Starlink or global inexpensive low latency internet

[SavingWithBabies]

I posted on my blog about Starlink a while ago. Today, a big long post about it made the rounds (Hacker News, etc):

https://caseyhandmer.wordpress.com/2019 ... -big-deal/

I'm still reading through the above blog entry but curious if others here have read it or are interested in it and what they think of the post.

Do you hold Alphabet stock in order to invest in SpaceX? That apparently is the only way to invest in SpaceX at least as of August 2017 as Alphabet holds 7.5% of SpaceX stock.

[Cheepnis]

Thanks for posting that link. I've made my way through quite a few of Casey's posts now. What a great and interesting blog!

I don't have many intelligent thoughts on Starlink other than the scope and extent of Musk's ambition is astounding. The space nerd in me definitely wants it all to work out. I'd love for there to be crewed missions to mars within my lifetime for no other reason that it'd be cool af.

[Seppia]

Musk being involved, I am expecting most of the claims to be false.
One day we will look back and think "how did we not see this coming?", similar to WeWork and Uber.

Those who were actually paying attention DID see it coming

[SavingWithBabies]

My knee jerk reaction is not to trust or believe in Musk at all. However, I have a hard time reconciling that with the fact that Tesla is actually shipping cars in quantity and is therefor a viable car company (as far as I can tell, happy to be convinced I'm wrong).

I watched the teardown report summary video that I think was posted here (cliff notes: fit and finish off sometimes, overall minor quality issues, lower half of car oddly done but upper half/electronics way beyond any other maker), the regular reports about fit and finish problems (that seem to have gone down), the autopilot issues, etc. But every single car maker has huge quality problems on a regular basis.

Then there is of course the financing side of it and government subsidies for electric vehicles (ie what if that was gone? would Tesla survive)?

But even with all of that, they are shipping cars and people are buying them. The alternative to SpaceX seems to be what Iridium is offering (which Casey mentioned in the blog or at least in earlier blog posts I think -- I have really only read this one and the prior one this one links to). And that is nowhere as big a vision as this. Musk was able to do it with Tesla[*]. And I think SpaceX is a much bigger leap that I will be amazed if he could pull off. But there is a chance he could, right?

* at least so far, still plenty of time for things to go pear shaped

[Campitor]

From what I've read, SpaceX has managed to drastically reduce the price per pound for payload making it a cost effective means for delivering satellites into space. I think his plan is very doable but of course his satellite system is at the mercy of weather. I imagine places where storms are numerous or spotty like in the tropics, will experience spotty service. SpaceX has already launched 122 satellites.

Energy, Transportation, and telecommunications are very lucrative markets and Musk has a hand in all three. And all 3 of those markets are very synergistic. The knowledge gained producing battery technology, satellites, electric vehicles, and rocket production/launches can be cross-utilized easily.

[Seppia]

battery technology? Teslay buys their batteries from third parties.

[Riggerjack]

But today, almost all of us get our internet from a tiny handful of geographically-isolated monopolies. In the US, AT&T, Time Warner, Comcast, and a handful of smaller players have divided up the country to avoid competition, charge exorbitant rates for bad service, and bask in near universal hatred.

There is a compelling reason, besides overwhelming greed, for anti-competitive behavior among internet service providers. The underlying infrastructure of the internet, microwave cell towers and optical fiber, are extremely expensive to build.

I work for one of those hated companies. That's our business model! I would say the reason for the anticompetitive behavior, is that is the industry model. Created and enforced by regulation. One cannot operate in this sector in any other way. If SpaceX wants to play here, they will be playing the same game, by the same rules. The league and referees are the same.

But I have a few quibbles with this space fanboy's post. First, low latency.

He talked about how speed of light in vacuum is faster than speed of light in fiber. True. Speed of light in fiber is slowed by about a third in glass. But how far does one need to send a signal to make up that difference in distance? Today, the question is irrelevant, because switching times in equipment is the determining factor in latency. I'm sure brand new tech on satellites will be awesome, and may overcome this hurdle, but it isn't automatic.

Second, I am familiar with a few uses of lasers that just didn't work at all when practical applications were tried.

Tetratech was a company that wanted to bypass the logistical problems if providing LAN/WAN capability to businesses in cities, by bypassing the street level, and just going point to point on skyscraper rooftops. It turns out that skyscrapers move, and pointing a laser from moving rooftop to a receiver on an independently moving rooftop was a problem many millions of dollars didn't solve. They eventually gave up and moved into microwave.

Boeing wanted some of that Star Wars money. So they poured immense amounts of money into lasers for shooting down missiles. And they got one. I saw it, it's mounted as a trophy outside one of the buildings I cabled. But talking to the engineers, there were hundreds of tests, and that was one of the very few successes. Atmospheric distortion, range, and closing speed were limitations money alone could not break.

I'm not saying this is as impractical as sharks with lasers on their heads, after all, this guy isn't Dr. Evil. But there are significant technical hurdles this guy is just assuming away.

The third concern is orbital debris. In a previous post, I pointed out that satellites and debris below 600km will deorbit within a few years due to atmospheric friction, greatly reducing the possibility of Kessler syndrome. I think SpaceX gets a lot of hate for wanting to launch thousands of satellites, as though their designers have never thought of debris. When I look at the details of the Starlink implementation, it is hard for me to imagine a better way of doing it for debris mitigation.

Ok, so we launch thousands of satellites, and keep up with attrition by shooting more up daily. And no worries, they just burn up when they are used up.

From a space fanboy perspective, I am sure that's all good. But from my perspective, as someone who has other uses for the atmosphere than as an impediment to launch, just how toxic is this?

Exhaust from launches, now cheaper than shotgun shells. Burnt up launch tubes, burnt up satellites, and all in the upper atmosphere. That's probably the best place to store oxidized aluminum and rare earths, right? No possible way that could have unintended consequences...

I like the idea, I think there are possibilities here. But I wouldn't invest in this. I like a speculative play as much as the next guy, but damn.

[jacob]

FWIW, in high frequency trading where latency is king, companies competing on hardware went from fiber (IIRC about 2/3 times c, the speed of light) to satellites (c but longer up and down distances) and since about a decade ago to their own dedicated microwave tower network (also c but along the curve of the Earth). Latency essentially moved from some dozen milliseconds to the microsecond range.

I'm guessing most ordinary internet usage doesn't require those kinds of speeds, even gaming. It takes about 200ms for a human eye to blink.

[Riggerjack]

Yeah, there was a kerfuffle a few years ago dealing with high frequency trading, and the jouralists' number of $1B to find a faster route from NY to Chicago...

If anyone actually wants to write some checks, I would be happy to cut those costs for a tiny sliver of the savings...

I imagine there may be room to cut a bit of time by routing through the starlink system for the longer intercontinental circuits, but it won't pay the bills...

[George the original one]

I'm guessing most ordinary internet usage doesn't require those kinds of speeds, even gaming. It takes about 200ms for a human eye to blink.

Simracing is one of those classic gaming things that uses the Internet. To coordinate up to 40 drivers reacting to each other while being scattered around the globe, a 200 ms ping is at the upper end of desireablity and us players are more comfortable in the 100 ms range (because weather and routers happen). You can play at up to about 300 ms ping, but the jitter & warping is nervewracking, so you have to leave extra space around anybody who shows up like that. Geosynchronous satellite ping is usually just under 800 ms... unplayable, but the software will still usually let you connect to a race server if someone is curious.

[Campitor]

battery technology? Teslay buys their batteries from third parties.

Oops, I meant the Gigabit Factory (https://www.tesla.com/gigafactory): In mid-2018, battery production at Gigafactory 1 reached an annualized rate of roughly 20 GWh, making it the highest-volume battery plant in the world. Tesla currently produces more batteries in terms of kWh than all other carmakers combined. With the Gigafactory ramping up production, Tesla’s cost of battery cells will significantly decline through economies of scale, innovative manufacturing, reduction of waste, and the simple optimization of locating most manufacturing processes under one roof.

SolarCity build solar panels but then that too is directly applicable to satellite technology. Elon may face the same regulatory burdens that increase the cost of rocket launches but at least he's producing a lot of products in-house which should reduce cost.

[SavingWithBabies]

@Riggerjack My understanding is the lasers are for communicating between satellites in the mesh network. So they wouldn't need to be going through the atmosphere and they would already be on the satellites which are going to require regular course corrections. So in a way, it's even worse than moving satellites but it's not exactly a problem they can be ignoring. That does raise the question of the uplinks though -- not sure what the idea is there.

The whole issue is the last mile. They can take the same run around the phone/cable companies by not having a physical last mile connection. So just like a mobile data (LTE) connection, they don't have to deal with the monopoly that local governments have granted. Yeah, we elected/appointed those people but they still seem to think of cable TV instead of the value of internet connections. And your monopolies play dirty -- something like this is the usual story.

When you can get a 10 GB internet connection in Japan for $60/month (with a multiyear contract) or go to Sweden and pay even less for a not-quite-as-impressive 1 GB connection, what we have here starts to look bad. Sure, I can get 1 GB connection from the local cable company even in the tiny city I'm living in but it's ~$110/month, $199 install and the quality of both the local connection and the backbone connection is fairly poor (it's asymmetric so the upload is low, they have all kinds of filters in place still so packages don't get maximum speed and the actual speed out of this region via the backbone is poor).

Sure, we can trot out the whole argument about how the USA is a big dispersed country and ... But it's starting to fall flat. At a certain point, the poor service behind those monopolies is going to be an economic liability to GDP.

What we might really need is a viable threat to those monopolies and it's coming in the form of SpaceX (not holding my breath) and faster mobile networks (seems like more of a solid bet). The latency even on 4G LTE is pretty bad[*] so hopefully 5G gets better.

@jacob Back when I was playing first person shooters, a nice connection was around 30-40ms[**]. Once you got above that, you started to be less competitive. Maybe around 90-100ms was when it started to get really frustrating. This might have to do with how interaction requires bidirectional communication so it's really 2 x latency that is important. Then there is some processing time on the game server. I'm not sure where I'm going with this except that lower than 200ms latency is important.

* Could be my consumer mifi device but I've noticed the routing behind LTE data is very suspect -- you usually appear to be coming onto the Internet from some network presence 100s of miles away. I suspect there is a lot of proxying/routing/packet manipulation.

** This was with the original Counter-Strike. Later versions added much more fudging (randomization) to the shooting so lower ping couldn't dominate as much and fiddled with the networking code to try to make things more fair.

[SavingWithBabies]

So if I was more fair to the monopolies, I would argue the problem is two fold:

* Our decision via our elected/appointed local officials to grant the monopolies.
* The impediments to improvements that local officials have put upon improvements (driving up the cost for others to roll fiber, etc).

We've basically made a deal with the devil and the viewpoint is we should be happy we're getting what we're able to. I can understand that line of thinking but outlook is dismal.

[fiby41]

When you can get a 10 GB internet connection in Japan for $60/month (with a multiyear contract) or go to Sweden and pay even less for a not-quite-as-impressive 1 GB connection, what we have here starts to look bad. Sure, I can get 1 GB connection from the local cable company even in the tiny city I'm living in but it's ~$110/month, $199

The cost per GB is roughly 10 cents in India. The plan that I'm on is under $25/year for 1.5GB/day.

[Riggerjack]

Sure, we can trot out the whole argument about how the USA is a big dispersed country and ... But it's starting to fall flat. At a certain point, the poor service behind those monopolies is going to be an economic liability to GDP.

Maybe I have been unclear. I work for Telecom. That doesn't mean I like the business model.

Utilities are best understood as tax collectors. Not like the collection of bureaucratic storm troopers like the IRS, but like the old school tax collectors from early American history. The ones who were less than moral in methodology, and were occasionally tarred and feathered. And yet, they were useful, so abuse in both directions was equally useful to politicians.

We have allowed internet to be regulated as a utility. The FCC has control over bandwidths, and the rules of competition in this market. They are looking after the public interests in the same way they always have. And if starlink can provide better service, at a lower price, AND generate more government revenue than other utilities, they may get a chance to prove it. But the whole time, there will be the lobbying arm of an entire industry, and congressmen "protecting their constituents" tipping the scales.

As for those measurements of speeds, they are nearly meaningless. How fast one connects to an individual node is important for marketing, but not terribly relevant to the user's experience. Traffic shaping and aggregation is just part of the game. Everyone does it, most don't do it with the end user's experience as the priority. Cell data all runs across the same land lines as everyone else, but the routing is handled by the cell company, nowadays.

My point was that all the techn-optimism in the world won't overcome the political pressures that can be applied, unless they understand that they are signing up to be tax collectors, and leave room in the business plan for those costs. Costs overlooked in the original blog post, while the revenues were exaggerated.

I was also impressed by the way all costs go down significantly, and none went up. Surely that is a sign of a balanced opinion... I'm wondering what happens to the cost of rocket fuel in this scenario. IIRC, there are only two solid rocket fuel manufacturers in the country, and that industry is very protected by defense interests. Pepcon, in Cedar City UT is one of them, unless they blew up again... But I don't know what fuels SpaceX is using.

Either way, it seems like an odd detail to leave out of such a thoughtful post.

[SavingWithBabies]

@Riggerjack Honestly, I went on a bit of a rant. I hadn't assumed you liked the business model. I was aware to some degree of the tax-collector nature of it but I had assumed non-physical data connections were less regulated/under the thumb. So by that I mean mobile data (LTE/4G/?) and wireless internet service providers (WiSPs) operating on other radio frequencies.

I know mobile data is not free with the spectrums being auctioned off market-by-market by the FCC. I think the frequencies the WiSPs are using are more open. Which really begs what spectrum/frequencies would SpaceX use and what are the costs? The article suggests perhaps Ka or V bands. So I had to go look those up:

V band = unlicensed so should be like the WiSP usage model
Ka band = wikipedia doesn't say so not sure yet

The reason why I think it's worth looking at that is if the FCC licenses it, the lobbying aspect is more important. But if it is unlicensed... Of course, the lobbying approach could be to try to get these bands licensed and then apply the screws that way?

But your main point, ignoring a potential huge cost area, stands and I don't dispute that.

[SavingWithBabies]

That is an interesting point about fuel. So if one can't invest in SpaceX, maybe one could invest in their potential dependencies? It looks like Kerr-McGee is one competitor but they are a privately held company. I wonder if all of them will be privately held. It looks like PEPCO became WECCO (Western ElectroChemical Company) and is now under AMPAC (American Pacific). AMPAC is also private.

I think @jacob mentioned the trend here about private companies keeping the good investment opportunities outside of the market. I had already started to think that might be the case but interesting to see it here. Although there might be other reasons to keep them private.

[Campitor]

... But I don't know what fuels SpaceX is using.

https://everydayastronaut.com/raptor-engine/ : SpaceX’s new Raptor engine is a methane fueled full flow staged combustion cycle engine...

Very interesting read. It goes into the details behind the Methane engine design.