How Rodney Rocha Thinks About Evidence Before High-Risk Decisions

Episode 7 of Headline Podcast was published on October 9, 2025, and features Rodney Rocha, the retired NASA chief engineer who worked on the Columbia debris assessment team. His central argument is uncomfortable for any leader who approves high-risk work with incomplete information: the organization should not wait for a worker or engineer to prove that a situation is unsafe before taking uncertainty seriously.

Rocha’s experience is not a license to freeze operations whenever evidence is incomplete. It is a demand for better decision discipline. The relevant question is whether the evidence supports the decision the organization wants to make, especially when the consequence of being wrong is severe and the people closest to the technical concern cannot easily challenge the hierarchy.

Why the burden of proof changes the risk decision

A risk decision becomes distorted when leaders demand proof that work is unsafe, because the person raising the concern must overcome uncertainty while the proposal to continue receives the benefit of the doubt.

On Headline Podcast, Rodney Rocha described the professional habit he had observed in safety-critical work. “Our profession’s culture is to show it’s safe enough to operate,” he said, explaining that the organization often does not go in the other direction and demand proof that it is unsafe.

That distinction changes who must produce evidence, how much time the review receives, and whether an unresolved technical question is treated as a control gap or as an inconvenience. A proposal to continue can appear ordinary simply because no one has yet translated the uncertainty into a formal decision point.

The practical test is simple. Ask what evidence justifies continuing, not only what evidence would justify stopping. This is particularly important in maintenance, commissioning, engineering change, aviation, mining, chemical processing, and any work where a hidden failure can outpace the organization’s ability to recover.

A point-of-work policy test can help leaders see whether the written control actually gives the field enough authority to challenge a weak decision.

What Episode 7 says about reasonable assumptions

Reasonable assumptions become dangerous when they are repeated as facts, because each handoff removes the uncertainty that made the original concern important.

Rocha’s episode theme includes the reasonable assumptions that masked deep uncertainty before the Columbia loss. An assumption can be useful when it is visible, bounded, and tested. It becomes a hazard when a team relies on it without recording the condition that would prove it wrong.

A project team may assume that a component will behave like a previous component, that a temporary control will remain in place, or that a known defect is too small to matter. None of those statements is automatically irrational. The failure appears when the assumption becomes the basis for authorization while the evidence required to challenge it remains undefined.

Risk officers can counter this pattern with four labels. Mark each critical statement as known, assumed, unknown, or verified. Then record which label is allowed to support the next decision. If an assumption is carrying the decision, name the test that will convert it into evidence or invalidate it.

This approach does not make uncertainty disappear. It keeps uncertainty from hiding inside confident language.

How warnings lose force as they move upward

Warnings weaken during escalation when the original technical concern is replaced by a short conclusion that removes the evidence, uncertainty, and consequence behind it.

A technical specialist may say that a condition is not understood, that the data set is incomplete, or that a failure mode has not been ruled out. A later summary may say that engineering has no objection, that the risk is acceptable, or that the issue is being monitored. The summary sounds calmer, but it may no longer represent the original message.

Rocha linked safety performance to the ability to tell the truth and empower others to do the same. His phrase, “Tell the truth, and empower others to tell the truth too,” is not a motivational slogan. It is an information-control requirement. Leaders cannot manage a risk that has been edited into harmlessness before it reaches them.

Use a short escalation record with five fields. Capture the concern in the originator’s words, list the supporting evidence, state the unresolved uncertainty, identify the potential consequence, and name the person who can decide whether work continues. The record should travel with the decision rather than disappear into a separate technical file.

When the concern changes during escalation, preserve both versions. That comparison may reveal where organizational pressure converted a warning into reassurance.

What high-risk teams should separate before approval

Before approving high-risk work, separate the hazard, the evidence, the assumption, the control, and the decision authority so that one confident sentence cannot conceal five different questions.

The table matters because risk reviews often combine these elements. A procedure is treated as evidence, coordination is treated as authority, and familiarity is treated as verification. Separating them exposes the exact point where the decision still depends on belief.

For supervisors, this can be done in a five-minute pre-job conversation. For engineers, it belongs in design review and change approval. For executives, it belongs in the governance question that follows any serious warning: what would make us change this decision?

A structured safety stand-down gives operations a practical way to create that pause without turning every concern into an indefinite shutdown.

How to make uncertainty visible in the next shift

Make uncertainty visible by assigning one owner, one interim control, one evidence test, and one review time before the next shift or work phase begins.

Many risk registers document the hazard but not the unresolved question. The result is a control plan that looks complete while the decision still rests on an assumption. A better entry says what is uncertain, why it matters, what temporary protection applies, and what evidence will close the gap.

Keep the time horizon short enough to influence work. A review date set for the next quarterly meeting will not control a decision that occurs tomorrow. Use a dated checkpoint such as the next shift handover, the next commissioning hold point, or the next engineering test.

The owner must also have authority. Assigning an EHS adviser to monitor a technical uncertainty does not solve the problem if the adviser cannot pause the work, obtain the missing evidence, or reach the executive who owns the approval.

Use the same record in the field and in the leadership meeting. Consistency prevents the organization from having one version of the risk for operators and another version for decision-makers.

Recommendation

Use Rodney Rocha’s evidence question on the next three high-consequence decisions and require leaders to show why the available evidence is enough for the action they want to authorize.

1. Write the proposed decision in one sentence.
2. List the facts that support it and mark every assumption separately.
3. Name the missing evidence that could change the consequence or control decision.
4. Assign an interim control and a review time before the next work phase.
5. Give the technical concern’s originator access to the person with authority to pause or redesign the work.

Do not use the exercise to reward the most cautious voice or to punish the person who challenges the plan. Use it to make the decision traceable. The goal is not perfect information. The goal is an honest account of why the organization believes the work is safe enough to proceed.

Critical-control verification is the next discipline when the decision depends on a barrier that must work in the field, not only in the approval record.

Conclusion

Rodney Rocha’s Episode 7 contribution changes the starting point for high-risk decisions. Leaders should not ask only whether someone has proved that work is unsafe. They should ask whether the organization has enough evidence, visible assumptions, and accountable authority to justify continuing.

Listen to the full conversation with Rodney Rocha on Headline Podcast, then apply the evidence test to one live decision before the next shift. That is how a difficult story becomes a practical risk-management control.