culture slow — and that the fast layers get the attention while the slow ones do the constraining. The best available framing for why technical signals decay in weeks and signals about human motivation decay in years.
Key work: Pace Layering: How Complex Systems Learn and Keep Learning
Forecasting accuracy and the Good Judgment Project
His research showed that forecasting accuracy is trainable and that the training is mostly bookkeeping: record predictions, score them, update methodically. This is the empirical backbone of the argument that dated, written intuitions compound into judgement while undated ones decay into anecdote.
Key work: Superforecasting: The Art and Science of Prediction
Author of "How to Measure Anything"
His central claim — that anything you care about can be measured well enough to reduce uncertainty, and that measurement is about uncertainty reduction rather than precision — is the direct answer to "but this can't be quantified." Essential for anyone trying to put a number on a vibe without pretending to false precision.
Key work: How to Measure Anything: Finding the Value of Intangibles in Business
Organisational sensemaking
His work on how organisations construct meaning retrospectively explains exactly why signal provenance decays: summaries travel, sources do not. The observation/interpretation split in our SPINE record exists to slow that construction down.
Key work: Sensemaking in Organizations
Strategic inflection points
"Only the Paranoid Survive" is the original practitioner account of detecting a structural shift before the numbers confirm it — including the honest admission of how long Intel's own leadership argued with the signal. The best case study of the cost of being early and the cost of being late in the same book.
Key work: Only the Paranoid Survive
The measurable decay of established knowledge
His work on how established facts lose accuracy over predictable timescales is the direct precedent for treating signals as having half-lives. If published science decays measurably, a team's undated working beliefs decay faster.
Key work: The Half-Life of Facts
Decision quality and knowing when to quit
Her work separates decision quality from outcome quality — the distinction that makes a kill ledger legible. "Quit" is the most direct treatment we know of the argument that stopping on time is a skill, not a failure.
Key work: Thinking in Bets; Quit
Originator of weak signal theory (1975)
The origin of treating low-amplitude, ambiguous indicators as legitimate strategic input rather than noise. Every framework on this site that records a signal before its meaning is clear is downstream of Ansoff's argument.
Key work: Managing Strategic Surprise by Response to Weak Signals
Pioneer of Naturalistic Decision Making
His research on how experts make decisions under time pressure and uncertainty is foundational. His "Recognition-Primed Decision" model is essentially the academic version of "trust the vibe, then validate it."
Key work: Sources of Power: How People Make Decisions
When can intuition be trusted?
Their joint paper "Conditions for Intuitive Expertise: A Failure to Disagree" (American Psychologist, 2009) is the single best articulation of when intuition can be trusted and when it can't. Essential reading for anyone doing vibe science.
Key work: Conditions for Intuitive Expertise (2009)
Creator of the Cynefin framework
His work on complexity, probe-sense-respond, and the difference between complicated and complex systems directly informs how we think about validation under uncertainty.
Key work: A Leader's Framework for Decision Making (HBR, 2007)
Commoncog — practitioner-oriented expertise research
His writing on tacit knowledge extraction, expertise acceleration, and naturalistic decision-making is the best practitioner-oriented work connecting academic NDM research to real business contexts.
Columbia professor, strategic inflection points
Her book "Seeing Around Corners" is about detecting strategic inflection points through weak signals and leading indicators. Directly adjacent to vibe science's focus on early signal detection.
Key work: Seeing Around Corners
Co-founder of ReD Associates
His book "Sensemaking" argues for humanities-based approaches to business problems as a counterweight to algorithmic thinking. The tension he identifies is one vibe science tries to resolve.
Key work: Sensemaking: The Power of the Humanities in the Age of the Algorithm
Coined "vibe coding" (February 2025)
Important to reference explicitly: vibe coding is tactical prototyping with AI. Vibe science is the discipline underneath — the systems-level inquiry into why some vibes are signal and others are noise.
Creator of Wardley Mapping
His work on situational awareness and mapping strategic landscapes under uncertainty is directly adjacent. Mapping the unknown is what vibe scientists do before the map exists.
Author of "The Uncertainty Mindset"
Former Google strategy. His research on how high-end restaurant kitchens organize for uncertainty is surprisingly applicable to innovation teams. A masterclass in structuring for the unknown.
Key work: The Uncertainty Mindset
Military strategist, creator of the OODA loop
The OODA loop is the closest military predecessor to the VSV loop. Boyd understood that speed of sense-making beats speed of action. Orient — where mental models meet new data — is where vibe science concentrates.
Systems thinker, leverage points
Her work on leverage points and system dynamics informs how we think about where small signals indicate large shifts. Not all signals are equal — Meadows showed us where to look.
Key work: Thinking in Systems: A Primer
Uncertainty, fragility, and optionality
His work on uncertainty, fragility, and optionality under fat-tailed distributions is essential context. Vibe science operates in the domain Taleb warns about — and tries to build antifragile practices within it.
Key work: Antifragile
Harvard professor, psychological safety
Her work explains why most teams can't do vibe science: they lack the safety to say "I have a feeling something's wrong" without data to back it up. Safety is the prerequisite.
Key work: The Fearless Organization
Developer, writer, AI tools practitioner
His ongoing documentation of AI tool usage patterns is the most honest, practitioner-level record of how humans and machines actually collaborate. Essential reading for the "fusion" angle of vibe science.
Wharton professor, AI and the future of work
His research and writing on how AI changes knowledge work is the closest academic work to the human+AI fusion thesis. Rigorous, practical, and constantly updated.
Key work: Co-Intelligence: Living and Working with AI
Frameworks and ideas that inform the practice.
Herbert Simon's finding that real decisions are made under limits of information, time, and attention — so the practical target is a choice that is good enough given those limits, not an optimum computed from data you do not have. The formal licence to act before the metrics exist.
Asking how often this class of thing succeeds in general, before reasoning about why this instance is special. Most frontier forecasts are inside-view stories with no base rate attached, which is why they cluster around optimism.
Julia Galef's frame for treating accuracy as a habit: mapping the terrain rather than defending a position. Practically, it is what makes a null-case writable — you cannot generate the boring explanation while defending the interesting one.
Stewart Brand's observation that systems have layers moving at different speeds. Applied to signals: tooling claims decay in weeks, human motivation in years, and confusing the two is how a team ends up re-litigating settled things while inheriting expired ones.
Amy Edmondson's distinction between failures that produce information (small, hypothesis-driven, in novel territory) and failures that produce only cost. The kill ledger is the artefact that makes the distinction visible after the fact.
Annie Duke's separation of the quality of a decision from the quality of its result. Under uncertainty, good decisions routinely produce bad outcomes. Judging process by outcome — resulting — is the fastest way to unlearn a working method.
Douglas Hubbard's reframing: a measurement is anything that reduces uncertainty, not something that produces precision. This is what makes it possible to quantify a vibe honestly, without pretending to a decimal place you have not earned.
Andrew Grove's term for the moment the fundamentals of a business change. Detectable early only through signals that contradict the current numbers — which is exactly why they get argued with rather than acted on.
Samuel Arbesman's observation that established knowledge decays at measurable, domain-specific rates. Applied at team scale, it argues that every recorded signal needs an expiry date rather than an indefinite shelf life.
The academic field studying how experts actually decide in the wild. Not how they should decide. How they do. This is the research tradition vibe science most directly extends into the AI era.
Dave Snowden's sense-making framework. Vibe science operates primarily in Cynefin's "complex" domain, where the relationship between cause and effect can only be perceived in retrospect. Probe-sense-respond is our operating model.
Gary Klein's model. Experts don't weigh options — they recognize patterns and simulate actions mentally. Vibe science adds: machines can now help with the pattern recognition part.
Igor Ansoff's original concept (1975) of detecting strategic discontinuities through early, ambiguous indicators. The entire premise of the VSV loop's "Signal" phase.
Karl Popper's principle that scientific theories must be disprovable. Our commitment that hypotheses must be falsifiable comes directly from this. Not as philosophy — as practice.
Gary Klein's technique. Imagine the project has failed, then work backward to identify why. One of the most practical vibe science tools.
John Boyd's Observe-Orient-Decide-Act loop. The military predecessor to sense-and-respond frameworks. The "Orient" phase — where mental models meet new data — is where vibe science concentrates.
Michael Polanyi's concept: "We know more than we can tell." The entire premise of the Vibe phase. Expertise manifests as pattern recognition that resists articulation. Vibe science creates structures to surface it.
Donella Meadows' hierarchy of places to intervene in a system. Vibe scientists look for the highest-leverage signals, not just the loudest ones. Where you intervene matters more than how hard you push.
Taleb's concept of systems that gain from disorder. Good experiment design in vibe science is antifragile: killing an idea makes the portfolio stronger, not weaker.
Amy Edmondson's research. Without it, weak signals stay private. The organizational prerequisite for vibe science to function. If people can't say "this feels off," you've lost before you started.
The primary sources. No summaries substitute for reading these.
Amos Tversky & Daniel Kahneman · 1974
The Science paper that named representativeness, availability, and anchoring. Still the most efficient way to learn the specific ways a confident read of an ambiguous situation goes wrong — which is the exact activity this site is about.
Glenn W. Brier · 1950
The scoring rule underneath every modern calibration practice. Worth reading once for the realisation that scoring probabilistic claims is a solved problem, and the only hard part is writing the claims down.
Kathleen M. Eisenhardt · 1989
Field study of decision making in fast-moving firms. The counter-intuitive result — faster teams used more information and considered more alternatives — is the empirical basis for arguing that speed and rigour are not a trade-off.
Stewart Brand · 2018
The essay version of the layered-change idea. Read alongside Arbesman if you are trying to work out what expiry date to put on a given belief.
H. Igor Ansoff · 1975
The origin of weak signal theory. Ansoff's argument that organisations must act on ambiguous early indicators rather than waiting for confirmation is the intellectual ancestor of the entire Signal phase.
Gary Klein · 2007
The two-page HBR piece that introduced the pre-mortem to a general management audience. Short enough to read before your next planning meeting, and specific enough to run it the same day.
Jeff Bezos · 2016
The primary source for the reversibility distinction that most product organisations now quote second-hand. Worth reading in the original for the argument about what large organisations get wrong by default.
Rahul Vohra · 2018
The canonical worked example of converting an intuition into a falsifiable, repeatable measurement — and then deliberately moving the number. Annotated as a case study in our experiments log.
Daniel Kahneman & Gary Klein · 2009
The foundational text on when to trust expert intuition. Two intellectual rivals find surprising common ground on the conditions under which gut feelings are reliable.
David Snowden & Mary Boone · 2007
The Cynefin paper that introduced probe-sense-respond to a business audience via Harvard Business Review. The paper that made complexity theory actionable for practitioners.
Donella Meadows · 1999
Where to look for high-impact signals in complex systems. Meadows' hierarchy of intervention points remains the best guide to understanding which signals matter most.
Vaughn Tan · 2020
On organizing for the unknown, drawn from haute cuisine kitchens. How the world's best restaurants structure teams to thrive under uncertainty — and what innovation teams can learn.
Kenneth Stanley & Joel Lehman · 2015
On the failure of objective-driven search in complex spaces, and why open-ended exploration finds more. The academic case for why rigid KPIs kill innovation.
Ethan Mollick · 2026
The argument that the chat-box era of AI is ending, and that work is moving toward agents that act on your behalf. Foundational context for how AI tooling will feel different from here on.
Ethan Mollick · 2026
Follow-up to Co-Intelligence: as AI systems begin acting more autonomously, the collaboration frame shifts from working with AI to negotiating alongside it. Direct implications for how vibe scientists design the human-machine loop.
Anthropic · 2026
Empirical look at how enterprises are moving from AI-as-copilot to AI-as-agent in production. Useful for calibrating what's actually being deployed versus what's being announced.
The models we publish on the Frameworks page are assembled from existing work, not invented. Here is what each one owes, and to whom.
No affiliate links. No partnerships. Just what works.
For experiment documentation and signal logging. The tool matters less than the habit. Pick one, use it consistently.
For qualitative research synthesis. Useful when you're clustering weak signals from interviews, support conversations, or user sessions.
For visual sense-making. Especially useful in the "Vibe" phase when you're trying to externalize a pattern you can feel but can't yet articulate.
For hypothesis stress-testing, signal clustering, and generating validation artifacts. The machine contribution in the Human + AI fusion. Best used as a thinking partner, not an answer machine.
Dave Snowden's tool for distributed ethnography and narrative-based research. Heavyweight, but the closest purpose-built tool to what vibe science aims to do.
For strategic situational awareness. Useful when you need to understand where in the value chain your signals are coming from and what stage of evolution a capability is in.
For networked thought and signal linking. Better than linear note-taking when you need to see connections between signals across time.
For rapid evidence gathering and literature review. When you need to validate whether a signal has academic or empirical backing before designing an experiment.
One selection rule: the value has to survive a model release.
Nothing here is a partnership, an endorsement or an affiliate link. Nobody listed has been contacted, asked, or paid. These are programmes that teach judgement under uncertainty — the part of the work that does not get cheaper when the tools do.
What we skip: credentials tied to a single vendor's stack, and certifications that certify attendance rather than judgement. Both expire faster than the models they are built around.
The Cynefin Co (Dave Snowden) · Paid, multi-day, run in cohorts
How to tell which kind of problem you are actually in — clear, complicated, complex, chaotic — and why the wrong method applied confidently is worse than no method. Includes narrative and distributed sense-making practice.
Feeds: The Vibe phase: deciding whether a situation is even analysable yet.
Where it stops: Heavily conceptual. It will sharpen how you classify a problem; it will not teach you how to run the experiment once you have classified it.
Good Judgment Inc · Paid corporate training; free practice via Good Judgment Open
Probabilistic reasoning, base-rate discipline, and scoring your own forecasts so overconfidence becomes visible rather than debatable. Descends from the IARPA forecasting tournaments.
Feeds: The Calibration Card. This is the closest thing to a formal qualification in the skill it measures.
Where it stops: Optimised for resolvable geopolitical-style questions. Early product signals often resist clean resolution, so the scoring discipline transfers better than the question format.
Hubbard Decision Research (Douglas Hubbard) · Paid seminars and in-house engagements
How to put a number on things people insist are unmeasurable, how to compute the value of additional information, and when to stop measuring because the decision will not change.
Feeds: SPINE and the Kill Ledger — deciding what evidence is worth gathering before you gather it.
Where it stops: Quantitatively demanding, and assumes a decision with a real cost attached. Less useful when you are still trying to work out what the question is.
Richard McElreath, Max Planck Institute · Free lectures and materials; book sold separately
Bayesian inference built up from causal reasoning rather than test-recipes. Full lecture series, problem sets and code are published free; the book is the paid part.
Feeds: The Signal phase: reading small, noisy, non-random data without pretending it is a clean A/B test.
Where it stops: It is a real course, not a primer. Expect months, not a weekend. Most teams need only the first third — enough to stop over-reading n=7.
Nielsen Norman Group · Paid courses, certification by course credits
Interview technique, study design, and qualitative synthesis at a level of rigour most product teams never reach. The unglamorous craft underneath every credible weak signal.
Feeds: Signal capture: getting evidence that is worth logging in the first place.
Where it stops: Method-heavy and consumer-research shaped. It teaches you to run a study well; it does not teach you when a study is the wrong instrument.
Rosenfeld Media · Paid workshops and conferences
Practitioner-run workshops on research operations, synthesis and design research leadership. Closer to how the work is actually done in organisations than academic training is.
Feeds: Turning scattered qualitative signal into something a team can act on and revisit later.
Where it stops: Quality varies by instructor — these are individual workshops, not a coherent curriculum. Choose by facilitator, not by brand.
CITI Program · Institutional subscription; often free through a university or employer
Consent, minimal risk, data handling and the basic ethics of testing anything on real people. The standard institutional module in academic and clinical research.
Feeds: Any probe that touches real users, health, finance or employment data — and the privacy protocols on the Tools page.
Where it stops: Written for institutional review contexts, not product teams. It sets a floor for what is defensible, not a process for shipping.
Publications, communities and associations working near this problem.
Credit, not affiliation. Each entry says what it is genuinely good for and where it diverges from what we are doing — because a recommendation without a boundary is just enthusiasm.
Newsletter and practitioner community — Lenny Rachitsky
The most consistent public archive of how product teams actually decide things, including the Sean Ellis PMF survey work we cite in our own experiment write-ups.
Where it diverges: Optimised for teams that already have users and metrics. Vibe science is about the period before that, where most of the playbooks there do not yet apply.
Publication — Cedric Chin
A sustained, serious attempt to make business expertise teachable, drawing on naturalistic decision-making rather than anecdote. The nearest neighbour to what this site is trying to do.
Where it diverges: Focused on how expertise is acquired over years. We are focused on what to do in the first two weeks, before expertise exists.
Newsletter — Ethan Mollick, Wharton
Published, dated observations about what AI systems can and cannot do in real work, with the experiments described well enough to argue with.
Where it diverges: Concerned with human–AI performance generally. We only care about the subset used to detect signal before metrics exist.
Curriculum and practitioner network
The closest thing to a structured professional curriculum for product and growth work, built by operators rather than academics.
Where it diverges: Programme-shaped and execution-oriented. Strong on running a known play; quieter on what to do when there is no play yet.
Academic association
The actual scholarly association for the field vibe science borrows from. Its annual conference and open abstracts are where the underlying research surfaces years before it reaches practitioner writing.
Where it diverges: Laboratory-first. Findings need translation before they survive contact with a startup on a two-week clock.
Academic association
SJDM's European counterpart, with the SPUDM conference series. Useful for tracking naturalistic and applied decision research outside the US.
Where it diverges: Same translation gap as SJDM: rigorous, slow, and not written for operators.
Public forecasting platform
A live, scored record of predictions on open questions. The best free way to practise calibration on things you cannot control, and to see what honest uncertainty looks like when it is written down.
Where it diverges: Only handles resolvable questions with a fixed close date. Most early signals are neither.
Research community and conference series
Where human-computer interaction research on how people actually work with AI systems is published, often with open-access papers and full method sections.
Where it diverges: Academic peer-review cycles run at roughly a year. Fine for foundations, too slow for anything about last quarter's models.
Open education organisation
Low-cost, structured courses and a large open literature library on design research and human-centred method. A reasonable entry point when formal training is out of budget.
Where it diverges: Breadth over depth, and design-discipline shaped. Treat it as orientation, not certification.