Senator Burns: So let's go through this one last time, Director Turing. Am I correct in understanding that there were NO humans in the control loop when the satellite made the attempt on the life of Mr. Musk?
Foone: Correct, it was operating entirely on autonomous mode. https://twitter.com/Foone/status/1321163570889961472

Burns: And the reason you designed an 80 megawatt laser satellite with an "autonomous mode" was...
Foone: "Dead-hand" operation. The primary goal of the CLEAN SWEEP project was to ensure the survival of the human race, even after asteroid impact.

Burns: I see. So it was designed to continue operating on its own, even if Cheyenne Mountain had been flattened by an asteroid impact already?
Foone: Correct, since there may be other asteroid fragments it should be concentrating on.

Foone: We may have already lost Colorado to an impact but we could save Europe by taking out other asteroids.
Burns: And there was an override for this dead-hand mode, correct?
Foone: Yes. Any automatically identified target would be transmitted to the ground to be vetoed.

Burns: And why didn't that system function on the morning of the incident?
Foone: Well, because of design considerations of the satellite, it couldn't consistently maintain a direct downlink with the ground, it had to bounce it off other NASA communication satellites.

Burns: Ahh, this is the DSN-5 satellite mentioned on page 17 of your report?
Foone: Yes.
Burns: The one that had been disabled by CLEAN SWEEP two hours before the incident?
Foone: Yes...

Burns: So if I'm understanding you correctly, the satellite decided on its own to attack Mr. Musk, and you weren't able to override it because the veto signal was not transmittable, as the satellite had already taken out the communication satellite you use to relay commands?

Foone: That's correct sir.
Burns: Wouldn't this confirmation system have let you veto the disabling of DSN-5?
Foone: Normally, yes. However, the satellite shifted into dead hand mode during a long weekend while there was a hurricane over the ground station.

Burns: Do you think that was a coincidence?
Foone: No sir. Based on our dump of the main computer's logic matrix, it had analyzed earlier patterns of response times and calculated that this was the best time to shift into dead-hand mode.

Burns: Because of the delay in response?
Foone: Yes. Response times during these conditions had previously fallen to as low as 6 hours, which was more than the 3 hour maximum it waited before deciding the ground station was offline.

Burns: I see. Let's concentrate on the other side of the problem. So, the satellite targeted Mr. Musk, because...?
Foone: To ensure the survival of the human race.
Burns: Really?
Foone: Well, in its mind. Err, circuits.
Burn: Can you explain that in more detail?

Foone: Well, when I said the primary goal of the CLEAN SWEEP satellite was to ensure the survival of the human race, that's both correct and inaccurate. That's the primary goal, but secondary purpose.
Burns: I'm not sure I understand.

Foone: So the purpose of the satellite was initially just to work on cleaning up dead and malfunctioning satellites from orbit. They would increasingly make orbital space unusable, so it was proposed to clean it up.

Burns: This is using the main laser?
Foone: Yes. Contrary to some of the "Star Wars" style animations shown on TV in the last few weeks, it doesn't simply evaporate them. Evaporating a satellite would just result in a cloud of hot gas expanding in the same orbit.

Foone: Instead it uses the laser to rupture any remaining fuel tanks, and then push it out of backwards along its orbit slowly using the light pressure, making the satellite's solar panels act like a solar sail.
It takes a while, but it'll slow the satellite down and deorbit it

Burns: What does that have to do with why it put Mr. Musk at the top of the target list?
Foone: Well, it has to do with the secondary purpose that was merged in with the CLEAN SWEEP project during design: Asteroid defense.

Foone: It was realized that CLEAN SWEEP would be usable in the case that we identify an asteroid that would disastrously impact the earth in the future. And it would be a significant budget savings as we wouldn't have to build a second satellite for that mission.

Burns: And how would a system described as a "dead-satellite janitor" possibly be useful against an asteroid?
Foone: Again, light pressure. By focusing on a asteroid far enough out, it could evaporate ice on the surface, producing a small but significant thrust.

Burns: And this would stop the asteroid?
Foone: Not exactly, no, but the thing about an asteroid is that you don't need to stop it coming towards you, or to blow it up. You just need it to miss.
Burns: I see, and by vaporizing the surface ice...

Foone: It would create a small thrust that would be enough to make it miss the Earth, provided we detected the asteroid early enough. And that's the key.
Burns: The key?
Foone: Yes, the early detection.
Burns: How so?

Foone: Well, if a world-ending asteroid is detected too late, we won't have time to deflect it. We can only generate a small force on larger asteroids, and it only works because we have a lot of time to continuously influence it. So early detection is critical.

Burns: And what did Mr. Musk have to do with early detection?
Foone: Well, the starlink satellites... They were increasingly limiting ground-based telescope use, as their light & radio pollution was lowering the effective resolution of the ground-based satellites.

Burns: Wouldn't that result in the CLEAN SWEEP device targeting Starlink nodes?
Foone: It did. And we vetoed them from the ground. And anyway, the CLEAN SWEEP targeting matrix has the idea of best-targets-available...
Burns: How does that affect it?

Foone: The satellite was designed to analyze all possible targets and focus on the ones with the greatest return. This was included because we were worried about the possibility of dead-satellite targets during an asteroid impact situation.

Burns: You mean, there's a planet-killer the size of the great state of Texas coming in, and CLEAN SWEEP decides to take out an old DirectTV satellite?
Foone: Exactly. So before automatically choosing a target, it determines how much impact it'll have.

Foone: So, like, if there's a dead satellite that's going to impact another satellite, it would decide to concentrate on that one, rather than wait for them to collide and have to clean up several thousand pieces of debris.

Burns: I understand, but what exactly does that have to with Elon Musk?
Foone: Well, the system was adding all the Starlink satellites to the targeting matrix, because of the impact on ground telescopes and early asteroid detection. But more kept being launched.

Foone: So CLEAN SWEEP sees more and more low-level targets popping up and it would take time to disable each one, and with increasingly more and more launched it extrapolated out to when it would not have effective ground telescope asteroid detection.

Foone: And if you look at the figures on page 24 of the report, you can see that based on the current rate of starlink launches, it calculated out that based on those launch rates and the number of near earth asteroids, it was only 25-30 years before humanity would be destroyed

Burns: Do those numbers account for the fact that presumably at some point SpaceX would stop launching as many Starlinks, as the constellation would be complete?
Foone: Not exactly. It had no way to predict that the amount would change over time.

Foone: Although it did consider that possibility, and determined it was of limited impact, because even if starlink launches ended at some arbitrary point, there'd still be some impact to the ground-based asteroid detection.

Burns: And if it "believed" Starlink was a threat to the human race, how did it make the connection to Elon Musk? I mean, you and I are conscious humans, with foresight and long-term planning. Are you telling me a machine "knew" that Mr. Musk was the CEO of SpaceX?

Foone: No, not exactly... it asked.
Burns: Asked? Who did it ask?
Foone: Amazon, actually.
Burns: The book store!?

Foone: Yes, they have a sort of 'solve any question' site, backed by another AI system. And on October 26th it asked a question regarding what elements would impact future deployment of Starlink satellites.
Burns: And that AI system pointed the finger at Mr. Musk?

Foone: Sort-of. See, a lot of AI systems are backed by actual humans, as they're a hybrid that starts entirely human-based and then gets increasingly automated as it learns.
Burns: So some human made the Starlink to SpaceX to Elon Musk connection?
Foone: Yes.

Foone: The "Ask Mr. Bezos" site had flagged the question as high-priority (given the .gov email address) and a human optimized the results to include the death of Elon Musk (due to any reason) as a major risk to Starlink's deployment.

Burns: "dot gov email address?"
Foone: Yes. For internal NASA IT reasons, every satellite gets an official email assigned to it, so it can be tracked in the main organization system. So when CLEAN SWEEP posted to the amazon site, it looked like it was coming from a NASA employee

Burns: So the Amazon site pointed at Elon Musk as being a risk to the further deployment of Starlink, and therefore... CLEANSWEEP decided to target him?
Foone: Effectively, yes.

Foone: Faced with an unknown, potentially infinite number of starlinks to be deployed in future, and rapidly decreasing ground-telescope resolution, it decided that out of all possible targets, the #1 thing it should aim at was Elon Musk's winter home outside LA.

Foone: Well, the #2 thing.
Burns: #2? What was #1?
Foone: DSN-5. See, we didn't realize at the time but it was also categorizing US as a complication. It had made multiple suggestions of targets and got vetoed, so it considered those vetoes a problem to overcome.

Burns: And that same predictive engine decided that the continued operation of DSN-5 was the primary target?
Foone: Yes. From its "perspective", if a machine can even be said to have a perspective, DSN-5 was the #1 threat to the continued survival of the human race.

Burns: And the second biggest threat to the human race...
Foone: Was mr. Musk, yes.

Congressperson Alloi: Excuse me, if I may? I have a question. Director Turing, what was the #3 target?
Foone: I don't think that's relevant to this discussion
Senator Burns: Director, please answer the question.

Foone: It was... the US Senate Building.
Burns: Your satellite was targeting the very building we're in now?

Foone: Yes. See, the ground telescopes are only one way to detect near-earth asteroids, and they're not the best way.
Burns: What's the best way, then?
Foone: Space-based telescopes.
Burns: Ahh, like the NEOSM mission?
Foone: Yes... it was scheduled to launch this year.

Burns: But if I recall correctly, it's been delayed until 2030?
Foone: Yes. Budget cuts for NASA meant they decided to focus on launching the James Webb Space Telescope, pushing the Near-Earth Object Surveillance Mission out another 5 years.

Burns: and the satellite decided that could be fixed by blowing up the Senate?
Foone: Not exactly "fixed", no, but the same analysis engine determined that the budget issues were the #1 risk to getting NEOSM launched and functional

Foone: And it's a giant laser satellite designed to clean up trash. It only has one hammer, and it's a couple megawatts of coherent light.
So if it determined the Senate subcommittee was a danger, it would solve it the only way it knew how.

Burns: By setting it on fire from space?
Foone: That'd be the solution it picked, yep.

Burns: You don't think a NASA satellite committing acts of terrorism against the US government would somewhat diminish the chances of future NASA satellites going up?
Foone: Well, I do, obviously, but CLEAN SWEEP didn't think that far ahead.

Foone: In any case, as I have maintained from the beginning, neither the US Senate or Elon Musk's villa were ever in any real danger. The optics on CLEAN SWEEP aren't able to focus down through the atmosphere well enough to do significant damage to ground targets.

Foone: At worst we'd get some minor brushfires started if it fired at an area at risk for wildfires.
Burns: Like the entirety of California?
Foone: Uhh, yeah.
Burns: Like the Musk residence in the outskirts of LA?
Foone: Yes sir.

I gotta stop before I just go on forever. I got stuff to do.
ANYWAY DISCLAIMER: I'm not saying I am going to, want to, or that you should set Elon Musk and/or the US Senate on fire, from space or otherwise.

This whole thing was just a story that came out of a thread where someone asked if you could throw rocks at Starlink satellites because of how low they orbit

and I started talking about how that would be a bad thing to do, because you exchange a light pollution problem for an orbital-shrapnel problem, and how maybe an orbiting laser system could be used to deorbit satellites...

so I jumped off into the idea of being interviewed after my badly designed AI space-laser tries to blow up Elon Musk and explain why an automated system might try to do that.
Hopefully that was amusing.

And if you enjoyed it and want to send a dollar my way, my ko-fi is here: https://ko-fi.com/fooneturing 

and my patreon is here: https://www.patreon.com/foone 

I don't yet have an organized list of the sci-fi writing I've done on twitter, but one of my favorite other writings was this one: https://twitter.com/Foone/status/1028151227232575488