SpaceX's 5-Step Design Process

How rapid failure and iteration will get Starship to orbit and beyond.

Thanks for subscribing.

Today, we’ll talk about SpaceX’s 5-step design process (and the recent Starship test flight and explosion).

Also this week:

• Who is the next LVMH CEO?

• Taylor Swift vs. FTX

• …and some fire memes (including TED Talks)

---

You’ve probably seen videos of SpaceX rockets taking off and landing. As a reminder, here is a GIF from 2019 of two Falcon rocket boosters landing after flight (they were reused for another launch). Wild.

Reusable rockets are the holy grail of space travel because…well, think about how expensive — and difficult — it would be to fly commercial if each airplane was only good for one trip.

Since 2018, more than 50% of SpaceX launches have been with a reused booster, which greatly reduces the cost of sending cargo and passengers to space. SpaceX is currently on a weekly-ish launch cadence and — over the past few years — has put more than 3,500 satellites into low-earth orbit. This satellite constellation is known as Starlink and provides internet coverage for 1m+ subscribers in more than 50 countries. By the end of the decade, Starlink wants to put 9,000 more satellites in space and the internet business could be worth $10B+ a year.

To reduce launch costs even further, SpaceX is building its next-generation rocket called Starship. And it is the opposite of small. The two-part launch vehicle combines a Super Heavy booster rocket (first stage) with a spacecraft also called Starship (second stage).

When stacked together, Starship is 390 feet tall and can carry 4-5x as much payload as the next biggest SpaceX vehicle (Falcon Heavy). On the latest episode of the All-In Podcast, investor Gavin Baker said that an operational Starship vehicle would reduce the variable cost of sending 100 metric tons to space from $15m down to $2m (an 80%+ cost reduction).

The cost reduction combined with Starship’s size completely changes what can be sent to space, per Pirate Wires:

For example, thousands of satellites can be launched into orbit, continuously, enabling a huge number of new satellite-based applications: hyper-localized weather forecasting, earthquake prediction, better-than-GPS location and navigation data, video communication from anywhere on earth, even air traffic control from space.

Constructing massive space telescopes like the James Webb Space Telescope will be easier and cheaper. The outer planets like Neptune and Pluto will become more accessible to our probes. Space tourism will become an industry; space hotels will actually be a thing. And, like, instead of watching rockets launch from a TV in their classroom, your kids will be sending science experiments up on those rockets, then watching them launch from a TV in their classroom. 

Starship is also the vehicle that Elon Musk and the 12,000-person SpaceX team are building to eventually ferry people to the Moon, Mars and beyond (the vehicle can carry up to 100 astronauts in one trip).

On Thursday, SpaceX conducted its first integrated test flight for Starship at Starbase in Boca Chica, Texas (no humans on board). Prior to the launch, Elon assigned a 50/50 chance that Starship would be able to clear the launchpad. However the test flight went, the SpaceX team needed data to improve future launches.

After flying for under 4 minutes and reaching a maximum altitude of 39km, Starship exploded over the Gulf of Mexico (watch the launch here). Despite the explosion, the test flight achieved two notable milestones:

• Clearing the launchpad: Remember, Starship is the largest and most powerful space launch vehicle ever built. Sending this thing off the ground — let alone into orbit — is a ridiculous undertaking. The propellant capacity for the first stage is 3,400 tonnes. Propellant = fuel = things that blow up. The Starship launch was like a bomb detonation (there were tremors and dust plumes miles from the launch site). A launchpad explosion gone wrong in front of 10m+ viewers could have set Starship backs years with regulators and the public.

• Passing the Stress Test: When sending a rocket into orbit, there is a point in the flight called “max Q”, which is when “the stress on the vehicle caused by thrust of the engines and the resistance of the atmosphere peaks”. Starship passed Max Q (more from The Economist on the pros and cons of Starship’s test flight).

The second stage (Starship spacecraft) was supposed to do a partial trip around the globe and splash down in the Pacific Ocean near Hawaii. But the stage 2 never separated and the vehicle was remotely blown up (to control the explosion).

Of course, completing the entire mission would have been great. but this Starship test flight was hardly a failure. And as much as it was a “failure”, the ability to fail — and iterate quickly — is baked into the SpaceX DNA.

Starship launch (Source: YouTube / SpaceX)

In August 2021, Tim Dodd from the Everyday Astronaut toured Starbase with Elon. During the 2.5-hour conversation, Elon laid out the design and manufacturing process for SpaceX (noting that manufacturing at scale is a much bigger challenge than design).

As Elon explained, the different SpaceX vehicles have different failure thresholds:

SpaceX has polar opposite design methods for Starship and Dragon [which carries crews as well as cargo to the ISS]. Dragon can never fail, and it must be tested in extreme amounts and has tons of margin.

However, to develop the world’s first fully and rapidly re-usable rocket, SpaceX must iterate rapidly, which leads to lots of failures. Falcon is in-between, where SpaceX can afford to have landing failure, but cannot experience a failure during ascent.

We are making rapid iterations. Every single ship and booster has had significant iterations. So, we actually want to push the envelope. Frankly, if you don’t push the envelope, you cannot achieve the goal of a full and rapidly reusable rocket. Yeah, it’s just not possible. You have to go close to the edge on margins, right.”

The SpaceX approach is quite different compared to that of another famous space program: NASA's Space Shuttle (which, sadly, had two fatal accidents). Why? Since the shuttles had people on board, NASA's test flights had to be much more conservative. Even though NASA engineers were aware of the issues with the shuttle, there was an imbalance between the risk and reward of flagging them.

“If you make a change and something went wrong [that’s a big punishment],” Elon says of NASA’s shuttle design process. “But if you make a change and it goes right, you only get a small reward. The biggest problem with the Space Shuttle was that its design froze. Due to all the Space Shuttle missions being crewed, design changes were high risk and low reward. Starship does not have anyone on board so we can blow things up.”

According to AREOFORM, previous NASA leaders have known that taking more frequent calculated risks is important for spacecraft innovation:

During the ‘90s, President Clinton appointed Daniel Goldin to be NASA’s administrator, he is the one who diagnosed and first defined the…paradox of how chasing safety led to greater and greater risks.

[Per Goldin], “There’s a paradox at work here that creates a downward spiral. Launching fewer spacecraft means scientists want to pile every instrument they can onto whatever’s going to fly. That increases the weight, which increases the cost of the spacecraft and the launcher. Fewer spacecraft also means we can’t take any risk with the ones we launch, so we have to have redundancy, which increases weight and cost, and we can’t risk flying new technology, so we don’t end up producing cutting edge technology.”

Rapid iteration is possible based on SpaceX’s 5-part design and manufacturing process. Here is how Elon explained it to Dodd:

1. Make the requirement less dumb: “The requirements are definitely dumb; it does not matter who gave them to you. It’s particularly dangerous when they come from an intelligent person, as you may not question them. Everyone’s wrong. No matter what you are, everyone is wrong some of the time. All designs are wrong, it’s just a matter of how wrong.
   
   Whatever requirement or constraint you have, it must come with a name (not a department). Because you can’t ask the departments, you have to ask a person. The person putting forward the requirement must take responsibility for it. Otherwise you could have a requirement that basically an intern [randomly came up with] 2 years ago.”
   

2. Try to delete part of the process: “If parts are not being added back to the design at least 10% of the time, [it means that] not enough parts are being deleted. The bias tends to be very strongly towards ‘let’s add this part or process step in case we need it.’ But you can basically make an ‘in case’ argument for so many things. And for a rocket that is trying to be the first fully re-usable rocket…you really need to run tight margins. Because if you don’t run tight margins, you will get nothing to orbit.”
   

3. Simplify or optimize: “Simplify and optimize the design. The reason this is the third step and not the first step is because the most common error of a smart engineer is to optimize something that should simply not exist. Why would people do that? Well, everyone’s been trained in high school and college that you gotta answer the question. If you tell a professor ‘your question is dumb’, you will get a bad grade. You have to ask the the question [whether something should exist or you will work on optimizing the thing that should simply not exist].”
   

4. Accelerate cycle time: “You’re moving too slowly, go faster. But don’t go faster until you’ve worked out the other three things first.”
   

5. Automate: “Then the final step is automate it. Now, I have personally made the mistake of going backwards on all five steps multiple times. On the [Tesla Model 3 battery pack manufacturing], I automated, accelerated, simplified then deleted.”

In the aftermath of Starship's test flight, there was a clear divide on Twitter over whether the launch was a success or a failure. The different reactions can be explained by three tweets that the SpaceX Twitter account posted.

The first tweet addressed the mid-flight explosion by saying, "Starship experienced a rapid unscheduled disassembly before stage separation." Many trolls - as there are always trolls - likened the phrase "rapid unscheduled disassembly" to Gwyneth Paltrow's euphemism for when she broke up with Chris Martin and called it a "conscious uncoupling."

Many media outlets grappled with this thread of the story and labeled the launch a failure. I will grant that "rapid unscheduled disassembly" is quite funny. However, the point of the Starship test flight — as well as future test flights — is captured in the next two tweets.

The reaction of SpaceX employees upon seeing Starship clear the launchpad also tells a pretty optimistic story. After a day, the SpaceX team has already identified more than 1000 discrete improvements (a big one: the launchpad was heavily damaged and has to be rebuilt with a water-cooled steel plate…which I have never heard of until this week but assume it is meant to handle the ridiculous heat generated by Heavy Booster’s 33 engines).

Putting Starship aside, SpaceX remains miles ahead of the competition. In 2023, the company will have an 80%+ share of the mass lifted into orbit (which is 4x Russia, China, Europe, Lockheed Martin, and everyone else combined).

Expect another Starship test flight in the coming months. Soon, we will hopefully see the Starship “land”. Well, technically, it can’t land because the vehicle doesn’t have legs like the Falcon rockets (legs would add too much weight). Rather, the launch tower for Starship is equipped with giant mechanical arms called Mechazilla that will “catch” Starship on its return.

There needs to be a lot of failures and iterations before that happens, though. Before both stages are fully reusable. Before Starship is launching daily. Before transoceanic transport. Before orbit. Before the Moon. Before Mars.

It all sounds like science-fiction, but so did “reusable Falcon rockets”. In the meantime, check out this animated GIF of Mechazilla at work. Wild.

PS. An interesting problem that SpaceX has to deal with is that the vehicle will someday have to take off from Mars, too. As a result, creating a launchpad that isn’t too complicated — and can be reconstructed on Mars — is important. Good thread from Dr. Phil Metzger, who is a Planetary Physicist who used to work for NASA:

https://twitter.com/drphiltill/status/1649531875692617728

---

Today’s SatPost is brought to you by Bearly.AI

Why are you seeing this ad?

Because I co-founded an AI-powered research app called Bearly AI. And I really like putting blue buttons in this email. If you press this blue button below, you can try AI-powered tools for reading (instant summaries), writing (ChatGPT) and text-to-image art (literally type some text and get a wild image).

---

Links and Memes

Who will run LVMH? Bernard Arnault is the world’s richest person, with a net worth >$200B (the Arnault family owns ~50% of luxury conglomerate LVMH). The Wall Street Journal has a story on his 5 children and who is likely to take over:

Once a month, Bernard Arnault gathers his children for lunch inside a private dining room at the [Paris] headquarters of LVMH

The meal, which lasts exactly 90 minutes, begins with the French billionaire reading aloud discussion topics he has prepared on his iPad, according to people close to him. Mr. Arnault then goes around the table, asking each of his five adult children for advice.

In public, all five children of Arnault’s children get along and his 47-year old daughter Delphine — newly appointed to run Christian Dior (LVMH’s second-biggest brand after LV) — is seen as a front-runner to lead the entire business. Whether or not an Arnault ends up taking over for Bernard, he has restructured the luxury conglomerate to make sure the family retains control:

In December he transformed Agache, the private holding company that ultimately controls LVMH, into a commandite, a French corporate structure that resembles a limited partnership and allows its controlling shareholder to wield significant power with a relatively small holding.

He also created a second entity, Agache Commandite SAS, that is owned by his five children, each with a 20% stake, according to France’s stock market regulator. The new company is empowered to take over the running of Agache and effectively end Mr. Arnault’s leadership of the company. Major decisions, such as dissolving Agache, require unanimous approval from his children.

Here are some other baller links:

• Taylor Swift: FTX offered Taylor Swift $100 million to endorse the crypto exchange, but she ended up passing. Why? According to The Block, Swift did some due diligence, including asking a question that is concerning the entire crypto industry: "Can you tell me that these are not unregistered securities?"
  
  Over a dozen other celebrities who endorsed FTX - Steph Curry, Larry David, Tom Brady, Shaq - have been named in a $5 billion class action lawsuit. Swift has been a savvy business operator for years, and her father Scott is a Wall Street veteran with a long investment career. Although the Swift family did not sniff out the major fraud, the fact that they asked any questions probably spooked FTX. In conclusion: you cannot fool Tay Tay. YOU CANNOT!

• Pachinko is a pinball-like popular game in Japan. It’s basically a legalized form of gambling that “earns more revenue yearly than all the casinos in Las Vegas, Macau, and Singapore combined, bolsters this view” as detailed in an interesting piece from Hidden Japan.

• Google DeepMind is the new AI unit at the search giant. It’s combination of DeepMind (which Google acquired for $500m+ in 2014) and its internal AI research team, Google Brain. Hacker News has a good discussion on the topic and people are generally pessimistic:

  Google still thinks of AI as a research project, or at best a way to produce better search results. They essentially created the entire current generation of the AI space and then... gave it away, because no one on the product side understood what they had actually built. Handing the reins to the DeepMind team – who have never launched a single product in their history – seems to be a doubling down on that same failed strategy.
  

  Google doesn't need more smart AI researchers, academics or ethicists. They need product managers who understand the underlying technology and can commercialize it. They need pragmatic engineers who can execute, launch and maintain services. That has always been their problem as a company.

• Steven Spielberg is one of the greatest directors ever. And his hardest-working year was 1993, when he juggled between dinosaur special FX (Jurassic Park) and grappling with a holocaust film (Schindler’s List) as detailed in this 17-minute YouTube video.

• BuzzFeed News (BFN) is shutting down. Despite BuzzFeed’s listicle reputation, BFN was known for top-notch reporting and even won a Pulitzer Prize (writers from the organization were regularly poached by the New York Times). As a parent, the one BFN article that has always stood out is a profile of the actor who created the wildly-successful Blippi YouTube character (the performer’s past may or may not include shock videos of him shooting diarrhea at a friend).

---

…and here some fire tweets:

The TED conference was just held in Vancouver. I didn’t go because no one invited me and I didn’t want to shell out $5k+ for tix. On a related note, is there a bigger head fake in corporate life than someone saying “I did a TED Talk” only for you to find out they did TEDx, which is the spin-off that organizers can pay to license the name?

Don’t get me wrong, there are a lot of good TEDx talks — here’s a very funny explainer on how to sound smart while giving a TED Talk — but this falls in the same bucket as people turning “Harvard Extension School” into “Harvard University” for their LinkedIn bio.

Netflix will be shutting down its DVD-mailing business after 25 years. In total, it sent out 5.2 billion DVDs. Revenue from the business line fell from $1.2B in 2012 to $146m in 2022, which is like 0.5% of Netflix total sales (still pretty wild that it’s a 9-figure business). Anyways, here is a solid meme:

---