Jouni

That's the reason I'm not reading the regulations, looking up references, and asking for definitions. When things fall outside my area of expertise, reading the original sources will usually make me more confused, not less confused. However, I do have some background in data privacy and data protection issues in academic and non-profit settings in the EU. I understand the general way the data protection people think about these issues. From that perspective, GDPR is not a major change to the way the EU approaches data protection. Most things that are illegal under GDPR were already illegal before it. GDPR simply gives the regulatory authorities more effective tools for enforcing the regulations. If an organization was already following the spirit of the law, they most likely just had to plan and document the ways they process personal data better.

I meant that storing and processing data in the technical sense is different from storing and processing data in the legal sense. Data protection is generally concerned with the long-term retention of data and the legal/significant effects of the decisions made using the data. If you simply process the data for short-term technical reasons and then discard it, you are often not processing it in the legal sense. A quick example: Routers inevitably process personal data. However, unless they resort to extensive logging, they are usually not processing the data in the legal sense. Words have different meanings in different fields. The context matters, and there are always hidden assumptions shared by the people in the field. When you read documents from outside your field of experise, you cannot simply assume that you can understand them, because you know the everyday meanings of the words.

The problem is that you are misinterpreting GDPR. You read it in a way typical to people in technical professions, who focus too much on the specific wordings and too little on the overall intent. Law doesn't work that way. It's not a technical specification but an ambiguous set of guidelines for dealing with the ambiguities of the real world. People spend years in law schools to learn the right way to interpret it. For us non-lawyers without large enough interests to hire a lawyer, it's better to focus on the intent of the law. If you don't want to do business with Europeans, you can collect the minimal amount of data to determine whether a person is European, make the decision, and throw away the data once you no longer need it. Assuming that you have some kind of online business, you probably don't even collect personal data from the GDPR perspective this way.

The EU position is that collecting personal data is inherently wrong. It is only allowed with explicit consent, when it is stricly necessary, or when it serves public interests. Anyone familiar with the 20th century European history should understand where this position comes from. The problem with companies such as Google is that their business models are unrelated to the services they provide. Because the information required for targeted advertising is not strictly necessary for providing web search, they cannot collect that information without consent. Consent is not a contract. It is only valid when it is freely given. If there is pressure to consent (e.g. by refusing access to an unrelated service), the consent is not valid. This is the reason why the EU heavily regulates contracts between unequal parties. Even when both parties have entered into a contract, it might not be based on freely given consent, and hence the terms of the contract might not be valid. Anyway, I'm not really defending the EU position, I'm simply describing it. Foreign cultures often have customs and values that seem alien to outsiders. The outsiders should take these customs and values into account when doing business with these cultures.

GDPR establishes that personal data belongs to the person themself. Those lawsuits claim that the business models of Facebook and Google are fundamentally incompatible with GDPR. The companies take and monetize somebody else's property without permit, which is not that different from piracy. If the courts hold those claims, the companies must either change their business models or withdraw from the EU. The EU assumes that their market is too large and too important for the major companies to ignore. They are probably right. It is generally understood now that it was a mistake for the Western technology companies to withdraw from China, instead of complying with the local regulations.

GDPR is 99% common sense, once you accept the principles behind it: Do not collect data, if you can avoid it. If you do collect data, plan in advance what you are going to do with it, and document this. The data you collect does not belong to you. In order to collect and process data, you need an explicit informed consent or a legitimate reason. Consent can obviously be withdrawn, but some of the data may be retained after withdrawal, if there is a legitimate reason for it. Fighting forum abuse sounds like one.

You are assuming a best-case scenario. No boosters are lost in accidents, every booster can be reused as many times as planned, there are no unforeseen issues that require changes to the boosters, and so on. In a more realistic scenario, SpaceX must maintain Falcon 9 production capacity until they can offer something else to their customers. This requires facilities dedicated to Falcon 9 production and workforce that is prepared to do Falcon 9 work at short notice.

Reducing the marginal costs does not mean that doing the same number of launches/year gets substantially cheaper. It means that doing more launches/year is not substantially more expensive. SpaceX is a company with high added value. They design, manufacture, service, and operate their rockets themselves. They even design and manufacture many of the components. In such companies, most of the costs are fixed. If they run a facility at 25% capacity, they have to pay almost as much as for running it at 100% capacity, because most of the costs are fixed. They can only save substantially by closing down the entire facility. There is potential for both high profit when the sales are good and for high losses if they invest in more capacity than what they need. Subcontractors are the standard way to transform fixed costs into variable costs. Instead of manufacturing Falcon 9 lower stages on their own, they could buy them from a subcontractor. Using subcontractors is expensive and unpredictable, but the subcontractor is now the one who has to take the risks with investments and overcapacity. If SpaceX needs 4x less lower stages than in the previous year, they simply buy 4x less lower stages, and the subcontractor has to deal with the consequences.

It doesn't work that way. SpaceX employs ~7000 people, which means something like $500 million/year. Their fixed costs are around $1 billion/year, which gives them the capacity to do maybe 25 launches/year. If they want to maintain the expertise and the production capacity they have slowly built over the years, they must pay most of that $1 billion/year, even if they don't launch any rockets at all. Marginal costs come into play if they want to expand their capacity. Right now, capacity increases would require hiring a lot of new people and building new facilities, which means tens of millions per launch. What SpaceX hopes is that, with increased reusability, their existing infrastructure would be able to support more launches per year. If their current total costs for 25 launches/year are ~$50 million/launch, reducing the marginal costs to $6 million/launch would reduce their total costs for 100 launches/year to ~$15 million/launch. Unfortunately, nobody wants those additional 75 launches/year at the moment. There are less than 100 launches/year in the world, and SpaceX cannot increase their market share significantly anymore. Many countries want to maintain independent launch capability, barring SpaceX from those launches. Even in the US, NASA and the military will continue using other launch vehicle providers, because they don't want to be stuck with a single company and a single technology. SpaceX hopes that orbital launches will be much more frequent in the future. That may happen, but the growth is very slow at the moment.

SpaceX wants to increase the prices, not to drop them. Their R&D is constrained by the amount of money they can bring in. Because the total number of launches in the world is not growing very quickly, the best way to bring in more money is to increase the price per launch. Lowering the marginal costs can also help, but the effect is limited, because they can't lower the costs below zero.

If they have to compete with other launch vehicle providers, they can charge $62 million for a launch. If the payload is too big for their competitors, they can charge $200 million for the same launch. Besides, marginal costs can be misleading when the market size is fixed and the market share is large. SpaceX already makes ~25% of orbital launches in the world. Unless they fire most of their employees and close down most of their facilities, they can't reduce their real costs per launch significantly. While that might increase short-term profits, it would also cripple their R&D and production.

Marginal costs are not particularly important when the total number of launches is low. SpaceX also needs revenue to cover R&D, facilities, and other investments. If they are competing for payloads that anyone can launch, it will take them much longer to break even.

Falcon Heavy is a real rocket that probably works, at least as far as we can extrapolate from a single data point. The BFR is a hypothetical rocket that may fly one day. If you want to launch something big and ambitious in the late 2020s, you should start designing the payload now. Right you, you can be fairly certain that there will be an operational launch vehicle at least as capable as Falcon Heavy, so you can design the payload around those capabilities. There may also be bigger and more capable launch vehicles, but you probably don't want to bet $1-2 billion on that. If SpaceX manages to get the BFR flying on schedule, we will start seeing payloads designed for it in the early 2030s. Until then, they can use the BFR to launch payloads designed for Falcon Heavy, which nobody else can probably launch. Without Falcon Heavy, the BFR would be in direct competition with ordinary launch vehicles such as Ariane 5/6, Atlas V, Delta IV, and Vulcan, because there would not be bigger payloads in the first 10 years or so. That would reduce the price they can charge for the launches significantly.

SpaceX is a business. If they can charge more and still win the contract, they will try to charge more.

When I use serial staging, 50-60% of launch mass is propellant for the first stage. Because I prefer launching with a low initial TWR, maybe 1.2 to 1.3, and because the sea level Isp of the appropriate engines is slightly below 300 s, the first stage lasts for 120-150 seconds. This produces 2000-2700 m/s. The staging altitude depends greatly on the propellant mass fraction, the intended orbit, and the TWR of the second stage. If I use boosters, the core stage lasts for 3-4 minutes, which is often enough to reach orbit.