Elon Musk went on firing spree over slow satellite broadband progress
SpaceX CEO Elon Musk recently "fired at least seven" managers in order to speed up development and testing of satellites that could provide broadband around the world, Reuters reported today.
SpaceX denied parts of the story, saying that some of those managers left of their own accord and that the firings happened over a longer period of time than Reuters claimed.
[...] Among the fired employees were SpaceX VP of Satellites Rajeev Badyal and top designer Mark Krebs, Reuters wrote. "Rajeev wanted three more iterations of test satellites," Reuters quoted one of its sources as saying. "Elon thinks we can do the job with cheaper and simpler satellites, sooner."
Reuters described a culture clash between Musk and employees hired from Microsoft, "where workers were more accustomed to longer development schedules than Musk's famously short deadlines." Badyal is a former Microsoft employee, while Krebs previously worked for Google."
Apparently, the test satellites work:
"We're using the Tintins to explore that modification," one of the SpaceX employee sources said. "They're happy and healthy and we're talking with them every time they pass a ground station, dozens of times a day."
SpaceX engineers have used the two test satellites to play online video games at SpaceX headquarters in Hawthorne, California and the Redmond office, the source said. "We were streaming 4k YouTube and playing 'Counter-Strike: Global Offensive' from Hawthorne to Redmond in the first week," the person added.
Also at SpaceNews and TechCrunch.
Related: SpaceX Deploys Broadband Test Satellites, Fails to Catch Entire Fairing
FCC Authorizes SpaceX to Provide Broadband Satellite Services
SpaceX Valued at $25 Billion... and More
SpaceX Starlink Satellite Prototypes Include Packed, Flexible Solar Arrays
(Score: 2) by takyon on Friday November 02 2018, @12:14AM (10 children)
Details? Or is that just your "feels"?
They could begin operating their ISP service with a fraction of the planned number of satellites, and just ramp it up over the next 5 years.
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 2) by Arik on Friday November 02 2018, @05:03AM (9 children)
Were they really playing Counterstrike through this system? Sure, I can believe that, but it sucked. And it's always going to suck, until we find a way to beat the universal speed limit of 186k miles/second. LEO satellites like this will be roughly 1000 miles away from the endpoints, minimum. That makes for a mile round trip considerably longer than 2k miles, if we assume the endpoints aren't next to each other and directly under a satellite, but let's just make it simple and call it 2k. At the global speed limit that takes a little over half a second to traverse - and that's an awful lot of milliseconds. AND it's also on top of any processing time added by the satellite itself, or by the additional trip remaining at one end of the journey unless both endpoints are served directly by satellite. You aren't going to enjoy playing that sort of game with that kind of lag and you know it.
So that part, at least, is clearly 'fluffery.' To put it politely.
The key clever idea with this project is to compete *on price.* Because it will certainly never compete on performance - BUT he's absolutely right that there's a lot of data being carried that doesn't need high performance. The watching videos example is better, because that IS something you could realistically do over satellite. Latency is much less important, you start it buffering and accept a small delay to get it started and no big deal. And that's a LOT of data.
But here's the thing - many many people would never be happy with this as their primary link. Because of latency - satellite traditionally hasn't offered the kind of bandwidth that would make his scheme work but I'll give him the benefit of the doubt and assume he'll get that part working. I just don't see him beating the speed of light. So this could work great for most of your traffic, but you'll still need a better connection to handle counterstrike, or video conferencing (as opposed to one-way video downloads) or for VOIP or any other application that's sensitive to latency.
And traditionally, partial solutions like this have a VERY hard time getting established. So it makes sense he's putting a lot of stress on early to market.
If laughter is the best medicine, who are the best doctors?
(Score: 5, Informative) by takyon on Friday November 02 2018, @05:37AM (2 children)
They plan on flying the satellites in two groups: one at 340 km (210 miles) altitude, and another at 1,200 km (750 miles).
https://en.wikipedia.org/wiki/Starlink_(satellite_constellation) [wikipedia.org]
I am willing to give it a shot, especially since I don't do much gaming at all. If total latency ends up being substantially worse than cable/fiber, I'm sure we'll hear many people complain about it.
A bigger issue for them might be keeping the satellites in orbit. At 210 miles, those sats will probably not be long-lived. In the long-term, maybe they will try to use the air-breathing electric thruster trick [soylentnews.org].
This service would also be one of the few ways for most people to give money to SpaceX, seeing as the company does not plan to go public until it is regularly transporting people to Mars. Otherwise, I can only think of the merch on their shop [spacex.com].
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 2) by Arik on Friday November 02 2018, @04:36PM (1 child)
If laughter is the best medicine, who are the best doctors?
(Score: 0) by Anonymous Coward on Friday November 02 2018, @07:54PM
I imagine they won't plan for these satellites to last more than 5 years. SpaceX can put them up relatively cheaply anyway, so replacing every so often shouldn't be a big deal.
And the ISS orbits at between 200-240 miles. I'm sure a company that can land rockets on a barge at sea can work out how to keep their satellites in a similar orbit.
(Score: 2) by rondon on Friday November 02 2018, @03:06PM (5 children)
I must be misunderstanding your math. Light can traverse 2000 miles in 11 thousandths of a second, not 500 thousandths of a second. Can you help me understand what I'm missing?
(Score: 2, Interesting) by Arik on Friday November 02 2018, @04:54PM (4 children)
However, in my defense, I did (somewhat hastily) check that my estimate was in the same ballpark as others before pushing post, and somehow that didn't save me.
From https://en.wikipedia.org/wiki/Starlink_(satellite_constellation) for instance;
"Internet traffic via a geostationary satellite has a minimum theoretical round-trip latency of at least 477 ms (between user and ground gateway), but in practice, current satellites offer latencies of 600 ms or more."
That's actually twice what I figured, but it's 30 times the refigured minimum you and I came to, so someone is still missing something here.
And an added 11ms each way won't make counterstrike completely unplayable but it would still be noticeable.
If laughter is the best medicine, who are the best doctors?
(Score: 2) by takyon on Friday November 02 2018, @05:44PM (3 children)
Obviously, you read my post above, so you know that most of the satellites are only at an altitude of 210 miles. Resulting in a 420 mile round trip (omg blaze it) rather than 2,000 miles. While you may be skeptical of this orbit, this is the plan they have filed with the FCC and they will have to stick to it for now. SpaceX's space debris mitigation plan says that the satellites have a 5-7 year useful lifetime and will be forced to deorbit by around one year after that.
They claim latency may be between 7 and 30 ms. A 22 ms ping would be just fine for gaming. 20-40 ms should work fine, above 50 ms would be more noticeable, and 100-150 ms or above would be bad. 11 ms would be just fantastic.
In conclusion, this is not a traditional geostationary satellite constellation. It's a new beast, and using it for good Internet service and even multiplayer gaming is plausible. SpaceX is well-placed to launch small satellites for its constellation. Call it a very literal example of "vertical integration"
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 1) by Arik on Friday November 02 2018, @07:53PM
Sure but you know as well as I do you won't get that. It's not 11ms total, it's an unavoidable 11ms on top of everything else. The only way you'd get close to that would be if both endpoints are connecting directly to the same satellite, and bringing them down so low makes the odds of that pretty low. Particularly since it means the satellites themselves will be moving in and out of range of endpoints quite swiftly.
I do think it's going to be very interesting to see how it turns out. If the price comes down enough this could be very attractive for some applications.
If laughter is the best medicine, who are the best doctors?
(Score: 2) by Tara Li on Friday November 09 2018, @01:47AM (1 child)
A satellite in a 210 mile orbit is only 210 miles from the transmitter/receiver when it is directly over-head, something that is a fairly rare instantaneous event. Figure in the cross-connection while in orbit, and yeah, a thousand mile trip from ground station to ground station (ignoring the ground station to server segment of the interaction - will Starlink provide ground station to any major data center, or will they try to convince the data centers to buy those connections?)
(Score: 2) by takyon on Friday November 09 2018, @02:15AM
https://en.wikipedia.org/wiki/Starlink_(satellite_constellation) [wikipedia.org]
They plan to have 7,518 satellites in the lower orbit. So while it will be rare for one to be directly overhead, they could be "nearby" often.
https://www.teslarati.com/spacex-starlink-internet-constellation-a-license-to-print-money/ [teslarati.com]
Apparently the optimists believe that Starlink at the HIGHER orbit will beat existing networks.
I can't answer your other questions yet. I'm just saying that when the service becomes available, I will look at the reviews. If real users indicate that the hype is to be believed, I will ditch whatever ISPs I'm currently using and throw money at Starlink. I think I would pay $70-100 a month for it.
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]