6 ms·

A lot of what you're saying isn't actually true here. Light moves significantly (47%) slower through optical fiber than through space. This drastically reduces

A lot of what you're saying isn't actually true here. Light moves significantly (47%) slower through optical fiber than through space. This drastically reduces the latency of long distance connections. We're talking NY round trip to Japan in the same time it takes NY roundtrip to the UK now. The cost of backhaul and ongoing maintenance of cables is a lot more than SpaceX launching their own satellites by an entire order of magnitude.

bargl 6y ago
>A lot of what you're saying isn't actually true here. You're talking latency not bandwidth. There is only so much spectrum to go around. My understanding is that there is only a certain amount of bandwidth to go around and there are diminishing returns on adding people beyond a set capacity. I am not arguing that fiber is faster, I'm arguing that fiber has is cheaper at the bandwidth/cost in more dense areas. I don't believe we will see satellite based communication in cities (at a large % of population) for a long time. Now I'm willing to believe someone who's worked closer to this than I have and to learn, but I've done enough in the satellite industry to not take what you're saying without seeing more evidence.
- busterarm 6y ago
  I worked for an ILEC providing cable/dsl/fiber service and just the rough napkin math around SpaceX's launch costs makes it look significantly cheaper. Besides the whole digging and laying cable costs, last mile work is expensive. You have to employ huge crews. Construction crews constantly dig up & cut your infrastructure "by accident" and the cost of recovering the fines from them is as much as the fines pay out. Having to send an emergency crew of three guys out for 18 hours of work to re-splice a 72-ct fiber every other weekend adds up. That doesn't happen in space.
  - bargl 6y ago
    >last mile work is expensive. What do you mean there?
    - busterarm 6y ago
      SpaceX isn't doing the install work work on these Starlink kits. They just sell you some hardware. If you're a terrestrial-based internet provider, you need a fleet of installers and they also have to do constant maintenance work. They have salaries and they make (lots of) overtime pay. At a lot of the smaller companies (like the one I worked at) they have pensions. We had a lot of guys who started with the company out of high school, retired at 20 years, came back as contractors for a year and then did another 20 years and got a second pension from the same company... And a couple of those guys are still contracting after their two pensions.
      - bargl 6y ago
        SpaceX can't scale regionally, they have to scale globally. The nature of their satellites is that if they want to accommodate 2x as many users in a city they have to double their capacity. The push I'm discussing is the cost to get a baseline internet out there for 10k people per 100 miles. I'm assuming the cost per last mile is higher in rural areas than in urban areas. If that's true right now the majority of our population is in cities so SpaceX will have trouble reaching them. That's the core of my point. For Starlink to hit everyone in a city they'd have a huge amount of wasted capacity in rural areas. Satellites also degrade naturally over time. They have a set lifetime and don't all last that long. Cables under the ground tend to last longer even if they require spot fixes. I'm assuming that bandwidth wise, the cables are more efficient at least until getting manufactured equipment to space becomes even more cheaper.
        busterarm 6y ago
        Yeah but we're talking about a company that has in their near-term plans the capability to _launch 4 rockets per day_. They're completely capable of adding capacity like that and having a high satellite count. I'm not saying that it doesn't have a cost, but they have no middleman. They don't need to buy space on somebody else's rocket. They also don't have to pay any ongoing maintenance cost beyond equipment replacement and it's much cheaper than doing it on land. I know that you have satellite experience but their launch costs are a fraction of what other companies have done till now. I worked for a suburban, regional ILEC and our operating costs were around 100 million a year to cover like two counties of maybe 100k customers and the local businesses. ... I actually do agree with you about costs delivering in dense markets as providing locally has certain "economy of scale" benefits. It's also not going to compete on the "Gigabit in every home" front, because I do agree that the bandwidth is more limited. Operating a local provider in general though is extremely expensive and doesn't really provide any additional benefit to the provider. SpaceX themselves have uses for this network. I also hope that I'm right here :)
  - kitsunesoba 6y ago
    The SpaceX approach also completely bypasses a ton of the nonsense local bureaucracy, stonewalling by existing ISPs, etc that a company would have to deal with to put in cables, towers, etc in the first place, saving both time and money.
    - busterarm 6y ago
      Yup. This is actually one of the more important factors for consumers that is underestimated. I've been pegged at 1000% excitement since I first found out about this project a year ago. This is the best thing to happen to this service industry since probably ISDN.
    - ckocagil 6y ago
      This is not a good thing. Eventually countries will heavily regulate satellites and then SpaceX will be the monopoly that keeps small players out. Instead of postponing the problem of getting rid of an evil monopoly, citizens should realize that last mile internet delivery is practically identical to water, sewage and power systems that can be owned by the city and leased out to providers.
  - 01100011 6y ago
    FWIW, last mile costs only matter if your last mile isn't wireless, more or less. Terrestrial internet doesn't imply wire/fiber to the premises, does it?
  - ckocagil 6y ago
    The costs of installing and maintaining infrastructure scale down to nothing with population density. Also you can upgrade a switch but not if it's on a satellite. I am very convinced that the two approaches will complement each other in the foreseeable future.
  - shaklee3 6y ago
    Please post your napkin math. That seems to disagree with all the other analysts' math.
pmlnr 6y ago
> Light moves significantly (47%) slower through optical fiber than through space The what? As far as I know c is a constant.
- akalin 6y ago
  c is the speed of light in a vacuum
- kibwen 6y ago
  The constant c is the speed of light in a vacuum. Different mediums have different speeds of light within them; in fact, it's actually possible for light to not be the fastest thing in a given medium (for more precise definitions of all those words): https://en.wikipedia.org/wiki/Cherenkov_radiation https://en.wikipedia.org/wiki/Cherenkov_radiation
- busterarm 6y ago
  When you refract light, you change its speed, slowing it down. https://en.wikipedia.org/wiki/Refractive_index#:~:text=The%20refractive%20index%20can%20be,of%20that%20light%20in%20vacuum https://en.wikipedia.org/wiki/Refractive_index#:~:text=The%2.... https://www.zmescience.com/science/physics/how-to-slow-trap-light/ https://www.zmescience.com/science/physics/how-to-slow-trap-...
- sannee 6y ago
  Altought the answers so far posted are not exactly incorrect, it is also important to note that the geometry (that is, a cylinder for optical fibers) of the waveguide used influences the group velocity of the propagating electromagnetic field. This will always end up being slower than a plane wave in the same material, usually only somewhere between 10%-90% (and this also depends on the wavelength) of the free-space group velocity.
- RJIb8RBYxzAMX9u 6y ago
  In a vacuum. See Cherenkov radiation[0]. [0] https://en.wikipedia.org/wiki/Cherenkov_radiation https://en.wikipedia.org/wiki/Cherenkov_radiation
- 6y ago
shaklee3 6y ago
SpaceX has to pay the same fiber cost in terms of latency as everyone else. They do not have gateways near every PoP, and will need to backhaul a nontrivial distance to add latency.
skrause 6y ago
The current generation of Starlink doesn't actually have any optical inter-satellite links which would forward your packets from satellite to satellite around the world. Your packets will go up to the satellite, immediately down again to the next ground station and from there to the destination through the normal fiber backbone.

bargl 6y ago

>A lot of what you're saying isn't actually true here. You're talking latency not bandwidth. There is only so much spectrum to go around. My understanding is that there is only a certain amount of bandwidth to go around and there are diminishing returns on adding people beyond a set capacity. I am not arguing that fiber is faster, I'm arguing that fiber has is cheaper at the bandwidth/cost in more dense areas. I don't believe we will see satellite based communication in cities (at a large % of population) for a long time. Now I'm willing to believe someone who's worked closer to this than I have and to learn, but I've done enough in the satellite industry to not take what you're saying without seeing more evidence.

busterarm 6y ago
I worked for an ILEC providing cable/dsl/fiber service and just the rough napkin math around SpaceX's launch costs makes it look significantly cheaper. Besides the whole digging and laying cable costs, last mile work is expensive. You have to employ huge crews. Construction crews constantly dig up & cut your infrastructure "by accident" and the cost of recovering the fines from them is as much as the fines pay out. Having to send an emergency crew of three guys out for 18 hours of work to re-splice a 72-ct fiber every other weekend adds up. That doesn't happen in space.
- bargl 6y ago
  >last mile work is expensive. What do you mean there?
  - busterarm 6y ago
    SpaceX isn't doing the install work work on these Starlink kits. They just sell you some hardware. If you're a terrestrial-based internet provider, you need a fleet of installers and they also have to do constant maintenance work. They have salaries and they make (lots of) overtime pay. At a lot of the smaller companies (like the one I worked at) they have pensions. We had a lot of guys who started with the company out of high school, retired at 20 years, came back as contractors for a year and then did another 20 years and got a second pension from the same company... And a couple of those guys are still contracting after their two pensions.
    - bargl 6y ago
      SpaceX can't scale regionally, they have to scale globally. The nature of their satellites is that if they want to accommodate 2x as many users in a city they have to double their capacity. The push I'm discussing is the cost to get a baseline internet out there for 10k people per 100 miles. I'm assuming the cost per last mile is higher in rural areas than in urban areas. If that's true right now the majority of our population is in cities so SpaceX will have trouble reaching them. That's the core of my point. For Starlink to hit everyone in a city they'd have a huge amount of wasted capacity in rural areas. Satellites also degrade naturally over time. They have a set lifetime and don't all last that long. Cables under the ground tend to last longer even if they require spot fixes. I'm assuming that bandwidth wise, the cables are more efficient at least until getting manufactured equipment to space becomes even more cheaper.
      - busterarm 6y ago
        Yeah but we're talking about a company that has in their near-term plans the capability to _launch 4 rockets per day_. They're completely capable of adding capacity like that and having a high satellite count. I'm not saying that it doesn't have a cost, but they have no middleman. They don't need to buy space on somebody else's rocket. They also don't have to pay any ongoing maintenance cost beyond equipment replacement and it's much cheaper than doing it on land. I know that you have satellite experience but their launch costs are a fraction of what other companies have done till now. I worked for a suburban, regional ILEC and our operating costs were around 100 million a year to cover like two counties of maybe 100k customers and the local businesses. ... I actually do agree with you about costs delivering in dense markets as providing locally has certain "economy of scale" benefits. It's also not going to compete on the "Gigabit in every home" front, because I do agree that the bandwidth is more limited. Operating a local provider in general though is extremely expensive and doesn't really provide any additional benefit to the provider. SpaceX themselves have uses for this network. I also hope that I'm right here :)
- kitsunesoba 6y ago
  The SpaceX approach also completely bypasses a ton of the nonsense local bureaucracy, stonewalling by existing ISPs, etc that a company would have to deal with to put in cables, towers, etc in the first place, saving both time and money.
  - busterarm 6y ago
    Yup. This is actually one of the more important factors for consumers that is underestimated. I've been pegged at 1000% excitement since I first found out about this project a year ago. This is the best thing to happen to this service industry since probably ISDN.
  - ckocagil 6y ago
    This is not a good thing. Eventually countries will heavily regulate satellites and then SpaceX will be the monopoly that keeps small players out. Instead of postponing the problem of getting rid of an evil monopoly, citizens should realize that last mile internet delivery is practically identical to water, sewage and power systems that can be owned by the city and leased out to providers.
- 01100011 6y ago
  FWIW, last mile costs only matter if your last mile isn't wireless, more or less. Terrestrial internet doesn't imply wire/fiber to the premises, does it?
- ckocagil 6y ago
  The costs of installing and maintaining infrastructure scale down to nothing with population density. Also you can upgrade a switch but not if it's on a satellite. I am very convinced that the two approaches will complement each other in the foreseeable future.
- shaklee3 6y ago
  Please post your napkin math. That seems to disagree with all the other analysts' math.

pmlnr 6y ago

> Light moves significantly (47%) slower through optical fiber than through space The what? As far as I know c is a constant.

akalin 6y ago
c is the speed of light in a vacuum
kibwen 6y ago
The constant c is the speed of light in a vacuum. Different mediums have different speeds of light within them; in fact, it's actually possible for light to not be the fastest thing in a given medium (for more precise definitions of all those words): https://en.wikipedia.org/wiki/Cherenkov_radiation https://en.wikipedia.org/wiki/Cherenkov_radiation
busterarm 6y ago
When you refract light, you change its speed, slowing it down. https://en.wikipedia.org/wiki/Refractive_index#:~:text=The%20refractive%20index%20can%20be,of%20that%20light%20in%20vacuum https://en.wikipedia.org/wiki/Refractive_index#:~:text=The%2.... https://www.zmescience.com/science/physics/how-to-slow-trap-light/ https://www.zmescience.com/science/physics/how-to-slow-trap-...
sannee 6y ago
Altought the answers so far posted are not exactly incorrect, it is also important to note that the geometry (that is, a cylinder for optical fibers) of the waveguide used influences the group velocity of the propagating electromagnetic field. This will always end up being slower than a plane wave in the same material, usually only somewhere between 10%-90% (and this also depends on the wavelength) of the free-space group velocity.
RJIb8RBYxzAMX9u 6y ago
In a vacuum. See Cherenkov radiation[0]. [0] https://en.wikipedia.org/wiki/Cherenkov_radiation https://en.wikipedia.org/wiki/Cherenkov_radiation
6y ago

shaklee3 6y ago

SpaceX has to pay the same fiber cost in terms of latency as everyone else. They do not have gateways near every PoP, and will need to backhaul a nontrivial distance to add latency.

skrause 6y ago

The current generation of Starlink doesn't actually have any optical inter-satellite links which would forward your packets from satellite to satellite around the world. Your packets will go up to the satellite, immediately down again to the next ground station and from there to the destination through the normal fiber backbone.