Hacker News Reader

Comment by bolasanibk | Hacker News Reader

SpaceX has made landing so routine that now failed landings are news.

I wonder if they're deliberate? Like if you're 100% you can land in "normal" circumstances, maybe try perturbing the parameters or input data a bit... to stress-test your algos, as if simulating wind, earthquake, different planet's gravity, different atmosphere, different rocket, engine failure...

Now that these are used to launch their own cargo (no not paid launches) I'm sure every loss hits them hard.

A failure on launch would hit them harder, but a failure landing the empty rocket costs the same as it always did, assuming they've got enough spare capacity to keep up the launch schedule.

The payload is smaller and they're more expensive than single-use.

It's cheaper if they land it multiple times, which they have. I'm not aware of the actual ratio they need to hit, although I'm sure they are.

Absolutely. Musk has openly shared that at spacex, failure is just a chance to learn. Successful landings don't teach you much, so the launch of an old core like this one (4 previous landings) is a good chance to experiment a bit.

>Absolutely.

There is no chance they are deliberately crashing rockets because landing them is too easy. That's lunacy.

Agreed. They do stress test old cores, but they make it clear that it's an intentional test before they do so. They don't shake up the parameters on a whim just because they've had a good stretch.

A failure is a change to learn, but a success is a chance to save 20 million dollars. Musk is philosophical about failures, but it's not a thing he's chasing.

I thought this launch was harder than the previous starlink missions. It's not like going to GTO but starlink payloads are really heavy and they mentioned the drone ship would be further down range due to a different orbit this time.

It will be good to hear what the problem was.

Based on the seas today as seen from the drone-ship camera, I wouldn't be surprised if they're willing to launch in conditions sub-optimal for recovery (but sufficient for launch) because the data they gather from a launch failure is still valuable and hitting a launch window is worth more to them than missing it and recovering stage 1.

Not quite the same as intentionally launching to crash, but it has a notional morphism to the site-reliability-engineering concept of "burning your error budget on a high-reliability system to satisfy other tradeoffs."

Reply on news.ycombinator.com