Improvisation Learning Harvest¶
Learning-harvest ritual — instantiates Constraint-Guided Improvisation
Reviews the novel moves people improvised in the field and decides which to promote into the sanctioned repertoire and grammar — turning one-off saves into reusable, reviewed practice and catching precedent before it sets unreviewed.
Improvisation Learning Harvest is the recurring session that looks back at the moves people actually invented on the spot and decides which of them should become part of the canon. Its defining move is a gate: each provisional, off-repertoire move is judged — promote it to sanctioned practice, keep it as a flagged last-resort exception, or drop it — so the repertoire grows from real experience instead of only from what was planned. It is deliberately two-sided. It captures the hard-won improvisation that worked and would otherwise evaporate, and it catches the improvisation that happened to work before it quietly hardens into unreviewed precedent nobody vetted. It asks "which of these should we keep, and how do we encode it?" — not, like a retrospective, "what happened and why?"
Example¶
At 2 a.m. during an outage, an on-call engineer improvises a mitigation that is in no runbook — re-routing traffic in a way no one had tried — and it holds. Weeks later the team runs its harvest over the improvised moves from recent incidents. This one gets examined: was it luck, or a real pattern that generalizes? They judge it sound, so they promote it — add it to the runbook (the repertoire), write a rule into the incident grammar (when this subsystem saturates, prefer the re-route before scaling out), and record the rationale and the conditions under which it does not apply.
A second improvised move from the same window — a hand-edit to the production database that also "worked" — is examined and deliberately not promoted: it is logged as a sanctioned-exception-only maneuver requiring sign-off, precisely so it does not slide into normal practice by having gotten away with it once. The repertoire grew on evidence, and a dangerous precedent was stopped from setting silently.
How it works¶
- Operate on provisional moves. The input is specifically what people did outside the sanctioned repertoire — the improvisations, not the by-the-book actions.
- Apply the promotion gate. Each move is routed to one of three outcomes: promote to sanctioned, keep as flagged exception, or discard.
- Encode what's promoted. A kept move lands as a repertoire entry, a new rule in the composition grammar, or both — not merely as a story told once.
- Record the decision, not just the move. Capture the rationale and the boundary conditions, so a later reader knows when the move applies and when it does not.
Tuning parameters¶
- Promotion bar — how much evidence a move needs before it is canonized. A high bar prevents codifying luck but grows the repertoire slowly; a low bar learns fast but risks turning a fluke into a rule.
- Harvest cadence — how often the session runs. Frequent harvests catch moves while the context is fresh but cost overhead; rare ones batch efficiently but lose detail and let drift normalize.
- Review scope — every improvised move versus only the consequential ones. Broad scope catches quiet precedent; narrow scope keeps the session tractable.
- Encoding target — whether a promotion updates the repertoire, the grammar, or both. Grammar rules generalize further but are far harder to get right.
- Exception aggressiveness — how firmly un-promoted moves are marked as exception-only rather than simply not adopted.
When it helps, and when it misleads¶
Its strength is that it is the archetype's forward learning loop: the repertoire evolves from what the situation actually taught, and unreviewed precedent — the failure where a one-off quietly becomes de facto policy — gets caught at a gate instead of setting by default. Because it revises the governing repertoire and grammar rather than just fixing the single incident, it is a form of double-loop learning.[1]
It misleads through hindsight: a move that happened to work reads, afterward, as obviously right, and over-eager promotion codifies luck into a brittle rule. The classic misuse is running it backwards — as a blame exercise aimed at the improviser, or as a rubber stamp for what leadership already meant to standardize, rather than an honest test of whether the move generalizes. The discipline is to separate "did it work this time" from "should it be a rule," to weigh the move on evidence of generalization, and to keep the recorded-but-not-promoted path open so exceptions stay visible without becoming policy.
How it implements the components¶
Improvisation Learning Harvest realizes the learning side of the archetype — the machinery that runs after moves are made:
provisional_move_learning_gate— it is the gate that decides which off-repertoire moves are promoted, kept as exceptions, or dropped.learning_record— it records each decision with its rationale and the move's boundary conditions.composition_rule_grammar— promoted moves are encoded as new rules for how moves may be combined, updating the grammar the archetype improvises within.
It does not drill promoted moves to fluency (that is Variable Scenario Rehearsal), run the blow-by-blow incident retrospective (After-Action Review), or state the fixed constraints a move must respect to be promotable (Constraint Backbone Brief).
Related¶
- Instantiates: Constraint-Guided Improvisation — grows the repertoire and grammar from moves the situation actually produced.
- Consumes: After-Action Review surfaces the improvised moves and their context; the harvest is the gate that then judges which to keep.
- Sibling mechanisms: Variable Scenario Rehearsal · After-Action Review · Constraint Backbone Brief · Shared Situation Model · Decision-Rights Matrix · Move-Announce-Acknowledge Cycle · Facilitated Turn-Taking · Call-and-Response Pattern · Tactical Reset Point · Minimum Safe Stabilization · Rollback or Pause Protocol
Editorial Notes¶
Form Classification¶
Form family: Decision, Gate & Allocation
Rationale: Improvisation Learning Harvest operates as a case-specific gate, selection, routing, prioritization, or resource disposition because it reviews the novel moves people improvised in the field and decides which to promote into the sanctioned repertoire and grammar — turning one-off saves into reusable, reviewed practice and catching precedent before it sets unreviewed
Independent corroboration: The frozen evidence defines Improvisation Learning Harvest as 'Reviews the novel moves people improvised in the field and decides which to promote into the sanctioned repertoire and grammar — turning one-off saves into reusable, reviewed practice and catching precedent before it sets unreviewed', so its operative form is Decision, Gate & Allocation.
Review outcome: Independent reviewer agreement; high confidence.
Origin Attribution¶
Primary origin: Organizational & Management Science
Origin pattern: Cross-disciplinary synthesis
Present-day reach: Multi-domain
Rationale: Capturing improvised action and revising governing assumptions follows organizational learning and Argyris-Schon double-loop learning.
Related originating lineages:
- Education & Pedagogy — Reflective practice materially turns experience into teachable future capability.
- Performing Arts & Theatre Studies — Improvisational practice supplies repertoire, responsiveness, and reflection on emergent moves.
Review resolution: Both reviewers independently assign organizational_management as the primary originating domain, so that shared primary is retained. Alternate domains are the union of reviewer-identified formative or independently originating lineages; later application settings alone are excluded. The final form materially composes methods or concepts from more than one formative domain. It has established independent use across several domains, but that does not make it domain-free. The encyclopedia entry makes that composition explicit.
Encyclopedia synthesis: The exact catalogued form synthesizes established practice rather than reproducing a single standard historical label.
Review outcome: Reconciled after independent review; high confidence.
Notes¶
The harvest's most valuable output is often the move it refuses to promote — the precedent flagged as exception-only before it could normalize. It pairs with Variable Scenario Rehearsal as the two halves of one cycle: the harvest decides what enters the repertoire, rehearsal makes it fluent. Where it sits next to an After-Action Review, the review establishes what happened and the harvest decides what to keep — complementary meetings, not the same one.
References¶
[1] Argyris, C., and Schön, D. A. Organizational Learning: A Theory of Action Perspective. Addison-Wesley (1978). Defines double-loop learning as revising governing norms and assumptions rather than only correcting errors within them. registry ↩