Evidence from the record · What it means for leaders
Ocean
Gate
When the standard becomes the obstacle.
Small enough to have no excuse
Every other study in this series examines a large organisation, which invites an easy escape: that these are failures of scale, bureaucracy and distance from the front line.
OceanGate employed a few dozen people. There were no layers in which to lose a signal, no divisions to misfile it between and no committee to defer it to. The chief executive knew everyone and was personally involved in the engineering.
The diagnosis is identical to Carillion’s, at roughly one-thousandth of the headcount. Wave One is not a large-company problem. It is a problem concerning how an organisation treats verification—and it arrives faster at small scale, not slower.
How this study is written
Five people died
The submersible was lost on 18 June 2023 during a descent to the wreck of the Titanic. Those aboard included the company’s chief executive, and the youngest was nineteen.
What follows examines how the organisation handled verification and dissent, because that mechanism transfers to businesses that will never build anything that goes underwater. Their deaths are not material for a leadership framework, and nothing here draws on them.
The record in numbers
The final investigation followed two years of evidence gathering at the highest level convened by the US Coast Guard.
Pages in the Coast Guard’s final report, published in August 2025
Years of investigation at the highest level convened by the service
The year in which the clearest internal warning was raised
Independent certifications of the submersible’s hull
What happened
Five years from the first documented warning to the loss.
A genuinely novel approach
The company sets out to build a deep-diving submersible using a carbon-fibre hull rather than the titanium or steel spheres used by every certified vessel of its kind.
The ambition is real, and the engineering problem is legitimate.
The internal warning—and what followed
A senior member of the marine operations team documents concerns about the hull and testing regime in a written report.
He is dismissed. A legal dispute follows and is settled.
And the external warning
A professional body writes to the company on behalf of a large group of industry figures, warning that its experimental approach could have serious consequences.
The company’s public position is that independent classification slows innovation.
Dives proceed, with incidents
The submersible completes dives to the Titanic. Investigators later find that the company continued using it after a series of incidents compromised the integrity of the hull and other critical components.
The vessel was not properly assessed or inspected after those incidents.
An economy with consequences
Under financial pressure, the company stores the submersible outdoors through a Canadian winter.
The Coast Guard finds that the resulting temperature fluctuations further compromised the vessel’s integrity.
The loss
The submersible implodes roughly ninety minutes into its descent, at a depth of around 3,500 metres, following a catastrophic loss of structural integrity in the carbon-fibre hull.
Two federal reports
The Coast Guard’s Marine Board of Investigation concludes that the loss was preventable.
The National Transportation Safety Board reports separately and reaches compatible conclusions about the engineering.
The finding that should travel furthest
Of everything in the Coast Guard’s report, one phrase applies to almost every organisation reading this.
The record, in the investigators’ words
“Glaring disparities between their written safety protocols and their actual practices.”
US Coast Guard Marine Board of Investigation, Report of Investigation, August 2025. The board described the company’s safety culture and operational practices as critically flawed and concluded that the loss was preventable.
Why this matters beyond submersibles
The documents existed. That was the problem.
OceanGate was not an organisation without procedures. It had written safety protocols. What it lacked was any mechanism ensuring that what was written and what was done remained the same thing.
A procedure nobody checks is worse than no procedure at all, because it produces the appearance of control.
And the second finding, on oversight
Confusion about who was responsible was useful
The report describes a company that used a mixture of intimidation, its scientific framing and its favourable reputation to remain outside regulatory scrutiny. It also exploited genuine confusion about which authority had jurisdiction.
Ambiguity about who is accountable is rarely neutral. It tends to be discovered by the party it benefits—and then maintained.
The standard was reframed, not refused
Nobody at OceanGate argued that the hull was safe when it was not. Something subtler happened.
The company’s stated position was that independent classification was an obstacle to innovation—slow, conservative, designed for established technology and ill-suited to anything genuinely new.
That argument is not absurd. It is made, with some justice, in every industry where standards lag behind practice.
What a refusal looks like
“We are not meeting that standard.”
Visible. Arguable. Correctable.
What a reframe looks like
“That standard is what’s holding this industry back.”
Invisible. Principled. Final.
The first invites a conversation about closing a gap. The second makes anyone who raises the gap sound as though they do not believe in the mission.
This is the mechanism worth carrying away, and it operates in organisations with no physical risk whatsoever. Verification steps—review, testing, sign-off, second opinions and audit—are slow by design. They exist precisely to be inconvenient at the moment you most want to proceed.
Calling a check bureaucracy is how a check gets removed without a decision.
What happens to the person who says it anyway
In 2018, a senior member of the operations team put his concerns about the hull and the testing regime into a written document. He was dismissed.
That is the most important fact in this study, and not because of what happened to him. It is because of what every remaining employee learned that week.
Investigators found a company in which the chief executive was the final authority in all matters, and in which people who did not accept his decisions faced consequences ranging from ridicule to dismissal.
After 2018, nobody needed to be told what raising a concern would cost.
The arithmetic of a single dismissal
One firing calibrates everyone
An organisation of forty people does not need a policy against dissent. It needs one example, once—and the lesson is permanent and requires no maintenance.
The reverse is also true. In a small company, one visible instance of someone being thanked for an unwelcome finding does more for the challenge culture than any written value ever will.
And the missing stop
The hull kept its job after failing at it
The Coast Guard found that the submersible continued in use after incidents had compromised the integrity of its hull and other critical components, without proper assessment or inspection.
There was no threshold at which the vessel came out of service by rule rather than by someone’s judgment. The only person whose judgment counted was also the person with the most to lose from a pause.
What was missing
The Shortboard model describes twelve attributes an organisation builds across three cumulative waves.
Wave One is Lean Dependability—the ability to do what you say you will do, honestly and without drama. Wave Two is New Growth. Wave Three is Perpetual Relevance.
Wave Two capabilities built on a hollow Wave One do not hold, because there is nothing underneath them to take the weight.
Compare this grid with Carillion’s. They are the same. One organisation had forty-three thousand employees and the other had a few dozen.
Dependability
Scrappy
resourcefulness
Finding economical ways to verify that something is sound without removing the verification itself.
Healthy
conflict
Making disagreement safe enough that unwelcome evidence can alter decisions before action is taken.
Ruthless
consolidation
Establishing mandatory stopping points when evidence shows that a critical asset may no longer be dependable.
Distributed
ownership
Separating commercial ambition, technical judgment and operational authority so that no one person controls every answer.
Continuous
reinvention
Repeatedly improving or replacing propositions as evidence reveals new limits and possibilities.
Structural
fluidity
Reshaping teams, responsibilities and capabilities around changing opportunities and emerging evidence.
Strategic
optionality
Preserving multiple testable routes instead of depending entirely upon one approach succeeding.
Platform
thinking
Building reusable capabilities that support multiple products, experiments and routes to growth.
Relevance
Pioneer
sanctuaries
Protecting future-facing ideas without exempting them from evidence, verification or challenge.
The awe-driven
mindset
Meeting uncertainty with curiosity and humility rather than allowing conviction to become certainty.
The pioneer’s
leap
Turning a future possibility into a disciplined commitment before complete certainty is available.
Supply-driven
optionality
Creating possibilities before demand is proven while retaining multiple routes forward.
The uncomfortable adjacency
This looked like Wave Three from inside
A pioneering mission. An unconventional approach. Impatience with established practice. A willingness to be laughed at by incumbents. From within, the story OceanGate told about itself was a Wave Three story—and everyone there believed it.
Wave Three without Wave One is not pioneering. It is the appearance of pioneering, built on a floor that has been removed rather than crossed.
The four Wave One gaps
Each rests on findings from a two-year federal investigation with subpoena power.
Healthy conflict
The dominant failure. Dissent was not merely unwelcome—it was terminal. Investigators describe a chief executive who was the final authority on everything and employees who faced real consequences for not accepting his decisions. After the 2018 dismissal, the organisation had no functioning way to be told it was wrong.
Distributed ownership
One person held the engineering judgment, the commercial pressure and the operational decision—and was also the pilot. No separation existed between the person who wanted to dive and the person who decided whether the vessel was fit to do so. That is not a small-company inevitability. It is a design choice.
Scrappy resourcefulness
Removing verification is not the same as being resourceful. Genuine resourcefulness under constraint means finding cheaper ways to establish that something is sound. Storing the vessel outdoors through a Canadian winter to save money, with the consequences described in the report, was the opposite: an economy made in the one place where it could not be afforded.
Ruthless consolidation
No threshold existed at which the hull came out of service. Incidents that compromised structural integrity did not trigger a mandatory stop because no rule required one—and the judgment rested with the person least able to make it neutrally.
The same result as a construction group of forty-three thousand people, a bank of eighty thousand and a fashion retailer with a supply chain it could not list.
Wave One has nothing to do with size. Small organisations do not lack these attributes less often—they simply reach the consequences sooner.
The company ended; the gap did not
OceanGate suspended operations and has been wound down. There is no turnaround chapter here.
What remains is the regulatory finding—and it is the part most likely to be overlooked.
The Coast Guard concluded that the framework governing submersibles and vessels of novel design was inadequate both domestically and internationally. It also found that the whistleblower protections available to the person who raised concerns in 2018 did not work.
The loss occurs and operations are suspended. The highest level of Coast Guard investigation is convened.
Public hearings take place. Former employees and industry figures give evidence on the record.
The Marine Board reports. It finds the loss preventable and the company’s safety culture critically flawed.
The National Transportation Safety Board reports separately on the engineering failure.
The report also records that, had the chief executive survived, its findings might have supported a criminal referral. He did not, and there is no accountability process left to run.
The structural point
A novel category is an unpoliced one
Any genuinely new thing sits outside the standards written for what came before. That gap is where innovation happens—and it is also where nobody is watching.
If your organisation is doing something new enough that no external body understands it, you have not escaped oversight. You have inherited it—and must supply the scrutiny that would otherwise have come from outside.
Small does not mean safe
A check removed for good reasons is still a check removed.
The comfortable reading of a large-company failure is that it happened because the organisation was too big to see itself. This case removes that comfort.
Forty people, one room and a chief executive who knew every detail—and the identical four gaps.
What small scale changes is the speed. A large organisation can carry a hollow Wave One for a decade or more because its mass absorbs the consequences. A small one finds out within a few years.
Three questions for your own team
Where do our written procedures differ from what we actually do—and who already knows?
What have we dismissed as bureaucracy that was, in fact, a verification step?
Who last disagreed with the most senior person in the room—and what happened to them afterwards?
If you only take one thing
Find out what your last dismissal taught everyone
In a small organisation, culture is set by a handful of visible events—not by documents. Think about the last person who left after a disagreement, and about what the people who stayed concluded from it.
They drew a conclusion. You just weren’t in the conversation where they drew it.
How we know this—and what we don’t
An unusually strong record for an organisation this small.
What makes this strong
A two-year federal investigation was conducted at the highest level convened by the Coast Guard, with subpoena power, public hearings and a 335-page report.
A separate transport-safety investigation was conducted independently and reached compatible conclusions about the engineering.
Former employees and industry figures gave testimony on the record, supported by recovered company documents and dive logs.
What to hold lightly
The central figure did not survive and could not answer the findings. That is a real gap: every account of his reasoning comes from other people.
Investigations conducted after a widely publicised loss operate under enormous pressure to produce a clear account.
The company no longer exists to contest the findings, so nothing here has been tested adversarially.
The honest limit
An outlier is still an outlier
Most small companies with strong founders and impatient cultures do not end like this. The great majority of organisations that cut a verification step get away with it—which is exactly why the practice spreads.
Take this as a clear illustration of a common mechanism, not as a claim that founder-led organisations are inherently dangerous. The mechanism is ordinary. The consequence here was not.
Sources
US Coast Guard Marine Board of Investigation, Report of Investigation into the Loss of the Submersible Titan, August 2025; National Transportation Safety Board final report, October 2025; public hearing testimony and exhibits, 2024; and contemporaneous reporting of the 2018 internal report, dismissal and subsequent legal dispute.
A few dozen people.
The same four things
missing as a company
of forty-three thousand.
Subscribe to Notes from the Water
Our weekly newsletter brings readers the latest insights, studies, and observations relating to institutional judgement and the business design