Provisional commitment and reserved optionality
Two capabilities most institutions lack. Provisional commitment is a stance: commit firmly to a direction, hold the frame lightly, keep options open where learning is expected, and set triggers for revisiting. Reserved optionality is its counterpart in the portfolio: keep the ability to change course, and spend it only on purpose.
Kinds of not-knowing
“Living with ambiguity” is usually treated as a personality trait, or as a coping skill the rest of us can learn. Complexity thinking treats it as a capability that analysts and institutions can build. It has three parts: telling kinds of not-knowing apart, resisting closure that comes too early, and acting without first resolving everything.
The first part is diagnostic. Stirling (2010) separates risk, uncertainty, ambiguity and ignorance, and each needs a different response. The six dials page sets the four out in full. The short version is that the “ambiguous” in VUCA covers all four, and that many institutional failures start with the wrong diagnosis.
Seizing and freezing
Kruglanski and Webster (1996) showed that time pressure, fatigue and threat raise the need for cognitive closure. People “seize” on an early answer and then “freeze” on it. Frenkel-Brunswik (1949) had already described intolerance of ambiguity as a stable personality variable.
Put the two together and the problem is structural. A crisis raises the need for closure at exactly the moment ambiguity is highest. Telling people to tolerate ambiguity will not fix a mismatch built into the conditions, so living with ambiguity has to be designed into institutions, not asked of individuals.
The school met this in Term 2, Week 4. The checkpoint had just been marked, staff were tired and complaints were arriving through the app. One teacher quietly rated borderline students up, and the complaints stopped. The question of what the ratings meant was left less answered than before. Holding it open would have cost more that week, which is what resisting closure usually costs.
Ambiguity as a resource
Ambiguity is not only a cost. Eisenberg (1984) showed that strategic ambiguity lets different groups unite around a shared commitment that each reads in its own way. Weick (1995) treated equivocality, the presence of several plausible meanings, as the raw material of sensemaking. Collapse it too early and you throw information away.
Stark (2009) found that organisations which keep several principles of evaluation in tension are better at spotting new opportunities. Research on the paradox mindset finds that people who accept tensions as lasting, not as problems to eliminate, tend to perform better and innovate more when resources are scarce (Miron-Spektor et al., 2018; Smith & Lewis, 2022).
Whole-class feedback spread in Term 2, Week 10 to two more departments, and each changed it on the way. Nobody wrote down a single definition of what it was for. That looseness is part of why it spread.
Stances for the in-between
Several traditions describe the threshold between one order and the next. Negative capability, Keats’ phrase for being “in uncertainties, mysteries, doubts” without reaching irritably for fact and reason, has been revived as a leadership capacity (Simpson et al., 2002). Turner (1969) described liminality, the phase of a transition when old structures are suspended and new ones not yet formed, as both dangerous and generative.
Gramsci (1971) named the political version, the interregnum in which “morbid symptoms” appear. Haraway (2016) urged staying with the trouble instead of reaching for redemptive or apocalyptic closure. Stein (2019) describes the present as a “time between worlds”. Together they describe the condition that crisis thinking diagnoses, and the stance it asks for.
Holding environments
Heifetz (1994), adapting Winnicott, argued that leaders working on adaptive challenges must create a holding environment: a setting that contains distress enough for people to do difficult work without fleeing into denial or early closure. Psychological safety is its form at the level of a team (Edmondson, 1999). Wong (2026) makes the same point about the metacrisis. The obstacle is often not a lack of data but a lack of the confidence and emotional safety to act.
A holding environment is the institutional design that lowers the need for closure enough for ambiguity to be worked with, not escaped. The staff meeting in Term 1, Week 10 was a small one. English could agree to one probe and decline two, and the meeting still ended with something to try.
Three stances
Indecision, conviction, and the one between
Provisional commitment is not a midpoint. It commits firmly to a direction and holds the frame lightly.
| Indecision | Conviction | Provisional commitment | |
|---|---|---|---|
| Action | Deferred until clarity arrives | Immediate and total | Immediate, staged and reversible where possible |
| Learning | Passive: waiting for information | Closed: evidence filtered to fit | Active: probes designed to learn |
| Frame | Many frames, none chosen | One frame, held as truth | One working frame, held as partial |
| Revision | Constant and unstructured | Resisted | Scheduled, and triggered by signposts |
| Failure mode | Drift, and options spent by default | Escalating commitment to a failing course | Over-hedging, if triggers are never set |
The second capability
Reserved optionality
The principle
A flexible ticket costs more than a fixed one. People buy it when they expect their plans to change. Reserved optionality is the same reasoning applied to institutions, and it recurs across finance, ecology, engineering, policy and strategy. When the future is uncertain and learning is expected, the ability to change course later has value. That value should be protected, not spent by default.
Complexity thinking supplies the deeper reason. In an adaptive system the future is not only unknown but under construction. What other people do next depends partly on what you do now. So the option to respond to what emerges is often worth more than the efficiency gained by committing early.
Why it differs from optimisation
Optimisation asks what is best, given a model of the future. Optionality asks what keeps us able to act well across the futures the model cannot see. The two can coexist. In practice optimisation usually wins, because its gains are visible and immediate, while the value of an option is invisible until the moment it is needed.
Spending options invisibly
Most option loss is unintended. Single-supplier contracts, monoculture platforms, narrow skill pipelines, lean staffing with no surge capacity and policy designs that hard-code assumptions all spend options. None of them looks like a decision to give up flexibility.
Complexity thinking adds the discipline of an option audit. Which options is this plan spending, and did anyone decide to spend them?
The commitment paradox
Optionality is not always the answer. Commitment has value of its own: it signals credibility, coordinates others, deters rivals and makes investment possible (Ghemawat, 1991). Options kept open indefinitely turn into indecision.
Some problems need options destroyed on purpose. Levin et al. (2012) describe super wicked problems: time is running out, those trying to solve the problem also cause it, no central authority exists, and policies irrationally discount the future. For these they argue for policies that “constrain our future selves”, through stickiness, entrenchment and positive lock-in.
Complexity thinking holds both sides. Preserve options where learning is expected and reversal is cheap. Create deliberate lock-in where delay is the larger risk and path dependence can be put to work. The skill lies in telling the two apart. The grid is a first cut at it.
The family
Eleven ideas about keeping the future open
Different fields arrived at the same idea under different names. Each name brings a usable method.
| Concept | Source | The idea | Use it for |
|---|---|---|---|
| Real options | Myers (1977); Dixit & Pindyck (1994); Trigeorgis (1996); McGrath (1999) | The flexibility to defer, expand, switch or abandon has a value that can be quantified under uncertainty. | Staging investments; valuing pilots |
| Quasi-option value | Arrow & Fisher (1974) | Where decisions are irreversible and learning is expected, keeping options until more is known has value. | Environmental, infrastructure and regulatory choices |
| Reversibility | Popularised as “one-way and two-way doors” (Bezos, 2016) | Make reversible decisions quickly and lightly. Make irreversible ones slowly and carefully. | Decision governance |
| Slack | Cyert & March (1963); Bourgeois (1981); Nohria & Gulati (1996) | Uncommitted resources absorb shocks and fund exploration. The benefit rises, then falls. | Budgets, capacity, staffing |
| Redundancy and degeneracy | Edelman & Gally (2001) | Redundancy duplicates identical parts. Degeneracy lets structurally different parts perform the same function, which resists common-mode failure. | Supply chains, workforce and infrastructure design |
| Modularity and loose coupling | Simon (1962); Weick (1976) | Semi-independent units contain failure and allow local adaptation. | Organisation and system architecture |
| Efficiency–resilience balance | Ulanowicz et al. (2009) | Sustainable systems sit in a “window of vitality” between brittle efficiency and stagnant redundancy. | Network and portfolio design |
| Barbell strategy | Taleb (2012) | Combine very safe bets with very exploratory ones, and avoid the fragile middle. | Portfolio and innovation strategy |
| Affordable loss | Sarasvathy (2001) | Commit only what you can afford to lose, and let opportunities emerge. | Entrepreneurial and pilot decisions |
| Adaptive pathways | Haasnoot et al. (2013); Marchau et al. (2019) | Sequence actions with signposts and triggers, and keep alternative routes open until tipping points. | Long-horizon policy and infrastructure |
| Robust decision making | Lempert et al. (2003) | Prefer actions that perform acceptably across many futures over actions that are optimal in one. | Strategy under deep uncertainty |
The grid
When to keep options open, and when to lock in
Provisional commitment and reserved optionality are two sides of one capability. The first concerns the mind and the second the portfolio. A leader who can hold ambiguity but whose institution has no options is trapped. An institution with options but leaders who need closure will spend them in the first crisis.
High learning, low cost of delay
Reserve options. Stage, pilot and defer irreversible commitments.
High learning, high cost of delay
Act on adaptive pathways. Move now, keep switching routes open, and set signposts and triggers.
Low learning, low cost of delay
Decide on efficiency grounds.
Low learning, high cost of delay
Commit firmly. Consider deliberate lock-in to make the commitment durable.
At the school
Both capabilities, across one term
The contract review in Term 2, Week 6
The finance committee wanted the five-year extension, and on its own terms the committee was right. The discount was real and the system was in use. On the grid, though, the decision sat where learning is high and delay is cheap. Learning was expected: the promised guidance on reporting for lower secondary had not arrived, departments still read the five-point scale two ways, and parents had just shown what they make of it. The cost of delay was low, because a one-year extension kept the system running. That cell says stage the commitment. Mrs Lim took one year at a higher price.
She had to defend it to the principal because the value of an option is invisible until it is needed. The premium shows on a budget line. What it bought, the chance to change the scale once the guidance arrives and once the school knows what a rating should say, shows nowhere. An option audit would have named what the five-year deal was spending: the five-point scale, fixed on every results slip the current Secondary 1 students would receive before they leave Secondary 4, and given up without anyone deciding to give it up. If the guidance changes nothing and the scale settles, the premium will look wasted. That is what an unused option always looks like.
The target nobody could drop
The 95 per cent target was written as a working figure. In Term 2, Week 5 the dashboard showed 96, nobody in the school trusted it, and the cluster newsletter cited the school as an exemplar of “feedback culture”. Praise turned a provisional number into a public commitment, and nobody in the school chose that. Commitment coordinates others, as Ghemawat says. Here it did so against the school: other people now counted on the figure.
By the end of the semester Mrs Lim had proposed dropping the format check from book checks, and the cluster still wanted the compliance figure. That is one commitment held two ways, and it sits uneasily with everyone who knows. Ms Tay’s proposal that the cluster stop collecting the twice-a-term ratings is the nearest thing to an exit, it does not touch the compliance figure, and it is not the school’s decision. The lesson for next time is modest. A target meant to be provisional should be published as provisional, with the condition for revisiting it written in from the start.
What to lock in, and when to look again
Not everything should stay open. Where little more will be learned and delay is costly, the grid says commit, and make it stick. After the checkpoint in Term 2, Week 4, the school knows enough to rule that no rating rests on one timed paper alone, and every term without that rule risks another student like Aisyah. The unrecorded quiz is a second candidate. Whatever the promised guidance says, a short quiz that finds misconceptions in weeks performs acceptably across its likely versions, which is the test Lempert et al. (2003) set for a robust choice. Locking in that it stays unrecorded is the harder part, because leaders keep asking for the data.
Everything left open needs a signpost and someone who owns it, or provisional commitment slides into over-hedging. The semester suggests the signposts. The guidance on reporting for lower secondary, when it arrives, reopens the scale and the contract. Any change the supplier makes to how ratings are shown brings the contract conversation forward. Green-pen responses that converge on the same phrase, as they did on “I will use more evidence”, mean the format needs changing. One group of students rated Emerging across every subject calls for a look at what the ratings judge, not at the students. And when a class asks for a probe to count, ask why before saying yes. None of these triggers fired on its own in this semester. Someone had to notice each one.
Reading
Where it comes from
Start with the first on the list.
Kruglanski, A. W., & Webster, D. M. (1996). Motivated closing of the mind: “Seizing” and “freezing”. Psychological Review, 103(2), 263–283.
Why pressure, fatigue and threat push people towards early answers, and keep them there.
Weick, K. E. (1995). Sensemaking in organizations. Sage.
Equivocality as the raw material of sensemaking, and the cost of collapsing it early.
Heifetz, R. A. (1994). Leadership without easy answers. Belknap Press of Harvard University Press.
Adaptive challenges and the holding environment that lets people work on them.
Simpson, P. F., French, R., & Harvey, C. E. (2002). Leadership and negative capability. Human Relations, 55(10), 1209–1226.
Keats’ negative capability recast as something leaders can practise.
Arrow, K. J., & Fisher, A. C. (1974). Environmental preservation, uncertainty, and irreversibility. The Quarterly Journal of Economics, 88(2), 312–319.
The original case for the value of keeping options when decisions cannot be undone.
McGrath, R. G. (1999). Falling forward: Real options reasoning and entrepreneurial failure. Academy of Management Review, 24(1), 13–30.
Real options reasoning brought out of finance and into pilots and ventures.
Haasnoot, M., Kwakkel, J. H., Walker, W. E., & ter Maat, J. (2013). Dynamic adaptive policy pathways: A method for crafting robust decisions for a deeply uncertain world. Global Environmental Change, 23(2), 485–498.
Signposts, triggers and switching routes, worked out as a method.
Levin, K., Cashore, B., Bernstein, S., & Auld, G. (2012). Overcoming the tragedy of super wicked problems: Constraining our future selves to ameliorate global climate change. Policy Sciences, 45(2), 123–152.
The counter-case: when the right move is to destroy options on purpose.
Unsettled
What I would still argue about
The questions this page does not settle.
- Who owns a revisit trigger when the person who set it moves on, and what happens to a trigger nobody remembers setting?
- How should a school account for an option it paid for and never used, so that the next renewal does not read the premium as waste?
- Can a target be published as provisional without being read as a lack of resolve by the office it is reported to?
- Where does a holding environment end and a habit of deferring hard decisions begin?