- Social licence is not a legal instrument, but its absence is more expensive than most legal disputes. A $3–5 billion project loses roughly $20 million per week in net present value when community conflict delays it.
- Across 200 infrastructure projects in 20 Latin American and Caribbean countries, 81% were delayed and 58% overran cost because of conflict — and 18% were cancelled outright.
- Social licence operates at three levels: legitimacy, credibility, trust. Projects that never get past legitimacy are vulnerable to any triggering event, however well approved.
- Four gaps manufacture failure: the approval gap, the trust gap, the timing gap, and the distribution gap. Each is a decision, not an accident.
- Engagement after design lock-in is the most common structural error. Community intelligence is treated as a communications input rather than an engineering one.
- The evidence on remedy is specific: where community-based monitoring and shared decision-making were implemented, delays were halved and cost overruns reduced by over 10%.
Most project teams treat social licence as a stakeholder management task that runs alongside engineering. That is the wrong frame and it explains the failure rate. Social licence is a delivery risk manufactured in four gaps — between legal permission and community acceptance, between legitimacy and trust, between when the community is consulted and when the design is locked, and between who receives the benefits and who carries the burdens. This guide names each gap, prices it, and shows what the evidence says actually closes it.
On this page
- Gap 1 — The approval gap: legal licence is not social licence
- The three levels of social licence
- The price of the gap
- Gap 2 — The trust gap: why trust is the only durable level
- Gap 3 — The timing gap: engagement after design lock-in
- Gap 4 — The distribution gap: benefits and burdens
- The consent gap: FPIC and Indigenous rights
- Can social licence be measured?
- What the evidence says actually works
- Misinformation and the new opposition
- A practical social licence checklist
- Frequently asked questions
1. Gap 1 — The Approval Gap: Legal Licence Is Not Social Licence
The distinction is the foundation of everything that follows, and it is routinely collapsed in project language.
A legal licence is granted by a government authority. It is documented, appealable, and enforceable. A social licence is granted by communities and stakeholders. It is undocumented, unappealable, and revocable at any time. It exists on a continuum from rejection through acceptance to approval to full identification with the project.
The gap between them is where projects die. The Northern Gateway pipeline received its certificates to construct and operate after a comprehensive regulatory review, and remained stalled awaiting a social licence. The Dakota Access Pipeline held regulatory permits and faced months of protest and international attention; the project's cost to its owners and co-owners reached at least $7.5 billion against an original budget of $3.8 billion, largely because of delays tied to the opposition.
The Moonlight Range Reversal
Queensland's Moonlight Range Wind Farm — 450 MW, 88 turbines, battery storage — was conditionally approved by the State Assessment and Referral Agency in late 2024. In January 2025 the Deputy Premier exercised ministerial call-in powers, and in May 2025 refused planning consent. The stated reasons included insufficient community consultation prior to lodgment, concern that a 300-person construction workforce would strain local housing, and minimal perceived local benefit. 88% of local resident submissions objected. A project with statutory approval was refused on social licence grounds.
That case is not an anomaly; it is the emerging pattern. Queensland subsequently introduced mandatory social impact assessments and Community Benefit Agreements for new wind farms, solar farms above 1 MW, and large battery storage above 50 MW. Social licence moved from an informal expectation to a statutory consideration.
The Legitimate Critique
There is a real objection to the social licence concept that engineers should understand rather than dismiss. In its strongest form: social licence “has somehow broadened to imply a right to veto an approved project, a result that potentially undermines the rule of law.” That critique is not wrong as a description of the risk. The practical answer is not to abandon the concept but to be precise about what it is: a delivery risk, not a legal veto. The community does not need a veto to impose a two-year delay and a 40% cost overrun. The veto is the legal system's job; the delay is the project's problem.
2. The Three Levels of Social Licence
Social licence is not binary. It operates at three cumulative levels, and a project's resilience depends on which level it has reached.
| Level | What it requires | What it survives |
|---|---|---|
| Legitimacy (baseline) | Legal compliance and conformance to community norms | Almost nothing — any triggering event can collapse it |
| Credibility | Consistent follow-through on commitments, transparent communication, demonstrated competence | Routine setbacks and minor incidents |
| Trust (highest) | Ongoing relationship characterised by mutual regard and a shared sense of purpose | Significant setbacks, including incidents that would destroy a legitimacy-level project |
Levels are cumulative, not alternative. A project cannot be at credibility without legitimacy, or at trust without credibility.
Why Legitimacy Is the Dangerous Level
A project that is legally compliant and nothing more is operating at legitimacy level. It has no credit with the community, so the first serious incident — a spill, a noise complaint that is not resolved, a contractor behaving badly — becomes the trigger for organised opposition. The project team experiences this as disproportionate, because from a compliance standpoint nothing has gone wrong. From a relationship standpoint, everything has: there was never a relationship to draw on.
3. The Price of the Gap
The cost of social licence failure is not soft, and it is not speculative. It has been quantified across multiple independent datasets.
| Source | Finding | What it measures |
|---|---|---|
| Franks et al., PNAS (2014) | A $3–5 billion capital project loses roughly $20 million per week in net present value during conflict-induced delay | NPV destruction rate |
| Franks et al., PNAS (2014) | Delays of 2–3 years and cost overruns of up to 40% | Schedule and cost impact |
| IDB, 200 projects in 20 countries | 81% of quantified cases delayed; 58% overran cost; 18% cancelled outright | Project-level consequences |
| IDB, same dataset | 42% required redesign; 7% relocated; 13% changed joint-venture participants | Scope consequences |
| IDB, same dataset | 95% recorded reputational damage; 30% redress payments; 20% fines; 27% concession amendments | Legal and administrative consequences |
| Documented case | Nine-month conflict-related delay costing $750 million in additional project cost | Single-project magnitude |
| Social management failure | Overruns of 11–68% of total budget attributable to inadequate social management | Budget impact range |
| New Zealand, 2015–2025 | More than $69 billion in infrastructure projects impacted by politicisation and loss of social licence | National pipeline impact |
Figures are not directly comparable across studies: they use different asset classes, conflict definitions, and discount rates. The consistent finding is that conflict produces schedule and cost impacts of a magnitude that exceeds most technical risks on the same project.
The Number That Reframes the Business Case
Twenty million dollars per week is not a public relations budget. It is a rate of value destruction comparable to the cost of a major earthworks package, accruing every week the conflict continues. A project that spends $5 million on genuine engagement to avoid a twelve-week delay at that rate has not spent money — it has avoided $240 million of loss. That arithmetic is why the business case for social licence does not depend on ethical arguments, although the ethical arguments are also correct.
The IDB dataset also contains a finding that is rarely quoted and matters for strategy: in 86% of cases firms took some action to address the conflict, but in 91% of cases the action was limited to press statements. Firms were far more likely to issue a statement than to conduct genuine consultation once conflict escalated. That is the behavioural signature of treating social licence as a communications problem: the response to opposition is messaging, not change.
4. Gap 2 — The Trust Gap: Why Trust Is the Only Durable Level
The second gap is between being credible and being trusted. Most project teams believe they are in the first category and assume it is the second.
The 2026 I2S research, built on more than 1,500 survey responses across a multi-year Australian study and analysed using structural equation modelling, identifies the drivers of community acceptance. The headline finding is that perceived benefits and perceived risks have the largest effect sizes on community acceptance — and, critically, that they are also the mechanisms through which other predictors transmit their influence.
| Driver | What it means in practice |
|---|---|
| Place-based approach | Address the specific perceived impacts and risks of this site, not generic project benefits |
| Clear social value proposition | Speak to local benefits and to broader societal liveability, not only to the sponsor's case |
| High-quality engagement | Build trust and quality relationships rather than transmitting information |
| Engagement–decision link | Engagement must visibly shape operational and technical decisions |
| Perceived benefits and risks | The largest direct effects, and the channel through which the other drivers operate |
Drivers from the I2S Social Licence for Infrastructure Model, derived from more than 7,000 community responses collected 2021–2023 and published in Corporate Social Responsibility and Environmental Management, 2026.
The Direction of Flow
One finding from the same research programme is counter-intuitive and strategically important: social licence flows bottom-up, not top-down. Broader social licence around an issue — climate action, for example — does not trickle down to a local-level social licence. The implication is direct: a renewable energy project cannot borrow legitimacy from the urgency of decarbonisation. Its social licence is determined project by project, community by community, and the actions of every actor in the sector affect the licence available to the next project.
That last point explains a pattern that frustrates developers. A transmission company with an exemplary engagement record can find its social licence degraded by the behaviour of a different company in the same region. Social licence operates at project, community, and industry levels simultaneously, and the industry-level licence is a shared resource that any participant can deplete.
5. Gap 3 — The Timing Gap: Engagement After Design Lock-In
The third gap is structural, and it is the one most within the engineering team's control. It is the gap between when the community is consulted and when the design is locked.
A 2026 study of civil works in socio-economically sensitive regions described the failure mode with unusual precision: engineering teams routinely design site layouts, logistics corridors, and construction sequencing without integrating real-time community intelligence, while community relations departments operate in parallel, responding reactively to grievances after technical decisions have been locked. That structural separation produces predictable failure modes: unanticipated land access disputes, contractor–community friction over local employment expectations, and logistics bottlenecks triggered by poorly understood cultural or territorial sensitivities.
The Structural Separation
The study's diagnosis is worth quoting for its precision: “The prevailing project delivery paradigm has proven inadequate for regions where community variables are not peripheral constraints but central determinants of execution success.” The problem is not that community relations teams are incompetent. It is that they are organisationally separated from the engineering decisions that create the impacts they are asked to manage.
The remedy the study validated is not a better engagement programme. It is a different delivery architecture, built on four pillars that embed community variables into engineering execution rather than running alongside it.
| Pillar | Function |
|---|---|
| Community-Relations Engineering | Embed community intelligence into design reviews and technical decision-making |
| Logistics-Efficiency Mapping | Anticipate contractor–community friction over access, employment, and movement |
| Structured joint-venture partner cadence | Align delivery partners on community variables, not just commercial ones |
| HSE-embedded design | Treat social risk with the same rigour as safety risk |
Framework validated across projects in West Africa, Southeast Asia, and Latin America, and on a live greenfield site preparation campaign.
The measured result: zero community-induced work stoppages over an eighteen-month civil works phase, a 40% reduction in logistics-related grievances against baseline benchmarks, and measurably improved decision alignment between engineering, procurement, construction management, and community relations functions. Community dynamics moved from a primary schedule risk to a stabilising factor for critical-path adherence.
6. Gap 4 — The Distribution Gap: Benefits and Burdens
The fourth gap is between who receives the benefit and who carries the burden. It is the gap that the climate transition is currently making wider.
The pattern is consistent across project types. A wind farm delivers electricity to distant cities while imposing noise, visual impact, and land-use change on local residents. A transmission line carries renewable power to a metropolitan load centre while crossing agricultural land. A mine supplies materials for the energy transition while affecting the water and land of the host community. In each case, the project's societal benefit is real and the local burden is also real, and the two are borne by different people.
The Climate Benefit Does Not Confer a Social Licence
Renewable energy developers “often assume that the climate benefits of their projects automatically confer social licence — a dangerous misconception.” Local communities experience the visual, noise, land-use, and property-value impacts regardless of the climate case. This is the clearest illustration of the bottom-up principle: the global benefit does not trickle down to local acceptance.
Distributional fairness is not only about money. The 2026 research on social acceptance identifies procedural fairness, distributional fairness, relationship quality, and environmental protection as the factors that drive trust — and in at least one model, environmental protection was the most significant variable. A community that perceives the project as environmentally damaging will not be bought with a community benefit payment, because the payment does not address the variable that is driving their assessment.
| Fairness dimension | What the community is assessing | Typical sponsor error |
|---|---|---|
| Procedural fairness | Was the process open, consistent, and free of manipulation? | Consultation presented as a decision already taken |
| Distributional fairness | Are benefits and burdens shared proportionally? | Benefits calculated at regional or national scale, burdens at local scale |
| Relationship quality | Is there genuine mutual regard, or transactional contact? | Engagement resourced only when conflict appears |
| Environmental protection | Is the project causing avoidable environmental harm? | Assuming compensation substitutes for avoidance |
Dimensions drawn from SLO acceptance modelling in the Brazilian Amazon and related studies. Environmental protection was the most significant variable in that model.
The policy response is moving toward mandatory benefit-sharing. Queensland's Community Benefit Agreements require major renewable projects to demonstrate how they will deliver long-term benefits to affected communities, alongside mandatory social impact assessments. The direction of travel is clear: distributional fairness is becoming a condition of consent, not a voluntary extra.
7. The Consent Gap: FPIC and Indigenous Rights
For projects affecting Indigenous peoples' lands, resources, or rights, the social licence question becomes a consent question. The standard is Free, Prior and Informed Consent (FPIC), and it is a higher bar than consultation.
| Instrument | What it requires |
|---|---|
| ILO Convention 169 | Consultation with Indigenous peoples in good faith, through appropriate procedures, with the objective of achieving agreement or consent |
| UN Declaration on the Rights of Indigenous Peoples | FPIC for projects affecting lands, territories, and resources, particularly for relocation and hazardous activities |
| IFC Performance Standard 7 | FPIC where projects affect lands and natural resources traditionally owned or used, cause relocation, or affect cultural heritage |
| World Bank Environmental and Social Standard 7 | FPIC where activities affect lands and natural resources traditionally owned or used, or cause relocation |
The instruments differ in legal force but converge on the same substantive requirement: a good-faith process conducted before activities begin, with full information, and without coercion.
Two Things Engineers Get Wrong About FPIC
First, FPIC does not mandate consensus. It requires a good-faith process aimed at agreement; it is not a veto held by any individual or faction. Understanding this prevents both the sponsor's fear of paralysis and the community's false expectation of absolute control. Second, FPIC is prior. One documented case records that “construction had already started prior to application of IFC PS7” — which means the consent process was retrofitted to a decision already made. At that point it is not consent; it is notification with a negotiation attached.
The prior requirement is the one most often breached, and it is the one that connects back to the timing gap. A consent process that begins after mobilisation cannot satisfy the “prior” element, no matter how well it is conducted from that point. The decision to sequence engagement before construction is therefore not a communications decision — it is a compliance decision.
8. Can Social Licence Be Measured?
For years the answer was “not rigorously,” which allowed social licence to be dismissed as intangible. That has changed. Quantitative frameworks now exist, and their results are instructive precisely because they are unflattering.
A 2026 New Zealand study adapted the mining-derived Boutilier and Thomson model to forestry, producing the first quantitative SLO framework for that sector. The instrument used 15 Likert-scale statements, classified SLO into four levels, and achieved reliability with Cronbach's α above 0.84 across nine wood supply regions.
| Metric | Result | Interpretation |
|---|---|---|
| Overall SLO score | 2.9 | “Acceptance” level — not approval, not trust |
| Institutional trust | 2.60 | The lowest-scoring dimension |
| Environmental approval | 2.75 | Second-lowest; the sector's weakest legitimacy signal |
| Reliability | Cronbach's α > 0.84 | The instrument is internally consistent |
Scores from the New Zealand forestry SLO framework, Business Strategy and the Environment, 2026. Lower scores on trust and environmental approval indicate where intervention is required.
What Measurement Actually Reveals
The value of the framework is not the headline score. It is the disaggregation. An overall score of 2.9 tells a sector it is at acceptance level. The subscores tell it why: institutional trust is the weakest dimension, and environmental approval is next. Those are actionable. A sector that measures only its overall reputation learns nothing it can act on; a sector that measures its trust and environmental dimensions learns exactly where to invest.
For construction specifically, the measurement question is now being addressed through the I2S Social Licence for Infrastructure Model, which is the first model built from infrastructure-specific data rather than adapted from mining. Its practical contribution is that it identifies which levers move acceptance — perceived benefits and perceived risks — rather than treating social licence as a single reputation score to be improved by general goodwill.
9. What the Evidence Says Actually Works
The evidence base on remedy is smaller than the evidence base on cost, but it is specific enough to act on.
| Intervention | Measured effect | Source context |
|---|---|---|
| Community-based monitoring and shared decision-making | Project delays halved; cost overruns reduced by over 10% | Franks et al., PNAS, extractives sector |
| Community-integrated delivery model (four pillars) | Zero community-induced stoppages over 18 months; 40% fewer logistics grievances | 2026 civil works study, West Africa / SE Asia / Latin America |
| Early social risk identification and grievance mechanisms | Reduced magnitude and frequency of disruptive conflict | Franks et al.; IDB dataset |
| Structured preparedness against 20 Priority Actions | Performance improved on 19 of 20 actions between 2024 and 2026 | Energy Charter Better Practice Social Licence Guideline review |
| Place-based engagement and a clear social value proposition | Identified as drivers of acceptance in the infrastructure-specific model | I2S, 1,500+ survey responses |
Effects are not directly comparable across sectors and study designs. The consistent finding is that structured, early, decision-linked engagement outperforms reactive communications.
9.1 Grievance Mechanisms: The Cheap Early Warning System
An effective grievance mechanism is the least glamorous and most reliable social licence instrument. The criteria are well established: it must be accessible (multiple channels, local language), transparent (tracking and public reporting on complaints), responsive (timely investigation and resolution), and independent (perceived as fair rather than company-controlled). The UN Guiding Principles on Business and Human Rights specify the effectiveness criteria, and the IFC Performance Standards require mechanisms for financed projects.
The reason grievance mechanisms matter disproportionately is that they convert a latent conflict into a manageable case. A complaint that enters a functioning mechanism is a data point. A complaint that has no channel becomes a organising grievance, and organising grievances become the campaigns that produce the delays and cancellations in the IDB dataset.
The Energy Charter Result, Read Carefully
Performance against 19 of the 20 Priority Actions improved markedly between the 2024 and 2026 reviews — a genuinely strong result for a voluntary framework. But the same review found that participants' social licence was “being impacted by external factors including other energy companies,” and that “action is needed from governments, regulators and policymakers to build industry social licence.” The lesson for any project team: your own performance is necessary but not sufficient. The sector's licence is a shared resource, and a competitor's failure depletes it.
10. Misinformation and the New Opposition
The opposition landscape has changed in a way that most engagement programmes are not designed for. The 8th annual State of Infrastructure and Engagement survey took a deep dive into misinformation and disinformation and how it affects project delivery, noting that interest from governments, investors, and developers had “skyrocketed” over the previous twelve months.
The scale is significant. By the end of 2025, at least 888 state laws and local ordinances restricting the siting of renewable energy projects had been adopted across 48 US states. In Australia, social licence failures and landholder backlash are delaying, redesigning, or cancelling renewable projects and transmission networks that are technically sound and financially viable.
| Driver | Mechanism | Implication for engagement design |
|---|---|---|
| Misinformation | False claims about health, property value, or environmental impact circulate faster than corrections | Pre-bunking and trusted local messengers outperform reactive rebuttal |
| Project politicisation | Projects become proxies for broader political conflict | Engagement cannot resolve a political dispute; it can only avoid becoming its subject |
| Cumulative impacts | Communities host multiple projects sequentially and experience the aggregate burden | Project-by-project assessment misses the lived experience |
| Place-based approaches | Practitioners know they deliver better social value but struggle within project-by-project regulatory regimes | Policy reform is required, not just better project management |
Survey themes from the 8th State of Infrastructure and Engagement survey, Australia and Aotearoa New Zealand, 2025–2026.
There is a hard truth in the politicisation data that project teams need to absorb. In Australia, almost 80% of infrastructure professionals agree or strongly agree that project politicisation results in unnecessary delays or cancellations. In New Zealand, over half cite politicisation as the main influence on delays or cancellations, with more than $69 billion in projects impacted in a decade. A project team cannot fix politicisation through engagement. What it can do is ensure that the project is not vulnerable to politicisation — which means having the social licence to survive a change of government.
11. A Practical Social Licence Checklist
- Treat the legal licence and the social licence as two separate workstreams with separate evidence. A completed approvals register is not evidence of acceptance. Report them separately to the project board.
- Assess which level you have reached — legitimacy, credibility, or trust. If you are at legitimacy, you have no resilience to a triggering event and should say so explicitly rather than assuming the relationship is fine.
- Move engagement before design freeze, not after. The decisions that determine the community's experience are made in feasibility and concept design. Engagement that begins at the design freeze is compensation negotiation, not consultation.
- Embed community intelligence as an engineering input. Community variables belong in design reviews, logistics planning, contractor selection, and daily execution protocols — not in a parallel stakeholder function.
- Test the distribution, not just the total benefit. Calculate benefits at the scale at which they are experienced. A regional economic benefit does not offset a local burden if the people carrying the burden are not the people receiving the benefit.
- Establish a grievance mechanism before mobilisation. Accessible, transparent, responsive, independent. Track and publish complaint volumes; a rising trend is the earliest warning of a social licence problem.
- For Indigenous-affected projects, sequence FPIC before any activity. The “prior” element cannot be satisfied retroactively. Starting construction before consent converts a consent process into a dispute.
- Measure the drivers, not just the reputation. Use a disaggregated instrument that separates trust, environmental approval, and procedural fairness. An overall score tells you nothing actionable.
- Budget engagement against the cost of delay, not against the communications budget. At $20 million per week of NPV loss on a large project, engagement is not an overhead; it is the cheapest schedule insurance available.
- Recognise that the sector's licence is shared. Your project's acceptance is affected by other companies' behaviour in the same region. Industry-level engagement is therefore part of project risk management, not a separate corporate function.
12. Frequently Asked Questions
What is social licence in construction?
Social licence is the ongoing, tacit acceptance of a project by the communities and stakeholders it affects. Unlike a legal permit, it is not granted by a government authority and cannot be enforced in court. It is earned through trust, transparency, and demonstrated benefit, and it can be withdrawn at any time through community opposition. It exists on a continuum from rejection through acceptance to full identification with the project.
Can a project proceed without social licence if it has all its permits?
Legally, yes. Practically, it is extremely risky. Opposition can manifest as protests, legal challenges, permit appeals, political pressure on regulators, media campaigns, and investor activism — each of which increases cost and delay. The question is not whether a project can proceed without social licence, but whether the economics still work once the cost of opposition is included.
How much does community conflict actually cost a project?
A project with $3–5 billion in capital expenditure can lose roughly $20 million per week in net present value during a conflict-induced delay, with delays of 2–3 years and cost overruns of up to 40% documented. Across 200 Latin American and Caribbean projects, 81% were delayed, 58% overran cost, and 18% were cancelled outright because of conflict. A single documented nine-month delay cost $750 million.
What are the three levels of social licence?
Legitimacy is the baseline: legal compliance and conformance to community norms. Credibility is built through consistent follow-through, transparency, and demonstrated competence. Trust is the highest level, emerging from an ongoing relationship of mutual regard and shared purpose. Projects at legitimacy level are vulnerable to any triggering event; projects at trust level can weather significant setbacks.
Why does engagement often fail to prevent opposition?
Because it usually happens after design lock-in. Engineering teams design site layouts, logistics, and construction sequencing without community intelligence, while community relations teams respond reactively after technical decisions are locked. By the time the community is consulted, the decisions that determine their experience have already been made, so engagement becomes a negotiation about compensation rather than a contribution to design.
What is FPIC, and when does it apply?
Free, Prior and Informed Consent is the standard for projects affecting Indigenous peoples' lands, resources, or rights, established under ILO Convention 169, UNDRIP, IFC Performance Standard 7, and World Bank ESS7. It requires a good-faith process, conducted before activities begin, with full information and without coercion. It does not mandate consensus, but it does require that consent be sought prior — a requirement that cannot be satisfied retroactively.
Can social licence be measured quantitatively?
Yes. A 2026 New Zealand framework used 15 Likert-scale statements to classify social licence into four levels, achieving Cronbach's α above 0.84. Its results were instructive: an overall score of 2.9 (acceptance level), with institutional trust at 2.60 and environmental approval at 2.75 as the weakest dimensions. The value is in the disaggregation — it tells an organisation which driver to invest in, rather than producing a single reputation score.
Does the climate benefit of a renewable project confer social licence?
No. Research is explicit that broader social licence around an issue such as climate action does not trickle down to a local-level social licence. Local communities experience the visual, noise, land-use, and property-value impacts regardless of the project's climate case. Social licence is determined project by project and community by community.
What actually reduces conflict and delay?
The strongest evidence is for community-based monitoring and shared decision-making, which halved project delays and reduced cost overruns by over 10% in the Franks et al. dataset. A 2026 community-integrated delivery model achieved zero community-induced work stoppages over 18 months and a 40% reduction in logistics-related grievances by embedding community intelligence into design reviews and execution protocols rather than running it as a parallel function.
What is the single most common mistake?
Treating social licence as a communications problem. In the IDB dataset, firms took action to address conflict in 86% of cases, but in 91% of those cases the action was limited to press statements. Messaging is the response of an organisation that believes the problem is perception. When the problem is the project's distribution of benefits and burdens, messaging cannot solve it — and the delay continues while the statements are issued.
Related reading on BuildersMeasures: Mega-Project Risk Management: The Risk Is in the Estimate, Not the Execution, Climate-Resilient Infrastructure: A Threshold Problem, Not a Load Problem, and Dam Engineering Basics: A Dam Is a Controlled Failure.