SpaceX’s Use of Agile Methods
The space tech industry has been stuck in a loop of re-inventing the same designs again and again for the past 40 years. You can see this in Boeing’s version of a “modern” spacecraft — which has the ludicrous name of “Starliner”. It is a mere space capsule that looks just like the command module of the Apollo rocket from the 1960s. Here are both, with the “Starliner” on the left, and the 1960s era Apollo Command Module on the right:
More than fifty years have passed since the Apollo Command Module was designed. Is the “Starliner” the best that we can come up with?
Meanwhile, here is what SpaceX is working on:
NASA and Boeing — the prime contractor for the “Starliner” — want us to think that the “Starliner” is an amazing leap forward, but — sorry — it is essentially a tin can with a life support system. Yes, it has a mission abort system, and yes its heat shield can be reused up to ten times, but it is just a fancy tin can: it can’t go anywhere on its own.
NASA has also contracted to develop another similar tin can, the Orion (at a cost of $6 billion) — that way we will have paid for two more tin cans since the retirement of the Space Shuttle — three actually, since they have paid SpaceX to create one as well.
Better have three, since tin cans are so revolutionary — no telling what might happen! But the aerospace contractors think that things are just as they should be: the gravy train is flowing nicely, creating lots of low-risk machines at enormous cost. No pushing the envelope here.
Seems like we are going backwards: from a tin can to a space shuttle and back to a tin can.
To make the “Starliner” go places, there is the new “Space Launch System” (SLS), which looks just like the Apollo Saturn V rocket from the 1960s. The SLS is on the left, and the Apollo Saturn V rocket is on the right:
Both were designed to have enough payload capacity to send ships to the Moon and to Mars.
Each launch of the SLS will reportedly cost $2 billion dollars — each launch!
It appears to me that the existing government-funded space exploration ecosystem, which includes the long entrenched aerospace contractors, don’t really want to do anything new — they just want to keep reinventing the same old things, to keep the money flowing to them but with minimal risk, because mission risk translates into funding risk, and there is less risk in doing things you have already done.
And now SpaceX has ruined things for them. Now they have to actually try things they have not done before. Suddenly everyone is talking about rockets that can land and be reused; and suddenly there is talk of going back to the Moon, and even going to Mars. Other nations have reignited their own programs, because no one wants to be left behind.
How can a private company do all these things — a private company that was recently a little startup? How did that little startup build its own rockets from scratch, and graduate to doing things that previously only nations could do (or so we were told), and become the tail wagging the dog of the entire aerospace industry? And how does that company, which is no longer so little, do things for a fraction of the cost claimed by the established space tech contractors?
Agile Plus Leadership
The leader of SpaceX, Elon Musk, started his career building software. He has brought ideas from that industry to his companies.
SpaceX is an Agile company. That is one reason why they can do things faster, cheaper, and better. And crucially, they apply Agile ideas thoughtfully — not as a methodology to mindlessly follow, but as a set of ideas to inspire them. But another important reason for SpaceX’s success is that there is a visionary person at the helm who involves himself deeply in every product, while executing a form of leadership that gets the most from people. Let’s look at the Agile dimension first.
Investment in the Design through Delivery Pipeline
SpaceX has invested a great deal of effort in automating the process of designing and validating through simulation, and delivering the machinery that they build through automation. They use traditional computer-aided design (CAD) tools such as CATIA, but they also invested in an end-to-end 3D modeling system from which they can view and simulate entire assemblies and automatically print parts. Importantly, the software is fast, even when handling complex assemblies, so that engineers do not have to wait, which encourages a rapid iterative design approach.
As Elon Musk says, they can “take the concept from your mind, translate that into a 3D object, really intuitively…and be able to make it real just by printing it.”
This is systems thinking — a core principle of DevOps — and it is leadership: Musk is thinking about how to empower his engineers and enable them to get things from design to test as quickly and easily as possible — in other words, how they can reduce cycle time.
Of course, judgment is needed about when to “print”, since parts are often of very expensive materials such as titanium and inconel.
Perfect Is the Enemy of the Good
The ability to go right from design to “print” is extremely important. While design and simulation are extremely important at SpaceX, they do not try to perfect a design before they try it. They design, and they simulate, but they also build and test often. They feel that they learn more by building something and pushing it to failure than they would learn in a hundred simulations.
They call this a “hardware rich” approach. It is what the Agile community calls a “fail fast” approach. Instead of spending months or years on design and then carefully building one perfect prototype, they build many, and they test them all in myriad ways. For example, their new Raptor rocket engine is the most advanced rocket engine in existence, with two pre-combustion chambers — one fuel-rich and the other oxygen-rich. SpaceX has made a sophisticated design work, not by trying to perfect it for years before building anything and getting it just right, but by making their best design and then trying it, pushing it to its limits where it blows up, measuring everything about what happens, and then going back to the drawing board and trying again — and again, and again.
They don’t spend forever trying to get each version perfect, which would be very expensive. Instead, they build what their design says, rapidly and without being too careful. They want to get the thing to the test stand, even if it has imperfections.
Rapid Pipeline from Design to Test
To reduce the cost of making parts whose design is always changing, they use 3D printing. This enables the design to evolve without requiring new dies to be fabricated: they literally just print a new design. This is really important because it enables a “red-green” cycle for development — something that is core to an Agile approach. It allows the design to rapidly evolve. One does not wait for a design to be perfect before trying it: one tries it repeatedly and the outcome of each trial — each test — helps the design to converge on perfection much more rapidly than if one tried to perfect it beforehand through analysis.
One result of this process is that the design is always changing, and so if one asks someone from SpaceX something about the current design, their answer today might be different than it is a month later. This might seem chaotic to an outsider, but it is a natural characteristic of an evolutionary and Agile design process.
For example, SpaceX has completely changed their design for a Mars vehicle. It started with the intention to use their Falcon Heavy vehicle to get a crew capsule to Mars; then they came up with the concept for what they now call the “Starship”. (A ludicrous name, one must admit, but less ludicrous than referring to a capsule as a “Starliner”.) Early designs for the Starship were based on a carbon fiber shell, but that approach was recently changed to use stainless steel, and only months after the change, SpaceX produced its first prototype test vehicle.
When using traditional methods, such a major design change would delay a program by years, because the new design would have to be carefully simulated and validated before building a prototype, which is expected to work perfectly. In contrast, SpaceX does not expect its prototypes to work perfectly — if at all. The Starship prototype that they built was destroyed during testing, and — as SpaceX asserts — that is what the prototype was for: to test the design and push the limits. They have more prototypes under construction now, with design changes that benefit from lessons learned in the prior test.
Reuse to the Limit
The reuse of equipment, methods, and ideas is core to Musk’s approach. Making SpaceX’s vehicles reusable has been a goal of his from early on. As a result, SpaceX is the first to build reusable orbital class rockets, and their Mars vehicle (the Starship) will be reusable. In fact, it is the same vehicle that they will use for lunar missions, and it will replace their current Falcon vehicles for launching satellites, avoiding the need to design and test a completely different vehicle for each kind of mission.
Ideas are reused as well. It is ironic that the recently announced Tesla truck — the “Cybertruck” — is made of thick stainless steel that can resist even a sledgehammer. That sounds like it might be heavy, but the brilliant breakthrough for the design is that the vehicle’s outer skin is also its structure: it does not have a separate internal frame as most trucks to. They call this the vehicle’s “exoskeleton”. Since Musk owns both Tesla and SpaceX, one cannot help but wonder if the idea to use stainless steel for the Cybertruck came from SpaceX’s experience using stainless steel as the body for its Starship vehicle.
SpaceX’s approach to building something new is to break the problem down into a series of smaller problems, and solve them individually. For example, the Starship will need to be able to dock in orbit with another Starship and transfer fuel. To figure out how to do that, SpaceX has signed a contract with NASA to develop the technique. NASA would like to learn how to do that as well, so it is a win-win.
To look at this through an Agile lens, each problem solved is a “minimum marketable feature” — one that can be used and reused for many of its future products.
SpaceX did not invent that approach: NASA used it to get to the Moon. NASA did not start out building a moon rocket. They started with Mercury, to learn how to get to orbit and back reliably. Then they shifted to the Gemini program, through which they perfected the ability to dock in space and do spacewalks. Apollo was next. What SpaceX has done is systematize the approach and use it for all of the systems that they create.
Sometimes when developing a new method, there appears to be more than one way to achieve it. SpaceX sometimes tries both, creating a team for each approach, and seeing how they each work out. The teams are required to “share notes” on a frequent basis: it is not a competition — it is a learning journey.
Leadership
Having leadership is essential for achieving big things. As I explain in my articles Agile at Scale Requires a Special Kind of Leadership and Scrum is Very Confused About Leadership, the Agile community does not have good models for leadership. Leadership is what made SpaceX what it is; it is what made Apple what it is; it is what made Amazon what it is. Whenever you see an organization that has done something truly great, there is great leadership behind it.
It is hard to find out what goes on inside of SpaceX, but I did get to see a video of Elon Musk conducting a discussion with technical staff about a particular technical problem. What struck me was that he walked around the room asking questions: How can we solve that? Will that work? Why or why not?
It reminded me of a Socratic process, whereby the best idea wins. It was intellectual. It was goal driven. It was ego-less.
This might be surprising because there are reports of Musk being volatile. For example, when Tesla was struggling to produce the Model 3, Musk reportedly spent a month at the factory, sleeping and eating there. There are anecdotes, reported in Wired, of Musk shouting at someone and firing them on the spot. That sounded like fear based leadership to me — not inspiring leadership.
People can behave differently in different situations however. I once asked Richard Barrett, author of multiple books on leadership, why it seems that organizations such as Apple, who have leaders such as Steve Jobs with notoriously toxic behavior, seem to be able to thrive. Barrett’s response was that he thinks that those leaders have other qualities that cause the organization to thrive, and that people put up with the toxic behavior because they want to be part of something great.
So, according to Barrett, those organizations that have toxic leaders but succeed anyway do not succeed because of the toxic leadership, but in spite of it, and only when the leader has some other abilities that are more important.
In any case, what Musk seems to be able to do — what I witnessed in the video — was get a group of experts to talk through a complex issue. For a company that is building leading edge things, that is an immensely valuable skill — perhaps the most valuable.
Chief Product Owner
Elon Musk is the CEO of SpaceX, but SpaceX’s description of Musk says that he “oversees the development and manufacturing of advanced rockets and spacecraft for missions to and beyond Earth orbit.”
An important person in the company is Gwynne Shotwell. As president and COO, she runs the company, but she is also very involved with customer relationships — specifically NASA and satellite customers. She is arguably the chief product owner for those customers. But for the company’s long term vision, Musk is the product owner. The vision to go to Mars is his.
Musk describes himself as the chief rocket engineer. He is not a CEO who spends all his time looking at financial projections and market performance numbers. Instead, he is focused on the products and the company’s mission. He claims to spend 80% of his time “doing engineering”.
Yet SpaceX was co-founded with Tom Mueller, a rocket engineer then working for TRW. Musk could have handed off the design of SpaceX’s rockets to Mueller, who was an expert, but instead Musk learned all he could from Mueller. Mueller led the design of SpaceX’s Merlin rocket engine — the engine that powers all of their current vehicles — and also led the design of the Draco thrusters used by the Dragon capsule’s crew escape system. But rather than leaving all that to Mueller, Musk stayed deeply involved with all of the decisions about their product technology, to the point where today he considers himself to be a “rocket engineer” and his actual title is “CEO and chief designer”.
One could not ask for a better Agile Product Owner. Musk has defined the vision for SpaceX — to “make humans a spacefaring species” — but he takes responsibility for the product — at all levels. He does not treat technical issues as outside of his purview. Everything about the product and how it is designed, tested, and produced is of interest to him. He has experts to turn to, but he wants to know everything he can about what the product does, and how it will work, and the entire ecosystem that produces the product, because all of that affects the mission — and the ability to attain his vision.
