Anonymous sources: Starship needs a major rebuild after 2 consecutive failures | Hacker News Reader

Anonymous sources: Starship needs a major rebuild after 2 consecutive failures

behindtheblack.com

40 pointsjethronethro24 comments

No surprises here. Consecutive apparently identical failures should obviously be throwing up red flags, but moving fast and breaking things was always about iterating. Admittedly, the line between moving fast and breaking things and throwing money into a flame trench is rather thin, but it's only a problem if they ignore their data. Data sure seems to indicate a new design is needed.

> Admittedly, the line between moving fast and breaking things and throwing money into a flame trench is rather thin, but it's only a problem if they ignore their data.

Not really. "Moving fast and breaking things" is relevant in software development circles, where a crash just means a restart and iterating means someone already on payroll has to spend some time fixing bugs.

This is not the case in real engineering fields. A crash has a far more profound implication on project viability, and a greater impact on cost and project deadline. Worst case scenario, it can cost people's lives.

These tradeoffs are far different in real engineering projects, to the point where in some domains an engineer can lose its ability to practice and even be thrown in jail.

Uh, clearly you don’t know the history of the space race? Or any wartime project ever?

> Uh, clearly you don’t know the history of the space race? Or any wartime project ever?

Comparing the tradeoffs if war with the tradeoffs of running a business is absurd. I can't put to words how silly this comparison is.

How many have died due to SpaceX’s actions again?

And the space race was never a declared (or undeclared, really) war.

If you're doing something that is a) very well understood and b) affects the safety of the public, then this is where engineering tends to be heavily regulated. It's not the definition of 'real' engineering though. And if you're trying to do something that is difficult in that it needs to be heavily optimised around thin margins, then it's very easy to spend way more time and money if you try to make it right the first time around, and you'll probably still fail compared to just building it and learning from the process, even if building it is a slow, expensive, and dangerous process.

> If you're doing something that is a) very well understood and b) affects the safety of the public, then this is where engineering tends to be heavily regulated.

No, not really. Regulation is a risk management tool. You regulate activities with a recognized risk to society in general and stakeholders in particular. Consequently, you have fields where it is obligatory to have paid inspectors to verify safety and quality based on the existing standards and guidelines, and even the bulk of the design process consists in running safety verifications to prove a project is safe and reliable based on standardized procedures.

This type of care for safety and reliability is not observed in software development circles. Faults and failures in general do not have a critical impact on society as a whole. The subject of testing and verifications, unlike in real engineering fields, is somehow a hotly contested debate. The primary implication of a fault or failure is downtime, whose only concerned party is the service operator. The only resources wasted addressing faults and failures are man*hour from people already on payroll. This leads to what would represent, in real engineering fields, critical process failures that could cause a project to fail and even cost lives.

As it is software development, critical failures are acceptable and tolerated. A major crash at worst costs some engineer some time to sort out and recover. Seeing a rocket explode can bankrupt some companies and end peoples careers.

It's mind numbingly stupid to extend software development principles and practices to real engineering fields. The resources spent are not the same, nor are the implications of failure. Stakes are much higher, and naturally so is scrutiny and consequences.

Exceedingly close to zero people are at risk no matter how severe my mistake, because I build things that don’t carry risk.

You are drawing the distinction along the wrong boundary: it's not software-vs-' real'-engineering, it's public-safety-critical vs not. There are in fact software engineers that work in the former, and most non-software engineering actually still works in the latter. And yes, a big, expensive expensive failure can be fatal to a company, but only if they've set themselves up to depend on it.

(I've in general seen this strange sort of reverse-chauvinism from a good number of software engineers, who seem to believe that other engineering fields work in some much more regimented and consistent way to software. From someone who's worked in both, including seeing how the sausage is made on safety-critical automotive and medical devices, trust me, they're not much different: all that regulation barely pulls up the minimum bar a teeny tiny bit, at great expense)

“ The one saving grace is that FAA red tape is clearly no longer an additional obstacle. ”

This is wrong - remember starship tests in 2022 were not FAA limited. This is just hard problem

I've been wondering for a while now if the previous FAA constraints actually helped Starship development because it served as a balance against the worst impulses of SpaceX management.

It will be interesting to see if the lack of governmental oversight ends up hindering the development of Starship.

Starship will cause even more damage to the ozone layer[1] and environmental damage to the surroundings of the launch site.

[1] https://universemagazine.com/en/starships-constant-accidents...

they did a stretch of the mk1 and added a whole bunch of systems, and moved the flaps and the most likely culprit is resonance cracking fuel lines and fittings, or tanks. Resonance or "shake" testing is a mandatory part of certifying high end satelites, and NASA and others have huge test rigs that simulate the vibrations that will be experienced durring launch. This is slow, and the easiest cures are to add mass dampaning, but that is bulky and heavy. lauching and iterating is just as viable. The sci fi look of exploded rocket parts on re entry is almost worth all the trouble. Third times a charm, or it's realy time to grind hard.

The green flames and so-fast-and-high-it-looks-slomo explosions are truly amazing.

This is hilarious. 15 years ago, every company blew up their rockets after one launch. Apparently reusability is so common now, we expect the opposite.

But also, I don't think this is unexpected. It hasn't flown any real missions or delivered any commercial payloads. It look them quite awhile to reuse a Falcon9.

This isn't a problem with reuse. Each starship tested is a new build at this stage.

So, by "rebuild", they mean redesign?

Yes, I think so. But there would also some partially built starships there that might be retrofitted or scrapped, as mentioned in the article.

A failed launch isn't the same thing as a successful expendable launch.

I have no idea what I have -3 votes, but this is also fascinating. It's pretty common knowledge that Saturn 5 never was reused for instance... really just the Space Shuttle was the first "resuable" (term used loosely) spacecraft.

Yeah, I'm not an Elon fan, the guy is incapable of comprehending that he's overextended his abilities, but that has little if anything to do with starship. The thing will blow up, until one day we notice that it hasn't in a while, and then there will be a useful heavy launch platform.

I really wonder how similar project "starship" is to project "FSD" in terms of a typical Elmos dreaming big while pumping hype. Which is, btw, also very similar to what Theranos did. Sooner or later it comes down crashing when it becomes apparent that you have overpromised and delivering really turned out to be impossible. Reality catches up, peoples belief in you fizzles and money dry out. Or, alternatively, anonymous sources start popping up with revealing uncomfortable truths.

Anyway, thunderf00t will have a blast with this article.

Probably a lot less dream than true FSD. SpaceX has operational reusable boosters. People have flown operational 'reusable' spaceplanes with tile based heat shields. Heck Starship itself successfully re-entered and soft-landed in the ocean multiple times. SpaceX has demonstrated the belly flop can work. What they need to do now is put all of these capabilities together and there are clearly some major teething pains with this. But it is certainly no Theranos.

imagine someone intends to make a machine that flies but instead actually makes a machine that jumps and the public goes: "well, they already clearly demonstrated that their machine is airborne, they only need to sort out couple of issues to stay airborne longer"

Reply on news.ycombinator.com