4 ms·

> Today, the tolerance for risk is much lower, yet SpaceX can launch a lot more often and a lot more reliably, and cheaper too. Strange, huh?

btilly 4y ago
Frequency and cost yes, but reliability is more complicated. ULA has 148 launches with NO accidents. SpaceX has put up more, but had an accident in flight with the Falcon 9 in June 2015, and another on the pad in September 2016. Mind you, ULA started with a known rocket design. While SpaceX has been iterating as it goes. But still, we don't have significant statistical evidence about which is actually more reliable.
- panick21_ 4y ago
  As you said, ULA when it started already had mature rockets. You need to compare rocket families not operators. And ULA had a number of issues in that time. For example a failure in an Atlas first stage and the only just managed to deploy the sat because the upper stage had just enough margin. And ULA had major delays, way longer then what SpaceX could afford for things like their Delta 4 Heavy launches. ULA of course also got 1 billion $ in subsidies per year to ensure launch safety, a luxury SpaceX never had. > But still, we don't have significant statistical evidence about which is actually more reliable. Actually we do. It you look at the launch rates and how often Falcon 9 has launched already the statistical significant of the results gets bigger and bigger. Unfortunately I can't find the list that shows this data and explains the methods used right now. But Falcon 9 is very clearly on top of the list.
  - philistine 4y ago
    Every time they do a run-of-the-mill Starlink launch that goes without a hitch, Falcon 9 becomes safer and more valuable. Every time Falcon 9 flies, the necessity for Starship to succeed soon decreases.
    - MichaelCollins 4y ago
      I doubt Starlink is commercially viable in the long run if they don't get Starship operational. The Starlink constellation requires continuous maintenance, replacement of old satellites that have expended their fuel and fall out of LEO a few months later. To do this with Falcon 9 instead of Starship, they need something on the order of 10x more launches, and each time they throw away the second stage. Furthermore the Starlink constellation 'only' has 3,000 satellites so far, but SpaceX has sought approval for 30,000 more than the 12,000 already approved. To launch and maintain a constellation that huge with Falcon 9? It might be commercially viable, but I doubt it, they already have a very rapid launch cadence with Falcon 9 starlink launches and they evidently think they need much more.
  - btilly 4y ago
    No, we really don't have significant statistical evidence. Even for the rocket family. And there is no need to look for the reference you can't find since this is all verifiable from public data. The ULA workhorse was the Atlas 5. Over its whole lifetime, both under Lockheed and ULA, it had 95 launches of which one was a partial failure (declared success). Wikipedia is good enough for that. The SpaceX workhorse is the Falcon 9. https://www.spacex.com/launches/ https://www.spacex.com/launches/ puts it at 176 launches with 1 failure and 1 partial failure. This is not counting one that blew up on the pad with the satellite aboard. The measured failure rate of the Atlas 5 is 0/95. Of Falcon 9 is 1/176. Note that 0/95 < 1/176. The measured "not perfect success" rate of the Atlas 5 is 1/95. Of Falcon 9 is 2/176 = 1/88. Note that 1/95 < 1/88. Also anyone with basic statistical knowledge will immediately see that the sample size is sufficiently low that we cannot draw any conclusion at 95% confidence. Now what Falcon 9 does have the crown on is the longest string of uninterrupted successes of any rocket family. They are now at 148. (Not counting, of course, the one that blew up on the launch pad with the satellite on board.) But not enough to be statistically strong evidence that they are best. Also ULA has over 150 launches across their rocket platforms without any being declared a failure - which exceeds SpaceX's current winning streak. That said, I agree with you about ULA's advantages, and I agree that SpaceX is the future. But I'm not going to lie about the math. And the math says that there isn't significant statistical evidence that SpaceX is better on safety.
    - panick21_ 4y ago
      The statistical measure I was talking about does some more complex calculation and takes into account the different iteration of the rocket as well. In low sample size cases such measures are a better predictor of the future. Insurance rates also reflect this same reality, as SpaceX has industry leading safety rating. If I can find this document again I will send it, but somehow I can't find the damn thing.
Karellen 4y ago
Falcon 9 flight 1 was delayed more than 10 times, for about 6 months, from it's initial NET launch date of Nov 29 2009 until the actual launch on June 4 2010. https://en.wikipedia.org/wiki/Dragon_Spacecraft_Qualification_Unit#Delays https://en.wikipedia.org/wiki/Dragon_Spacecraft_Qualificatio...
- philistine 4y ago
  That’s a great point, because it bellies the core strategic problems with SLS. When your rocket has already flown a lot, and getting it human-rated does not slow your rate of launches, you can afford to take your time not to kill anyone. The same plan is being put in place with Starship, which will fly a lot before they put people in it. SLS is one and done. It’s the old Apollo and Shuttle strategy, with the post-Columbia and Challenger risk management. It’s a bad strategy.
  - Karellen 4y ago
    The first 19 Falcon 9s were "one and done". But even after the first, those next 18 still had a fair few launch delays between them too. Getting your launches right is always a challenge for any new rocket family, and delaying your early test flights when (not if) you see unexpected telemetry data during the countdown should be seen as normal while you're learning how your rocket works in the real world. Even with Starship, where cratering the first few test article hops was considered pretty likely, they had a bunch of holds and scrubs during the countdowns to those launch attempts. I expect the next test flight not to have zero delays either. And SLS is being cautious. Delays should be expected, and not seen as some terrible bit of mismanagement. "Go fever" is real, and is a bad thing. https://en.wikipedia.org/wiki/Go_fever https://en.wikipedia.org/wiki/Go_fever
    - philistine 4y ago
      When I said one-and-done, I meant that they're flying an empty one and the second one already has people in it. That's the losing strategy.
altacc 4y ago
I think you might not be taking a significant numbers of factors into account as these are not comparable programs. Artemis & the SLS have very different requirements to SpaceX’s falcon. The closer comparison is to Starship & Super Heavy, which has also not yet launched. The Artemis program is ahead of SpaceX.
- natch 4y ago
  Compare SLS and SpaceX, not Artemis and SpaceX. SpaceX became a part of Artemis when it was selected to take Astronauts to the moon. SLS will only take them into orbit near the moon. SpaceX will take them to the actual surface of the moon. That’s a kinda important part of Artemis.
  - philistine 4y ago
    Yeah. It’s maddening to imagine that the spaceship to bring them to the moon is smaller than the lander that will put them on the moon.
  - MichaelCollins 4y ago
    That landing ever actually occurring seems about as likely as pigs flying (or pork barrels flying?) NASA selecting the absurd Starship lander proposal seems like a tacit admission of this. The lander contract is a 'safe' contract for SpaceX to win, despite their absurd proposal, since the whole program is very unlikely to ever get to the stage where the lander is actually put to use and sent to the moon.

btilly 4y ago

Frequency and cost yes, but reliability is more complicated. ULA has 148 launches with NO accidents. SpaceX has put up more, but had an accident in flight with the Falcon 9 in June 2015, and another on the pad in September 2016. Mind you, ULA started with a known rocket design. While SpaceX has been iterating as it goes. But still, we don't have significant statistical evidence about which is actually more reliable.

panick21_ 4y ago
As you said, ULA when it started already had mature rockets. You need to compare rocket families not operators. And ULA had a number of issues in that time. For example a failure in an Atlas first stage and the only just managed to deploy the sat because the upper stage had just enough margin. And ULA had major delays, way longer then what SpaceX could afford for things like their Delta 4 Heavy launches. ULA of course also got 1 billion $ in subsidies per year to ensure launch safety, a luxury SpaceX never had. > But still, we don't have significant statistical evidence about which is actually more reliable. Actually we do. It you look at the launch rates and how often Falcon 9 has launched already the statistical significant of the results gets bigger and bigger. Unfortunately I can't find the list that shows this data and explains the methods used right now. But Falcon 9 is very clearly on top of the list.
- philistine 4y ago
  Every time they do a run-of-the-mill Starlink launch that goes without a hitch, Falcon 9 becomes safer and more valuable. Every time Falcon 9 flies, the necessity for Starship to succeed soon decreases.
  - MichaelCollins 4y ago
    I doubt Starlink is commercially viable in the long run if they don't get Starship operational. The Starlink constellation requires continuous maintenance, replacement of old satellites that have expended their fuel and fall out of LEO a few months later. To do this with Falcon 9 instead of Starship, they need something on the order of 10x more launches, and each time they throw away the second stage. Furthermore the Starlink constellation 'only' has 3,000 satellites so far, but SpaceX has sought approval for 30,000 more than the 12,000 already approved. To launch and maintain a constellation that huge with Falcon 9? It might be commercially viable, but I doubt it, they already have a very rapid launch cadence with Falcon 9 starlink launches and they evidently think they need much more.
- btilly 4y ago
  No, we really don't have significant statistical evidence. Even for the rocket family. And there is no need to look for the reference you can't find since this is all verifiable from public data. The ULA workhorse was the Atlas 5. Over its whole lifetime, both under Lockheed and ULA, it had 95 launches of which one was a partial failure (declared success). Wikipedia is good enough for that. The SpaceX workhorse is the Falcon 9. https://www.spacex.com/launches/ https://www.spacex.com/launches/ puts it at 176 launches with 1 failure and 1 partial failure. This is not counting one that blew up on the pad with the satellite aboard. The measured failure rate of the Atlas 5 is 0/95. Of Falcon 9 is 1/176. Note that 0/95 < 1/176. The measured "not perfect success" rate of the Atlas 5 is 1/95. Of Falcon 9 is 2/176 = 1/88. Note that 1/95 < 1/88. Also anyone with basic statistical knowledge will immediately see that the sample size is sufficiently low that we cannot draw any conclusion at 95% confidence. Now what Falcon 9 does have the crown on is the longest string of uninterrupted successes of any rocket family. They are now at 148. (Not counting, of course, the one that blew up on the launch pad with the satellite on board.) But not enough to be statistically strong evidence that they are best. Also ULA has over 150 launches across their rocket platforms without any being declared a failure - which exceeds SpaceX's current winning streak. That said, I agree with you about ULA's advantages, and I agree that SpaceX is the future. But I'm not going to lie about the math. And the math says that there isn't significant statistical evidence that SpaceX is better on safety.
  - panick21_ 4y ago
    The statistical measure I was talking about does some more complex calculation and takes into account the different iteration of the rocket as well. In low sample size cases such measures are a better predictor of the future. Insurance rates also reflect this same reality, as SpaceX has industry leading safety rating. If I can find this document again I will send it, but somehow I can't find the damn thing.

Karellen 4y ago

Falcon 9 flight 1 was delayed more than 10 times, for about 6 months, from it's initial NET launch date of Nov 29 2009 until the actual launch on June 4 2010. https://en.wikipedia.org/wiki/Dragon_Spacecraft_Qualification_Unit#Delays https://en.wikipedia.org/wiki/Dragon_Spacecraft_Qualificatio...

philistine 4y ago
That’s a great point, because it bellies the core strategic problems with SLS. When your rocket has already flown a lot, and getting it human-rated does not slow your rate of launches, you can afford to take your time not to kill anyone. The same plan is being put in place with Starship, which will fly a lot before they put people in it. SLS is one and done. It’s the old Apollo and Shuttle strategy, with the post-Columbia and Challenger risk management. It’s a bad strategy.
- Karellen 4y ago
  The first 19 Falcon 9s were "one and done". But even after the first, those next 18 still had a fair few launch delays between them too. Getting your launches right is always a challenge for any new rocket family, and delaying your early test flights when (not if) you see unexpected telemetry data during the countdown should be seen as normal while you're learning how your rocket works in the real world. Even with Starship, where cratering the first few test article hops was considered pretty likely, they had a bunch of holds and scrubs during the countdowns to those launch attempts. I expect the next test flight not to have zero delays either. And SLS is being cautious. Delays should be expected, and not seen as some terrible bit of mismanagement. "Go fever" is real, and is a bad thing. https://en.wikipedia.org/wiki/Go_fever https://en.wikipedia.org/wiki/Go_fever
  - philistine 4y ago
    When I said one-and-done, I meant that they're flying an empty one and the second one already has people in it. That's the losing strategy.

altacc 4y ago

I think you might not be taking a significant numbers of factors into account as these are not comparable programs. Artemis & the SLS have very different requirements to SpaceX’s falcon. The closer comparison is to Starship & Super Heavy, which has also not yet launched. The Artemis program is ahead of SpaceX.

natch 4y ago
Compare SLS and SpaceX, not Artemis and SpaceX. SpaceX became a part of Artemis when it was selected to take Astronauts to the moon. SLS will only take them into orbit near the moon. SpaceX will take them to the actual surface of the moon. That’s a kinda important part of Artemis.
- philistine 4y ago
  Yeah. It’s maddening to imagine that the spaceship to bring them to the moon is smaller than the lander that will put them on the moon.
- MichaelCollins 4y ago
  That landing ever actually occurring seems about as likely as pigs flying (or pork barrels flying?) NASA selecting the absurd Starship lander proposal seems like a tacit admission of this. The lander contract is a 'safe' contract for SpaceX to win, despite their absurd proposal, since the whole program is very unlikely to ever get to the stage where the lander is actually put to use and sent to the moon.