Grounded Is Not Broken

Japan's H3 rocket failed because of a part that had broken months earlier, in a factory oven. What that failure, my own dead startup, and a hard month at Blue Origin teach about damage that hides between the layers.

By Peter Plötner. Aerospace engineer and Wayfinder Life Coach. More about Peter →

You can pass every review, hit every milestone, and still be coming apart between the layers. Japan's H3 rocket did exactly that.

Last December, an H3 lifted off from Tanegashima carrying Michibiki 5, a navigation satellite Japan had been waiting for. The launch looked perfect. Then the satellite came loose inside the rocket. By the time the two stages separated, investigators later concluded, it had probably already broken free. It damaged the second stage on its way, and the engine never made its second burn. The mission was lost.

The part that failed was the payload adapter, the ring that holds the satellite to the rocket. It is a sandwich: carbon fiber sheets glued to an aluminum honeycomb core. And here is the finding that stays with me. The adapter did not break in December. It broke months earlier, in the factory. During curing, the oven temperatures drifted past their limits and weakened the glue between the layers. Engineers call the result delamination: the layers quietly let go of each other, on the inside, where no eye can reach. From the outside, the part looked fine. It held its shape. It was mounted on a rocket. It was already broken.

The failure was old before it was visible

That is the uncomfortable pattern, and it is not only a rocket pattern. The damage happens when nobody is looking. It hides where inspections do not reach. And it shows itself at the worst moment, under full load, as a catastrophe that looks sudden and is not.

So when something blows up, in hardware or in a life, I have stopped asking “what went wrong today.” Today is rarely the answer. The better question is: when did this actually break?

Watermelon green

Leaders make this harder without meaning to, because the reviews we run inspect surfaces.

In French project slang there is a name for a certain kind of status report: a watermelon. Green on the outside, red on the inside. Every organization has them. There is a scene in the TV show Silicon Valley where a delay shrinks every time it moves up one level of management, until the CEO hears there is no delay at all. It is funny because every engineer has watched it happen.

It is not a lying problem. Information moving toward power loses its red, because people soften the truth for whoever decides their future. You will not fix that with sharper questions in the review. You fix it the way I wrote about in You Don't Prevent the Failure. You Make It Safe. by making red safe to say out loud. The move is simple to describe and hard to do. When red shows up, become curious instead of angry. Treat it as a challenge the team gets to grow on, not as a problem with a guilty party. People watch very closely what happens to the first messenger. That moment decides the color of the next report.

My own delaminated vehicle

I flew one of these myself.

Years ago I co-built a startup, a search engine that helped non-profits find funding. I believed in the mission completely: support people who do good work. We never found funding ourselves. It took me years to find that irony funny.

Money was not the pressure. I had started a new job by then, so my life was fine on paper. Maybe that was part of the problem: with no forcing function, the project could keep running in the evenings forever. Long after the numbers said what they said, I kept patching and polishing, telling myself the next version would turn it around. I had built it, so it had to fly.

Engineers know the trap by name, the sunk cost fallacy. Naming it does not release you from it. The honest difficulty is that there are two failure modes and they feel the same from the inside: quitting too early, and quitting too late. For a long time I had no way to tell them apart.

What finally ended it was not a feeling. It was finishing the investigation. When I understood the problem all the way down, I found two root causes that no feature could fix. The funding bodies had no incentive to make their processes simple and transparent, so the complexity we were fighting was not a bug in their system; it was load-bearing. And the market had a paradox at its center: the non-profits that had funding did not need us, and the ones that needed us could not pay. Once you see that clearly, there is no version of the product that flies. The flaw was not in our layers. It was in the ground we had built on.

The analysis was the easy half. The hard half was accepting it. There is a strange grief in understanding a problem so well that you prove your own project cannot work. But that acceptance is what an investigation board does: state the findings without blame, close the report, and carry the lessons to the next vehicle. When we finally decided to quit, nothing collapsed. The catastrophe I had braced against for years never came. The learnings and the hours went somewhere new.

I tried other vehicles for the same mission, automating paperwork for non-profits, a few other projects. Then I found coaching, and noticed something none of the other vehicles had. Coaching gives me energy. After two hours of it there is more life in me, not less. If I am honest, I do it for my own joy first, and the supporting comes second. It took me a while to accept that this is not selfish. It is what the right vehicle feels like. (Learning to read signals like that took me months, and I told that story in Tight Chest. Less Tight Chest.)

Six months to the pad

Here is why this essay is not about failure.

On June 12, six months after losing Michibiki 5, the H3 flew again. A new, lighter variant on its first flight, six university satellites on board, second stage exactly where it should be. The path from the failure to that launch was not hope. It was tracing: find the root causes, plural, because real failures rarely travel alone. Fix the processes that let them through. Inspect down to the layer where the damage actually lives. Then fly.

Blue Origin is in the hard middle of that same path. In April, a New Glenn upper stage missed its orbit. In May, while they prepared the return to flight, the next booster exploded during a ground test and damaged their only launch pad. Nobody was hurt. The industry has gallows humor for days like that; engineers call it a rapid unscheduled disassembly. The joke is how the people closest to the fire cope. The serious part is what they do next: trace, fix, rebuild. I am confident that we will see New Glenn fly again, hopefully soon.

I work at a launch site, around rockets and pyrotechnics. News like the explosion in Florida lands differently when your own day is spent near hardware that can do the same thing. It did not make me doubt flying. It did something more useful: it woke me up. At a launch site, years can pass with nothing going wrong. Sometimes decades. That long quiet is exactly when feeling safe becomes dangerous, because the quiet is not proof that the danger left. (That slow drift is the subject of Nothing Went Wrong for Ten Years.) The rule I carry from cave diving applies at work too: never do anything you do not personally judge to be safe, because in the end you are responsible for your own life. The morning after the news, I drove in and spent the day on a rocket, like any other day. Just a little more awake.

If something of yours is on the ground

Maybe a project of yours is grounded right now. A company, a role, a plan that did not survive contact with reality. The questions worth asking are the investigation board's questions, not the courtroom's.

What actually stopped working, and when did it really stop? Probably much earlier than the day you noticed.

What would you change about how you work, so the same crack does not form again?

And the question I avoided for years: do you have to give up the goal itself, or only the way you have been trying to reach it? My goal survived. The search engine did not.

Six months after losing a satellite, the H3 stood on the pad again. Grounded is not broken. Grounded is tracing. And tracing ends with a launch.

Frequently asked questions

What is delamination, and why do inspections miss it?

Composite parts are built from layers glued together. Delamination means the glue lets go inside the part, where surface checks cannot see it. The H3 adapter delaminated in the factory, months before launch, when curing temperatures drifted past their limits; the part looked perfect and was already broken. Lives do the same: the outside holds its shape long after something inside has let go.

How do I tell “this isn't working” from “I am giving up too early”?

Finish the investigation instead of consulting your mood. Trace the problem to its root causes and write them down. If what you find are fixable causes, in your product, your skills, your process, you are probably quitting too early. If what you find are structural causes that none of your work can change, the way a market paradox or a missing incentive cannot be patched from outside, then continuing is not persistence, it is avoidance of a finished report. As a cross-check, watch your energy over weeks: the right work drains you on hard days and still pulls you back.

What is a watermelon KPI, and how do I find mine?

A status that is green outside, red inside. You will not find them with harder questions, because information moving toward power loses its red. You find them by making red safe: greet bad news with curiosity instead of anger, treat the red as a challenge the team grows on, and make sure the first messenger visibly does not regret it.

Is letting go of a project the same as failing?

Only if you confuse the goal with the way you pursue it. My goal was supporting people; the search engine was one way, and it had quietly broken. Letting that way go is what freed the goal to find a way that works. JAXA did not give up on navigation satellites when the adapter failed. They fixed the process and launched again.

What if I am grounded right now?

Then you are in the tracing phase, and tracing is work, not shame. Ask when it actually broke, which is usually long before you noticed. Fix the process, not just the resolve. Blue Origin lost a mission and a booster within six weeks and still plans to fly this year. Being grounded is not the end of flying. It is the work that earns the next launch.

The slow drift that makes feeling safe dangerous is the subject of Nothing Went Wrong for Ten Years, and learning to read the energy signal that tells a living vehicle from a dead one is the work in Tight Chest. Less Tight Chest. If you want a place to start, the Essential Self Diagnostic is fifteen questions that take about sixty seconds, a quick read on which parts of your life are still carrying their load and which have quietly let go.