Hacker News Reader

20% is way too high. Even ignoring the (important!) human factors, that sort of failure rate would raise the expense tremendously all on its own. We are finally seeing a glimpse of true reusable rockets, which are the only way to do things in space at anything like a reasonable cost. A 20% failure rate means your reusable rockets can't be reused very much.

Along similar lines, even if you ignore the fear and grief factor, a 20% fatality rate means you'll never create experienced astronauts.

I also really doubt that putting the risk that high would be particularly useful. Even the worst non-man-rated rockets out there have a failure rate of something like 10%, and those failures generally wouldn't kill the astronauts if their spacecraft had a launch escape system. The Space Shuttle's fatality rate was about 2%, and that was only because it was a vastly over-complex, fragile machine with no realistic abort modes in many phases of flight. No other spacecraft made or planned could fail in the ways the Shuttle did.

A government effort probably will be expensive, but only because they're slow to adapt. It will probably be quite some time before we see a(nother) reusable government rocket.

However, a private effort won't go anywhere, because there's no business case for a moon base, nor even a "dream big" case for an Elon Musk type figure. (Mars is a better target by nearly every measure for the second one.)

The best bet IMO would be a public-private partnership, with governments providing the basic goals and funding, and private companies providing the technology and services.

Space Shuttle was not a major breakthrough -- we were in space and had orbital stations for a long time by then. It was supposed to be a cheaper, more convenient way to do things that we already did.

Soviet space program, Apollo program are better examples and both had ~10-20% fatality rates, at least during the initial phases (for example, Apollo 11 (successful moon landing) was 6th Apollo mission; one out of those 6 (Apollo 1) was fatal).

Ancient seafarers, voyagers, explorers faced more dangers and had higher fatalities than today's office clerks. If we are to keep making new voyages we should accept that there is danger involved. If we are spending $X and could make one step with success probability .9999 or 1000 steps with success probability .8 we should seriously consider both options, not let bureaucrats reject the riskier one out of hand. My 2c.

Apollo 1 was a ground test that was retroactively given a mission name to honor the dead. If you want to count it toward the fatality rate then the denominator needs to include all the other ground tests they did, not just the flights. A more complete view would also include Mercury and Gemini.

The Soviets lost a few people in flight during the early days, but the last one was in 1971. There's no reason to think we should have a similar fatality rate in 2017, with modern materials, processes, and control systems.

I agree in theory that if a 20% fatality rate were somehow required, it would be acceptable for things like this. My argument is not that it's unacceptable, but that it's so high as to be a hindrance rather than a help, and that it's far beyond what's necessary even in a "damn the torpedoes, full speed ahead" type of program.

We don't need airliner-level "one in a billion" safety, but something around 1% seems highly reasonable.

Of the Gemini astronauts, I believe fully one-sixth died in training accidents, mostly flight accidents. https://en.wikipedia.org/wiki/Spaceflight_non-fatal_training...

Didn't the Space Shuttle go 1000 times (yes, thousand times, not thousand percent) over budget in terms of cost of launching 1kg of payload into space?

I don't think so. According to the all-knowing Wikipedia, the lowest cost estimate during development was $635/lb in modern dollars. The actual cost was something like $8,000-$20,000/lb depending on whether you're measuring marginal or total costs.

Thanks for looking it up. I think Wikipedia has the variable costs? If I remember right, the article I got the shocking estimates from amortized the fixed costs of developing the space shuttle over the entire life of the program---and since they did much fewer missions than planned originally, you get the crazy overrun.

Anyway, even your figures are crazy.

Part of the reason SpaceX can be cheaper is not because they are smarter---but because they don't have to source parts from each and every congressional district.

Wikipedia has both, thus the range. But I misstated the high end: if you take all costs into account then it's $27,000/lb. Which is an awful lot, to be sure.

> 20% is way too high. Even ignoring the (important!) human factors, that sort of failure rate would raise the expense tremendously all on its own.

Historically, a 20% mortality rate is not a huge barrier. Power and Plenty discusses an episode in the early European spice trade in which somebody sent several ships to India for pepper. All but one of them were lost, giving mortality much higher than 20%, but the voyage made a hefty overall profit from the single surviving ship.

Rockets are a lot more expensive than sailing ships, and there's no equivalent to pepper on the moon.

> Rockets are a lot more expensive than sailing ships

How did you make this comparison?

Do you think the ship was more or less expensive than the lives of the men crewing it?

It seems fairly obvious to me that a wooden sailing ship, even built by hand with old technology, would be far easier to construct than a moon rocket.

It's hard to compare costs across such a great gulf of time, but consider for example that the Spanish Armada consisted of 130 ships, and was fielded by a country of about 8 million people at the time. If a 16th-century ship cost as much as a moon rocket today, then that would be roughly like the USA building 5,000 moon rockets. Moon rockets probably cost about $1 billion or so (Saturn V was a bit under in current dollars, SLS will be quite a bit over) so that would be about $5 trillion. Less than a decade later, Spain was able to throw another 140 ships at the English, so it wasn't a one-off thing either. And that's ignoring naval or merchant ships that didn't participate in the Armada.

I'd wager the ships were much more expensive than the crew at the time. Human life was quite cheap until recently (and still is in many/most parts of the world). A couple hundred man-years of of skilled laborers versus a couple hundred random nobodies?

OK, I agree that the ships are relevantly cheaper than the moon rockets. On the other hand, there are many organizations -- and individual people -- who could easily afford to lose a billion dollars on a moon rocket today. It was common (in some sense) for a galleon-owning merchant to be completely ruined by his single ship failing to come back. So I don't see that the cost of launching rockets is so high as to prevent the activity.

It seems like all of the work to support the idea "20% mortality is too high" is being done by the observation "there is no pepper on the moon". It may be too high on a cost-benefit basis, but that's because the benefits are very low.

But if you think that the benefits are high, perhaps because an initial high-cost high-mortality phase is the prerequisite to a long-term and profitable low-mortality phase, or even a long-term and profitable high-mortality phase, it seems pretty easy to accept 20% mortality.

The benefits not being that big are a big part of it. Yes, if there was some hugely compelling reason to create a moon base then a 20% loss rate would be fine. But there isn't. There are no resources that would make it worthwhile, and I don't see the "national pride" argument supporting that kind of cost. That's what Apollo was built on, and that got canned after only six landings. And they had the USSR to compete with, so I'd think the "national pride" motivation was much stronger than it would be now.

20% is way too high over long term, but I suspect over short term (the first dozen missions) it will be much higher than this.

Also nobody will be forced to go, I'm pretty sure astronauts are very clever people who know the big risks they are taking

20% over the first dozen missions would be 2-3 lost missions in that first dozen, which would be unprecedented in space flight.

Your last sentence confuses me. I went out of my way not to moralize the risk, and only discuss how it would pose problems in terms of cost and creating experienced people.

> Private companies are more likely to achieve this, especially if they are not hamstrung by regulations.

On what do you base this? If it's too expensive and dangerous for governments, with the ability to leverage overwhelming resources if they chose to and not care a bit whether any revenue were generated, how is it more likely that a severely resource constrained company likely limited by a profit (or at least constrained loss) motive will succeed?

Both profit-motives and a re-election motives can disrupt efforts. But I think being small is significant leverage for achieving new things, and small + technology is compounded leverage.

What makes you think "small + technology" is an advantage in this space?

I think that a small effort must approach things more from a bootstrapping of minimal effective steps. I've found this often incompatible with the design-by-comitee approach of government procurement bureaucracy. A government approach tends to put more importance on not making any mistakes than doing things fast and efficiently.

At a recent government placement I could buy post-it notes with the department budget, but not tissue paper. I could buy an iPad to test my software, but not a new monitor (that's IT!). I could buy an Adobe CC licence, but I couldn't install it or upgrade it without an IT admin. Process can eliminate judgement & autonomy.

> At a recent government placement I could buy post-it notes with the department budget, but not tissue paper. I could buy an iPad to test my software, but not a new monitor (that's IT!). I could buy an Adobe CC licence, but I couldn't install it or upgrade it without an IT admin. Process can eliminate judgement & autonomy.

In fairness, that also describes many large private entities.

On talking to a few folks I know who work for NASA. Government officials in most areas are VERY averse to risk taking. Managing and sometimes working on a mission with a fatal crash is likely the end of the career. Business as usual and not reaching far enough has no career downside.

And safety is expensive -- getting the system from .99 to .999 P(success) leads to a large increase in complexity and makes developed technologies more difficult to reuse -- the systems designed for high P(success) are usually monolithic, tightly integrated, built for a single purpose things, not modular systems of reusable components.

This has nothing to do with resource constraints; this is pre-conditioning / expectations that are not going to change for a large government organization. Unless there is a highly visible, universally accepted national need.

> Private companies are more likely to achieve this, especially if they are not hamstrung by regulations.

Private companies are only interested if there is short-term profit in these missions.

Public companies are only interested in short-term profits.

Technically a private company by an eccentric billionaire (musk? bezos?) could be perfectly happy to burn billions to build a moon base

I think we're talking about > 50 billion dollars. I'm not sure even those guys have the kind of resources you would need.

Not yet.

Good point. public and private funding are complementary to each other so there is a healthier Eco system for technology/social progress. Funding from (some time bureaucratic or socialist) government without concerning profit is necessary part of the system. Let me give you a few examples.

Quite a few many pioneers like Hinton, Bingo ,maybe LeCun(was working with Hinton before moved to NY) are happened to be from Canada. I realized the answer for the mysterious coincidence until I watched the video that Hinton explained why he moved to Toronto: There was not enough funding from US other than Military at the time while there were some funding in Canada that didn't demand much instant result from his research.

The healthcare system in Canada is also better than the states although it has a lot lot lot of problems. I believe socialist Canadians care more about people than efficiency of the system initially but it ended up with only 6 or 7 payers for the cost and less overhead instead of more than 2 thousands of payers along with less regulated but very complicated system.

The hi speed train system in China funded by government totally changed the concepts of distance , time, space , contribute so much to economy beyond calculation. I guess if California government can do the same thing, Silicon Valley would spread to be a bigger area and a lot of developer can afford housing by living 100 miles away from where they work.

BTW, I myself was quite pro small government/free market but became more moderate not because I'm influenced by my follow Canadians but by looking it at "High Order" level of efficiency.

Especially when part of that money comes from government subsidies: http://www.latimes.com/business/la-fi-hy-musk-subsidies-2015...

The only figure in the article that's related to SpaceX is

> On a smaller scale, SpaceX, Musk's rocket company, cut a deal for about $20 million in economic development subsidies from Texas to construct a launch facility there. (Separate from incentives, SpaceX has won more than $5.5 billion in government contracts from NASA and the U.S. Air Force.)

$20m is pocket change as subsidies go. And if you consider the contracts that NASA has with SpaceX as subsidies then you'd have to think that about Boeing and BAE and all the other defense contractors.

Cost-plus procurement being corporate welfare is a valid position, but it's a much bigger issue than SpaceX.

20% is impossible. When dealing with so many factors, thousands of failure points, is boils down to either safe (1%) or nothing (100%) with a very slippery slope between. It is basic algebra. When you multiply lots of factors you must either aim very low or accept near-infinity. There isnt much advantage to midpoints. The individual factors before multiplication would be nearly identical, with slight errors quickly adding up to total unpredictability. The 20% rocket wont be much cheaper than the 1%. You will spend far too much effort bouncing around between 1 and 100 trying to hit the 20.

No doubt savings can be made, but the overall reliability of such a complex system isnt directly and smoothly related to cost.

I wonder whether you'd have regulatory problems if people died on the Moon compared to rockets blowing up in the atmosphere. The latter can easily harm people who haven't consented to potentially being blown up for science. But nobody has the authority to regulate what happens on the Moon, so you should be okay to use the El-Cheapo Inflatable Habitat, right?

Just don't use any humans, and there's no risk of loss of human life.

Regardless of regulations, as engineer you should be comfortable of flying a rocket that you designed/built/engineered. Would you go for that if odds of dying would be as high as 20% (one of five) ?

That statistic is based on a very small samples size. In the span of less than a year the SpaceX failure rate is now about 7%. In another year it will probably be 2% if things go well.

From an engineering and process perspective, early models are more failure prone.

That would depend on other factors. For a test flight, I'd suggest we volunteer a chimp. If the plan was to establish a colony, and it meant my descendants might be some of the first humans to explore the wider galaxy...

Well, let me put it this way: absent clinical immortality, I can't imagine being male, childless, and unwilling to take that chance.

> as engineer you should be comfortable of flying a rocket that you designed/built/engineered

I don't think this is true. Some examples:

- If I were a flight engineer (which I'm not), I would be happy to design a plane that I was confident a pilot could fly within safety levels that were acceptable to me, regardless of whether I felt I could personally fly it as safely.

- I've seen a modern blacksmith write about certain female customers' desire for chest armor molded into the shape of breasts. This is fairly dangerous -- such armor effectively aims a large metal spike at the center of your chest -- and completely unnecessary. He showed a photo, with a comment along the lines of "I made this woman a boob plate [his term] because she asked for it to be made that way. I live in fear that one day she will trip and crack her sternum. But wearing the armor makes her feel sexy, and that matters too."

- If I design something with a 10% failure rate, and someone else is happy to operate it knowing it has a 10% failure rate, and it really does have a 10% failure rate, why is it my place to tell them "I wouldn't operate that myself, and therefore you can't either?"

Even with very expensive risk avoidance these kind of missions can be ridiculously failure prone. Just look at how many Mars missions have been lost. Is there really much difference between a process that produces a 20% failure rate and one with 80%-100% (lost all your data)? This would seem like a difficult balance to strike.

Yes and No.

Both countries are currently in desperate need of sizeable public works programs to keep people usefully occupied. Cost is proxy for human activity, but not necessarily a useful one. If we really are looking at massive near term automation, this is exactly the kind of proposal we can expect to see get off the ground...

I'm pretty sure there is minimal overlap between the set of people you'd want to hire for a lunar base program and the set of people in need of public works programs.

Really? Because there seems to be no shortage of unemployed young, educated people in Europe.

What would you have them do?

Crew? No, you want middle-aged types, preferably past reproductive age, for that. Obvious reasons familiar from decades of discussions about astronaut selection include radiation exposure, elevated risk of casualties and psychology.

Design? No, nobody sane would let inexperienced hands design hardware for human space flight and settlement.

Hardware production? Maybe some of that. A few hundred people, maybe even a few thousand, mostly watching over the CNC machines doing the actual work?

I feel that you are vastly underestimating the amount of, and variety of work required in this kind of endeavour. Apollo annually took ~1% of the GDP of the United States to get off the ground. And most of the people who worked on it were not astronauts or rocket surgeons.

> Both countries are currently in desperate need of sizeable public works programs to keep people usefully occupied.

Not really. The ECB will offset any inflation extra government spending might create. That's what inflation targeting does.

And on the lower side, the ECB will cut rates and do QE if total spending and thus inflation fall short of their target.

If the problem is loss of life and not material, I don't think we should keep with the "recruit the best of the best and send them there". Do it Australia style: select prisoners and launch multiple missions: if 7 out of 10 fail, you still have 3 bases working. Too bad for those who died.