Hacker News Reader

According to the other article I read about it (https://www.forbes.com/sites/jonathanocallaghan/2019/09/02/s...) they knew about it earlier but Starlink refused to move their satelite.

So, they were playing chicken? And Starlink won?

Probably because Aeolus is a more expensive asset than a StarLink sat, their threshold is also higher.

No.

Jonathan O’Callaghan published his extremely biased and inflammatory clickbait before waiting for SpaceX to present their side of the story (time from ESA tweet to Jonathan’s story was about three hours).

What actually happened is that SpaceX communicated their intention to take no action based on a much earlier estimate of extremely low collision risk. A later update from the US Air Force got lost in the mail filter, so SpaceX never read it otherwise they would have contacted ESA immediately.

Here is Loren Grush with a more mature and less inflammatory take on the topic: https://www.theverge.com/2019/9/3/20847243/spacex-starlink-s...

In time we will get a more thorough understanding of what happened, but the ESA’s fundamental point is emphasised by SpaceX’s experience: having to coordinate satellite operations through email that is handled by humans is fundamentally unworkable, and something needs to be done about automating everything before the megaconstellations take to the sky.

They didn't "refuse" since they were never asked to move their satellite. They chose not to, possibly because the risk did not meet their threshold for avoidance.

There's a bit more analysis over at the SpaceX subreddit [1]. Basically it seems that they are dodging one of the Starlink satellites that's going through deorbit testing. Some speculation over there as to why but likely it just seems the two companies have different risk appetite.

[1] https://www.reddit.com/r/spacex/comments/cypvjv/this_morning...

That subreddit is such a weird place.

It's 100% the best source for information about SpaceX, about future launches, past launches, history, future, everything. If you want the latest information on SpaceX, you cannot find a better place. Well moderated, well maintained, up to date info- they achieve what we all thought the internet would be capable of.

But it's also, imho, a bit too far into the "SpaceX/Elon can do no wrong" echo chamber. Well, some of the folks there are. They want what SpaceX is promising (Starship! Mars colonies!) so very badly. Heck, I do too. But it's not a great place to criticize the company. I've found that people there judge SpaceX on their best intentions and everyone else on their actual results.

That said, if I type "re" into my URL bar, it autocompletes to "reddit.com/r/spacex/" immediately. Because seriously, what a great source of information for space nerds like me.

>should be continuously running simulations that are constantly fed fresh TLEs from tracking stations, that check whether or not anything is going to collide

I'm ready to bet that they do this already, and that collision detection is automated indeed. However given the very high cost of orbital maneuvers (can't refuel a sat, so there's a limited quantity of delta v corrections you can do) there's probably a backlog of possible collisions to be checked and acted on manually. The ops team gets to it when they get to it.

I think there is somewhat of a wild west on orbit. Even in the case of known, active missions the parties in a "potential collision" may have different opinions about actual likelihood of the collision: "collision likely" usually means an estimate that both bodies will occupy the same 10-50m volume, but they may or may not actually collide. Based on this each party may decide whether they want to spend limited fuel on maneuver. Or take their chances. Or wait if the other party blinks first.

Add a ton of sats launched by spooks from multiple governments and agencies (who do not want positions of their sats known and may maneuver more frequently) add a whole new layer to this. Just my 2c -- I am not an expert; current setup may be better managed.

>Half an orbit is mightily late.

It makes perfect sense. If you did the manoeuvre earlier, there's more chance that drift in both satelite's orbits will bring them back closer again. By doing it half an orbit before, they can work from the most precise measurements of the two satelite's relative positions and calculate a minimum burn for an avoidance trajectory.

Also a burn at a given point offsets your relative altitude by the maximum amount at the opposite point in the orbit, half an orbit away. So you'll want to burn half an orbit (or an odd number of half orbits) from the point of intersection.

Yes, I knew how burns affect orbits, I was just surprised that they did that burn at the last possible optimal moment, and not, say, 10 and a half orbit before.

Why is half an orbit late? Admittedly, my knowledge of orbital dynamics mostly comes from kerbal space program but isn't half an orbit exactly when you'll have the most maneuverability relative to the collision point? 1/2+n orbits and you're just spending more time in the alternate orbit without increasing your ability to avoid the collision.

It's late because it's the last moment you can do this cheaply in terms of Δv. I expected them to do it n + 1/2 orbit before potential collision, with n = being at least 1, and more like 10 or 20.