Canon 3X: Explore/Expand/Extract
Kent Beck · email · 2026-07-30 source ↗ Reader 🖥️
Because a growing idea passes through three phases — Explore, Expand, Extract — that each demand categorically different, even incompatible approaches to finance, team size, strategy, personnel, and risk, applying one phase's playbook to a project actually in a different phase kills the idea; managing a portfolio well means correctly matching each project's approach to its actual phase, and prioritizing Expand-phase work even above currently more profitable Extract activity.
practical · mixed — Primarily expository/prescriptive — a framework document laid out as Risk/Goal/Strategy/Tactics/Exit for each phase, explicitly framed by the author as 'no theory, just stuff.' But it rests on one compressed theoretical claim (the S-curve as a tug-of-war between two feedback loops) and its central sentence is a normative warning ('applying the approach from one phase to an idea in another phase kills ideas'), which makes it argumentative at its core even though the surface form is a reference table.
Structure
| Part | Role |
|---|---|
| Emergence — the S-curve as a tug-of-war between a reinforcing growth loop and an inhibiting loop | theoretical foundation |
| 3X's — naming Explore/Expand/Extract as the three phases and stating the central thesis that each requires its own approach across nine dimensions | central thesis statement |
| Explore — Risk/Goal/Strategy/Tactics/Exit breakdown | elaboration/evidence for the central thesis |
| Expand — Risk/Goal/Strategy/Tactics/Exit breakdown | elaboration/evidence for the central thesis |
| Extract — Risk/Goal/Strategy/Tactics/Exit breakdown | elaboration/evidence for the central thesis |
| All Of The Above — portfolio management across simultaneous mixed-phase projects | conclusion / practical implication |
Key terms
- Growth loop (reinforcing loop) Growth Loops & Systemic Feedback
A self-reinforcing feedback mechanism that makes an idea's continued growth easier the more it operates — the engine of the early, accelerating part of the S-curve. Must be found, not designed in advance ('it needs to be new or somebody else would already be operating it'). - Inhibiting loop Growth Loops & Systemic Feedback
A self-limiting feedback mechanism that eventually takes over from the reinforcing loop, slowing and capping growth; in Expand there are multiple, successive inhibiting loops ('potentially-fatal') that must be overcome one at a time. - Explore
The phase whose goal is finding a new growth loop through rapid, low-cost, maximally creative experiments run by tiny, dependency-free teams that quickly discard failures. Risk: nobody cares and the idea dies for lack of fuel. - Expand
The phase of keeping a found growth loop operating long enough to scale, by intensely focusing on whichever growth bottleneck is currently emerging — even at the cost of throttling growth and discarding non-essential features. Risk: the idea can't scale. - Extract
The phase of harvesting profit once the ultimate inhibiting loop has capped growth, via small, safe optimization experiments rather than further aggressive growth-seeking. Risk: the idea can't sustain itself.
Propositions
- explicitThe growth of any idea — product, company, or movement — traces a logistic (S-shaped) curve.conf: high
- explicitThe S-curve is produced by a tug-of-war between a reinforcing growth loop (which makes growth easier as it accumulates) and an inhibiting loop (which eventually slows and caps growth).conf: high
- explicitA successful idea moves through three sequential phases: Explore (find a new growth loop), Expand (keep it operating and out-survive the inhibiting loops long enough to scale), and Extract (harvest profit as the ultimate inhibiting loop caps growth).conf: high
- explicitdevelopedEach of the three phases requires its own distinct approach across finance, team size, project management, personnel, technology, risk management, implementation, marketing, and sales — no matter how superficially similar the phases look from the outside.conf: highextends ←Reject Change, Sometimes · p3 — Elaborates the prior 3X essay's claim that each phase requires its own distinct approach by supplying the specific volatility/risk-strategy dimension (symmetric-flip/short-vol for Extract, skewed-flip/long-vol for Explore, ruin-avoidance for Expand) that the earlier essay left unspecified.
- explicitApplying the approach appropriate to one phase to a project that is actually in a different phase kills the idea.conf: high
- explicitdevelopedExplore's core risk is that nobody cares and the idea starves for lack of fuel; the appropriate response is rapid, maximally creative, cheaply discardable experiments run by tiny, dependency-free teams, exiting only once one experiment clearly outperforms the rest.conf: highextends ←90% of My Skills Are Now Worth $0 · p3 — Beck's prescription here ('try a little bit of a lot of ideas' because the 90/10 split can't be known in advance) is a personal, skill-level instance of Explore's own tactic of rapid, cheaply-discardable experiments — applied years before the 3X framework was formally published.
- explicitdevelopedExpand's core risk is failing to scale; the appropriate response is intense, sequential focus on whichever growth bottleneck is currently emerging — including deliberately throttling growth and discarding non-essential features — exiting once growth's cause and effect become predictable.conf: highsupportsStarving Genies · p1 — Canon's own Expand-phase tactics ('intense focus on the next emerging growth bottleneck', 'throttle growth, discard non-essential features') are the general framework of which starving-genies' claim — that Expand-phase growth is a staircase of successive rate-limiting resources to overcome one at a time — is the specific restatement applied to AI providers.extends ←Reject Change, Sometimes · p6 — Adds a specific mechanism — investing to reduce the probability of a catastrophic 'death' outcome, alongside increasing the probability of clearing the bottleneck — to the earlier essay's more general claim that Expand's core risk is failing to scale and the response is focus on the current growth bottleneck.
- explicitExtract's core risk is failing to sustain profitability; the appropriate response is safe, incremental optimization rather than further aggressive growth-seeking, exiting only once there is no further return on investment.conf: high
- explicitA healthy portfolio runs projects in different phases simultaneously by design — Extract products fund a portfolio of Explore projects — and when a project enters Expand, it should be treated as the priority even over currently more profitable Extract activity, though this is far easier to state than to execute.conf: high
- explicitThe Canon article series (of which this is an entry) is explicitly intended to explain Beck's ideas as plainly and unambiguously as possible — no analogies, no persuasion, just the facts.conf: highsupportsThe Beginnings of an Idea: XP is Long Volatility · p7 — This Canon entry is the concrete teachable-form endpoint that xp-is-long-volatility#p7 predicts 3X eventually reached via years of incubation and repeated bad explanations — its stated method (plain, unambiguous, no analogies, no persuasion) is the polished product of exactly that refinement process.
Arguments
You must not apply one phase's approach to a project that is actually in a different phase. strength: high
Explicit premises:
- Each phase (Explore/Expand/Extract) demands a categorically different approach across finance, team size, project management, personnel, technology, risk management, implementation, marketing, and sales.
- The phases look superficially similar from the outside even though they require incompatible approaches.
Implicit premises:
- An organization can, in practice, correctly identify which phase a given project is actually in.
- The cost of a phase-mismatched approach is severe enough to kill the idea outright, not merely slow it down.
Inferences:
- Because phase identification itself is never operationalized, an organization that misjudges its own phase will apply the 'correct' approach to the wrong problem and still kill the idea — the framework's central risk can be triggered by misdiagnosis alone, independent of how well the chosen approach is executed.
A portfolio should treat an Expand-phase project as the priority, even over more profitable Extract-phase activity.
Explicit premises:
- Extract products fund the portfolio's Explore projects.
- Expand is the phase in which a found growth loop must survive a sequence of potentially fatal inhibiting loops before it can be captured at all.
Implicit premises:
- Missing the scaling window during Expand is effectively irreversible, whereas Extract's ongoing optimization can absorb temporary underinvestment without permanent loss.
- Extract's returns are more immediately legible and measurable than Expand's, which is part of why they compete unevenly for resources.
Inferences:
- Organizations that let Extract's more legible, immediately-measurable returns crowd out Expand investment are making a systematically biased resourcing error, not a neutral trade-off — which is consistent with Beck's own admission that the prioritization is 'simple to say and apparently nearly impossible to execute.'
Narratives
- Beck introduces the S-curve's mechanism as 'a tug-of-war ... between two feedback loops' and immediately flags it parenthetically: '(oops, analogy, sorry)'.
→ Implicitly concedes that the Canon series' stated method — plain, unambiguous, analogy-free explanation — breaks down at the exact moment it needs to explain an abstract systems concept; the self-aware slip is the closest thing in the piece to a narrative, and it undercuts the piece's own methodological claim rather than supporting a substantive proposition.
Questions addressed
- ✅ Why do growth strategies that worked earlier suddenly stop working as an idea matures, even though growth still looks continuous?
Because the phase changed underneath the strategy — the smoothness of the S-curve is 'a dangerous illusion'; an approach tuned for one phase becomes actively harmful once the project has moved into a different phase. - ✅ How should an organization allocate resources and attention across a portfolio of projects at different maturity stages?
Fund Explore projects from Extract profits, and prioritize an Expand-phase project even above currently more profitable Extract work when they compete for resources. - ⭕ How does an organization determine, in real time, which phase a given project is actually in?
Gaps
- incompleteThe framework gives no prospective way to diagnose which phase a project is currently in. Every exit criterion offered is retrospective — Explore's exit is 'one experiment works way better than others,' Expand's is 'cause and effect of growth become predictable,' Extract's is 'no further return on investment' — each recognizable only after the transition has already happened. Since Beck's highest-stakes claim is that applying the wrong-phase approach kills ideas, and phase identification is the one capability that claim depends on, the piece asserts the stakes of misdiagnosis while leaving diagnosis itself unaddressed.
Read next: Read against lean-startup / product-discovery literature on leading indicators of product-market fit, which attempts exactly this real-time diagnosis problem; also reread Explore's own tactic ('quickly discard failures') for whether Explore's fast-failure discipline is itself an implicit workaround for the missing diagnostic — i.e., you don't diagnose the phase, you make Explore cheap enough that misdiagnosis is affordable. - illogicalThe Canon series promises to explain ideas 'no analogies, no persuasion, just the facts,' yet the load-bearing explanatory device of this very essay — the S-curve as a tug-of-war between two feedback loops — is an analogy, self-flagged mid-sentence ('oops, analogy, sorry'). The piece doesn't reconcile this with its stated method or revise the promise, leaving open whether an abstract systems concept like a feedback loop can be explained non-analogically at all, or whether 'no analogies' is a rhetorical framing device rather than a constraint the writing actually follows.
Read next: Compare with Canon TDD (the first entry in the series, referenced but not read here) to see whether it holds the no-analogy constraint more successfully, or whether the constraint reliably breaks down whenever Beck needs to explain a systems-level abstraction rather than a concrete practice. - incompleteBeck names the practical failure mode of portfolio management ("prioritize Expand even over profitable Extract ... simple to say and apparently nearly impossible to execute") but doesn't explain the mechanism behind the difficulty — is it incentive misalignment (Extract's returns are more legible and immediately rewarded), organizational structure (separate teams/budgets per phase competing for the same pool), or plain risk aversion? The failure is asserted, not analyzed.
Read next: Cross-reference extract-tbm-430-incubate-compound-refinance-liquidate for a more developed treatment of portfolio-allocation mechanics across investment-lifecycle stages, and organizational-behavior literature on why legible short-term metrics crowd out illegible but longer-term-critical investment.
Closes gaps elsewhere
- Starving Genies · g1 — Beck's own canonical statement of 3X (p9 here) explicitly assumes a portfolio runs mixed phases simultaneously by design ('some Extract products that pay the bills & pay for a portfolio of Explore projects') — the framework does not assume phase uniformity across projects, it is built to manage non-uniformity. This answers the gap's question about whether the model requires a shared phase: it doesn't, even though starving-genies' application of it to three AI providers reads as if it does.