The theft, with the receipt
On June 30, 2021, Elon Musk posted four words, three letters and a pirate flag: “Rapidly Reusable Rockets, R R R.”
Within hours the internet did what the internet does. People went looking for a coin called $RRR, decided it must be a token launch, and traded on it. That is one way to read three letters from a man who builds rockets.
I read it a different way, and I took it. The three letters are his. The principle underneath them is his. What I did was point them at something that is not a rocket, and then argue with the order until it described a house.
Ours is RRR — Reuse › Resale › Recycle — route each recovered material to its most valuable recoverable state.
That is the whole borrowing, stated up front, because the only honest way to take an idea is to say where you got it.
What SpaceX actually worked out
The insight is not that reuse is nice. Everyone thinks reuse is nice. The insight is arithmetic.
On September 29, 2011, at the National Press Club, Musk said SpaceX would attempt a fully and rapidly reusable rocket. That was four years before anything landed. The first Falcon 9 booster came back to a pad in December 2015. The first booster to fly a second time went up on March 30, 2017, carrying SES-10 on a stage that had already flown a cargo run the previous April.
The reason he kept saying it is the sentence he keeps returning to: making life multiplanetary is fundamentally a cost-per-ton problem. Not a physics problem, not an engineering problem in the end — a cost-per-ton problem. And you cannot solve a cost-per-ton problem while destroying the most expensive object in the operation every single time you run it. His analogy is better than mine: it would have been impossible to colonize America with expendable boats.
Expendable boats. Hold onto that.
A house is a cost-per-ton problem too
Here is what demolition does to a building. A crew spends a day turning a structure that a hundred people assembled into a pile, the pile goes into containers, the containers go to a facility, and the family that owned it borrows money to buy new material to rebuild the thing that just left in a truck. Every board in that container was already paid for. Every board leaves anyway.
That is the expendable boat. We are not throwing away a rocket, we are throwing away a house, and we do it about a million times a year in this country without anybody calling it a launch failure.
The tonnage is not a rounding error either. The EPA counted 600.3 million tons of construction and demolition debris in a single year against 292.4 million tons of household trash — more than twice as much, with no bin, no curbside day and almost no public argument attached to it. I wrote about that measurement problem separately in The Waste Stream Twice the Size, and the short version is that the industry’s recovery rate looks respectable only until you ask what the recovered material actually became.
Which is the same question rocketry had to ask. Not “did you get it back?” — “did you get it back in a state worth having?”
The three Rs, in priority order
Most people have met three Rs before, on a poster, in a classroom: reduce, reuse, recycle. That version is a suggestion. It tells you what is virtuous. It does not tell you what to do with the specific piece of Douglas fir in your hands at four in the afternoon with a truck coming.
Ours is a routing instruction, and it is strict.
| Destination | What it means | |
|---|---|---|
| Reuse | Held on site | The material goes back into the rebuild as the thing it already is — the house supplies its own replacement stock. |
| Resale | Builder’s Open House | What the rebuild cannot absorb goes to another builder through the secondary market. |
| Recycle | Processing | What nothing else will take is broken down to feedstock. |
A recovered rafter that becomes a rafter beats one sold as lumber, which beats one ground for fiber. The call is made per piece, and it is made at the moment of recovery, because a decision deferred is usually a grade lost.
Read the ladder downward and you are reading value leaving the material. A rafter is worth the most while it is still shaped like a rafter and rated like a rafter. Cut it down and you are paid for wood. Grind it and you are paid for fiber. Nothing about the tree changed at any point. What changed is how much of its original usefulness survived the way we took it out.
Why the order is the whole thing
The letters are borrowed. The order is the work.
Recycling is where most of the building industry stops, and it is the rung that flatters the statistics most. Crushed concrete counts as diverted. It is also the end of the line for that material — a foundation wall crushed into base course does not become a foundation wall again, or a retaining wall, or anything with structure in it. To a landfill operator it is diverted tonnage. To a road crew it is base course. To the building it came from it is the last thing it will ever be.
So Recycle sits at the bottom on purpose. It is the floor of recovery, not the goal of it. An industry that crushes well and reuses badly is not a circular industry — it is a linear one with a good press release.
This is exactly the distinction SpaceX had to hold. Recovering a booster and scrapping it for aluminum would still have been recovery. It would not have been reusability, and it would not have moved the cost per ton at all. The rocket had to come back in a condition that let it fly again. That is the difference between getting the material back and getting the material back as itself.
The ML Systems method is designed to recover up to 80–90% of a home’s materials — that number is MODELED, a target, not a result — but the recovery percentage is not the interesting figure. RRR is what decides whether that recovery was worth anything. You could hit ninety percent and still have a house full of gravel.
Where the ladder stops
I would rather name the hard case than pretend the ladder is smooth.
Most of a house climbs it without much argument. Lumber, fixtures, windows, doors, trim, hardware, metals — they reuse, or they resell. The material comes off in a condition that still has its shape and its rating, and where it goes next is mostly a question of whether the rebuild can absorb it.
Then you reach the foundation, and the ladder stops working. Concrete resists clean separation in a way framing does not, and it is also the largest single mass in the building. Our own method documentation says so plainly: the foundation is the phase that does not yet fit. Concrete has exactly one widely available destination, and that destination is the bottom rung.
That is a known constraint, not a solved one, and it is the subject of the next piece rather than this one.
What I actually took
Not a technology. Not an engineering approach. Rockets and houses have almost nothing in common mechanically, and I am not going to pretend otherwise or that I am in any league with the people who land boosters on ships.
What I took was three letters, a priority order, and one economic insight: the cheapest material is the one you already own and do not have to make again. That sentence is true of a booster and it is true of a stud. Everything else in this company — the routing, the grading, the marketplace, the order the house comes apart in — is downstream of taking that sentence seriously in an industry that has never had to.
Musk put a pirate flag on it. I am fairly sure that means I can have it, as long as I say out loud where it came from.
The honest part
No ML Systems deconstruction has been performed yet. The company is bootstrapped and pre-revenue. The 80–90% recovery figure is MODELED — a design target from our own math, not a measured result from completed projects — and 51% resale is a soft goal, a direction rather than a metric the system enforces. The method described here is a designed sequence, not a report from a finished job.
The SpaceX facts are public and dated, and I have given the dates so they can be checked: the National Press Club announcement on September 29, 2011; the first booster landing in December 2015; the first reflight of a used booster on March 30, 2017; and the post of June 30, 2021 that gave three letters a pirate flag.
ML Systems LLC · Rhode Island · NAICS 236115 · mlsystemsri.com