Hacker News Reader

Comment by bpodgursky | Hacker News Reader

If the SLS blew up during testing, it would set the program back a literal decade (or more).

If the James Webb telescope failed during deploy, it would be a $10-$20 billion loss -- and it would never be re-attempted.

NASA needs to re-learn from SpaceX's ability to iterate on cheap designs and flush out flaws quickly. I say re-learn, because they DID know how to do this correctly during the Gemini / Apollo eras (TONS of rockets blew up before they attempted human spaceflight).

But culturally, something went deeply wrong, perhaps when we reached the Space Shuttle era, and they lost the willingness to ever accept failure or pushing limits -- the same process we tell tell children is critical to ever learning.

View on HN

Safi Bahcall's Loonshots describes the phase-change transition which happens in organisations as they grow in size and value safety over innovation.

https://us.macmillan.com/books/9781250185969

https://www.worldcat.org/title/loonshots-how-to-nurture-the-...

New Books Network interview: https://newbooksnetwork.com/safi-bahcall-loonshots-how-to-nu...

It would be interesting to apply those ideas to American society in general.

Giving things a "red hot go" doesn't exist in the vocabulary of bureaucrats. If you have no personal vested interest in success, the only thing to optimise for is to minimise risks, because failures are rewarded with a dressing down at a minimum, career dead end at worst.

Elon Musk is known for rewarding for success, not punishing for failure, but instead reserving punishment for those that didn't even both trying.

Meanwhile, "not even trying" is the default mode of operation for 90% of public servants I've worked with...

Yea but consider the cost. Just slinging weld on a tank costs time and material. How much money was wasted by not properly manufacturing this tank. I don't understand how he has succeeded with such rube goldberg engineering.

Isn't this the second pressure vessel to fail on pressure testing? Doing it wrong once is OK. Doing it wrong twice means you learned nothing from your first failure.

It’s a pressurized tin can. It’s not a rocket. The entire thing could get nuked from orbit and SpaceX would still lose less money cleaning up the radiation than they would if Starship failed - potentially with people on board.

They’re at a “throw stuff at the wall and see what sticks” stage in the engineering cycle. They’re using (and abusing) processes never meant to be used for spaceflight, so in a lot of ways nobody there really knows how to do things correctly without trying and testing it. Imagine trying to learn a language like Go without being able to access any documentation on it. You would waste a ton of time and utterly fail at simple tasks too.

People have been building pressure vessels since the days of the steam engine and cryogenic pressure vessels since the early 1900s. Robert Goddard's first liquid engine ran on liquid oxygen, that was in 1926. The metallurgy and processes are well understood, an explosion shouldn't have happened, yet Musk managed to have one twice in a row. Fire engineering and management, Space X is beginning to look like Boeing!

Its easy when you don’t have to worry about weight. Just make it thicker and stronger. Steam engines don’t often explode, but they aren’t going to make orbit either.

Actually, German TÜV was instituted after some boilers in the steam age blew up the factories and killed many workers. Overbuilding boilers is expensive.

> Fire engineering and management, Space X is beginning to look like Boeing!

If you don't own any SpaceX stock and aren't planning on flying on one of their spaceships when the time comes, why do you care?

If I had to choose I’d rather believe someone that accomplished rocket reusability with prices ridiculously low compared to all the competitors rather than someone that has no idea about the difference between a steam engine and a rocket and probably never heard about the rocket equation.

That's is the fandom talking. There was no indication that Space X expected failure twice in a row.

Elon Musk presided over a failure in static pressure testing of a steel vessel. No vibrations, no dynamic loading, no fluid flow, just simply pressurizing a steel tank. No exotic materials or difficult techniques like aluminium welding were involved, just garden-variety steel, the best-understood material of all. How do you manage to fail at constructing a piece of equipment as simple as this?

Care to state terms of a bet to see who will succeed first at commercial space transport for some number of humans? I'll have someone write up something and commit to the other side of the bet.

We can discuss terms here.

The Russians are flying commercially into space right now, and Soyuz vehicle has an excellent safety record, > 97 % success, no loss of life. With their recent track record no one should trust Boeing or Space X - Starship, Starline and Dragon all revealed what a shambles their engineering is.

>Yea but consider the cost. Just slinging weld on a tank costs time and material

Sure, but its what, A few 10's of thousands of dollars for steel and welding for the test? If it saves a day of double or triple checking to just ""see if the damned thing blows up "" you have to weigh success odds and lessons learned vs. overhead costs. I wonder what a few days or weeks of overhead costs a company like SpaceX that would have had to be spent to make sure the test succeeded.

As a rule of thumb, calendar time is generally more expensive than equipment and man-hours - even in hard tech.

> Doing it wrong twice means you learned nothing from your first failure.

And yet falcon 1 failed how many times before they got falcon 9 up?

> Doing it wrong twice means you learned nothing from your first failure.

Edison failed thousands of times building an electric light bulb filament.

I think an important angle of "consider the cost" is that materials in R&D are pretty much ALWAYS cheaper than a team and time. For example here if the whole rocket was a 4mm thick 9m X 118m cylinder, the actual stainless steel would be like $700K [1]. While the rough headcount ceiling cost at SpaceX (7000 employees) is around $5M per day. And this way they get actual valuable engineering experience out of it, rather than just arguing in conference rooms like a more conservative company would have done. I think fail-fast prototypes are super valuable and are what good R&D looks like.

[1] https://www.wolframalpha.com/input/?i=%28pi+*+9m+*+4mm+*+118...

If your compiler throws an error twice, it doesn’t mean you are a failure, just that something else went wrong.

Besides all the points other people already made, the money saved by being able to switch to deploying the Starlink constellation via SuperHeavy/Starship (very likely to be cheaper in $ per satellite-to-orbit) instead of F9 earlier makes taking risks and iterating faster even more appealing.

How do you iterate cheaply on a large telescope mirror and associated precision components?

You're right. Iteration comes when you plan on building lots of something. If there was a plan to have 100 JWT type telescopes in orbit, then you build lots of prototypes and narrow down the manufacturing process. When there's only one of something, we see what we have now: years of delays and over-budget because the thing (JWT) absolutely has to work or the entire project is a failure.

> If the SLS blew up during testing, it would set the program back a literal decade (or more).

It just did, a few weeks ago: https://www.nasa.gov/exploration/systems/sls/nasa-engineers-....

Reply on news.ycombinator.com