I have a house on a farm. I only have relatively high speed internet (25mb/s) b/c someone set up a line of sight system pointing at a mountain as a co-op. My neighbors' house about half a kilometer away is blocked by a hill so they can't use it. When it's cloudy, I have to use my phone as a mobile hotspot.
Coming from San Francisco, I didn't realize how many people face this reality of poor connectivity. I don't care much about who is doing this as much as I care that someone is doing it at all. Most people don't realize how much of an internet access disparity exists. Especially as more education moves towards digital, internet access is a fundamental utility.
My parents still live in the house I grew up in. It's about 30 miles from the nearest big city, a top-10 U.S. city.
- They had dial-up until the early 2000s.
- Then they had that terrible asymmetric satellite service where the upstream service was also dial-up. That lasted a couple of years.
- Then they got some kind of two-way sat service. Still terrible latency, but tolerable for the time.
- Finally, sometime in the early 2010s cable television service made its way to their neighborhood. Television, not internet.
- About 2015 they were finally able to get cable internet, which they have today. It's an amazing 5mb/sec up/down.
When I go there I just tether to my phone because its faster. They never got DSL. I bet they'll never get fiber. Starlink could be a serious improvement to them even at 2x the price they pay now.
Data caps and throttling make it prohibitively expensive.
I’ve done it for weeks in the past, and (what I personally consider normal usage) burned through my monthly cap in a week, leaving me with large overage charges that wouldn’t be sustainable long-term. I haven’t heard of a US carrier-provided plan that’s less than several hundreds of dollars, and if there is one it’s likely unavailable in rural areas
There's actually resellers that are made for rural areas with unlimited data. They are quite expensive, but not several hundreds of dollars expensive. I'm pretty sure I have seen some less expansive but here's one: https://unlimitedville.com/plans
For a year my family used 4G as our home internet (this was around 8 years ago). At the time it was incredibly expensive because you pay by the gig used, and the pay scale is built around using a couple of gigs. Watching a couple of episodes on netflix bumps you into a higher tier for the month. Downloading and then uploading a single 1GB artifact for work could bump you up 2 tiers. I've looked around a bit, and the situation doesn't look much better today; you can pay for a certain data cap, and then pay something like $10 / month / GB over that.
I was, long ago, a microwave radio tech. It's cool to me that line of sight issues still create issues. For the most part, my old line of work is dead because tech marches on. My immediate reaction is that your neighbor should be able to use a troposcatter shot[1] to fix his issues, but I suppose the provider has bigger fish to fry.
I'm a bit sad about it, because tech advances have technically made this sort of thing dead easy.
Well, "dead" meaning broad use. I'm aware it's niche used here and there. Any troposcatter shots? Or is that dead from a price/performance point vs satellite or other?
I moved from Sunnyvale to Los Altos Hills and lost access to broadband. Comcast even told me they offered service at the address, then reneged when I moved in and asked to set it up.
More options would be great, and underserved areas aren't as far out as people often think they are. You can be five miles from Google's headquarters and unable to get decent internet.
My neighborhood (14 homes) is working towards building a fiber island with wireless backhaul elsewhere. We're the houses right around the Rhus Ridge trailhead. But there are even-closer-to-urbanization neighborhoods without access, e.g., Wildcrest and Wildflower.
Isn't it possible to install a mast with a pair of directional antennas on the house to get this 4G signal? Plug the antennas in an all-in-one modem/router/access point and enjoy great Internet in the house.
4G is amazing these days, I recently moved from a big city (coax 500 Mbps) to a house that only gets 4G and I barely feel the difference.
No, the house is too low behind a low hill for a short mast, and a very high mast wouldn't be ok. The way to do it would be 4G at the top of the hill and a cable or relay to the house, but as it's a populated village that would need cooperation among residents.
If there was plenty of cooperation and interest among the residents, they could install a better shared solution for the whole village than 4G even. There was a company willing to install something like that. But not enough people signed up to make it worthwhile.
I agree with you about 4G. I switched from ADSL+ to 4G last year in the middle of a city, and it's been a good improvement in download speeds, and an x10 improvement in upload. Netflix works excellently. Though there tend to be latency spikes, and new connections to websites for some reason are quite slow. I actually switched to save costs - unlimited data on 4G was less than half the price of bottom-of-the-range ADSL too.
When I want the best connection out of every option, such as a big download. I switch to my mobile phone, which seems perverse but there we are. I get ~90Mbit/s download via my phone. It's faster than wifi at the office, faster than ADSL, and faster than 4G at home too - the phone has a better modem than the standalone 4G router.
However I've tried 4G in London, which is the largest city in the UK, and it was awfully slow there. Too much density I suspect. So it's not a solution for everywhere. I think I'm lucky with it where I live. Around the same city, there are plenty of "notspot" shadows where there's no signal, not even 3G.
It's really cost. An extra hotspot device is >$100/month. My line of sight is $50month. My phone has unlimited data as well but for things like a security system, it's not something I want to rely on primarily if I leave and need to bring my phone with me.
I'm in a busy city so I don't think I will have a Starlink option, but I'd gladly pay twice as much to fund this than keep paying money to our evil overlord Comcast.
Keep in mind, Comcast owns NBC as well. So probably we’ll get a lot of negative coverage of Elon Musk on NBC and MSNBC in the process.
It’s much easier to lobby against a public villain than a public hero.
So every tweet by Musk will be amplified out of context.
Wouldn’t surprise me even some allegations of foreign impropriety will surface and soon having starlink becomes a national security unless it’s sold to Comcast or it’s offered in a way that won’t compete with Comcast.
I know it's very hip and cool to be cynical about journalistic independence these days but newsrooms still take the concept very seriously. There's simply no mechanism by which the CEO of Comcast could simultaneously a) meaningfully affect NBC's coverage of Elon Musk and b) not have it be an open secret and scandal within the industry.
I've worked on the product side of legacy media companies before and trust me, given the disorganization I was seeing in the editorial departments at my peer orgs, they couldn't have pulled this off even if they wanted to.
NBC's coverage of Elon isn't any more negative than it's sister orgs and the simple, occam's razor explanation is that his ridiculous behavior is absurdly good at generating clicks rather than any organized media endeavor against him.
On “journalistic independence”, NBC told Ed Schultz not to cover Bernie Sanders and why Schultz and Cenk both left the network [1]
Im sure we can agree that ratings are one of the key drivers and pushed by management. Here’s how cable media profited by covering Trump. This Trump coverage resulted in 160% increase in ratings and in return gave Trump $3 billion [2]in free advertising ( I’ve seen $5billion number in total free coverage as well).
Look it's one thing to hate on big corporations for existing actions, but's entirely something else to invent conspiracy theories. In reality Comcast isn't threatened by Starlink because Starlink market is aimed at less populated areas so Comcast has no need to fight it. It would be completely oversaturated by cities and cities are easy to invest in building network infrastructure.
Now, I could imagine other satellite internet companies to fight it, but they're much less powerful.
Comcast has repeatedly lobbied for legislation to prevent the creation of municipal broadband providers; it's not that much of a stretch that they could try to use regulatory capture to attack Starlink as well.
A new worldwide internet provider being led by a pioneer of innovation is a major threat to comcast. Especially since Musk is known for caring about providing a valuable product at a fair price to the masses (the Model 3).
Elon Musk aside, who wouldn't drop comcast/verizon in a second if there were another viable option at a better price?
Starlink doesn't have to completely match Comcast one-to-one, they just have to be a credible competitor. That increases the bargaining power of the consumer. Particularly as Starlink grows and as OneWeb and especially Project Kuiper comes online.
These low-latency satellite constellations will reduce the local oligopoly/monopoly rents (and thus profitability) of Comcast, AT&T, Verizon, CenturyLink, Cox, Frontier, etc... And the money that would've been profit will go to consumers' pockets, even if they don't themselves subscribe.
Given the capacity of the network, it is more likely that you would be paying 20x-100x as much, per byte, and would get perhaps the same factor, as the denominator, in terms of data cap. Comcast has several times the capacity of the entire Starlink global constellation in each real PoP.
That would be selfish, sort of like buying the hot Christmas toy when you don't even have kids. The amount of available bandwidth at any given time is fixed. If everyone does that then SpaceX would either need to raise prices (so the people with no other option will have to pay more) or implement some other kind of throttling.
It's good if Starlink has enough demand to make the service a success, so SpaceX will want to expand it. Beyond that, you're just pricing other people out, until they grow to the point that they can actually serve everyone.
I am assuming a lot of lag between giving them money and the launch of the satellites. (And other bandwidth improvements needed.) Similarly to how buying the hot Christmas toy doesn't automatically cause the manufacturer to make more in time for Christmas.
I suppose owning the launch systems helps, but there is still going to be lag.
That sounds correct from a supply chain and like, physical reality perspective, but hopefully there would be ways to increase the bandwidth density to counteract the launch problem. If there's not, and there's like a year turnaround on launch, then your concern is founded.
Perhaps the reader would be better off making a donation or investment into SpaceX instead, such that they receive money to better compete with Comcast in exchange for no obligation of service :)
Isn't it a phased-array antenna? It can adjust itself at the speed of electronics, without moving parts, so that should not be an issue in any reasonable scenario if the software supports it.
A larger problem for use on sea might be that first generation Starlink sats do not have satellite-to-satellite links, so the boat and a ground station would have to be within range of the same satellite.
You still have to orient the whole antenna into the direction of the right part of the sky. From then the phase antenna will be able to track the individual satellites.
That sounds like it'd still wobble a lot, compared to a land mount. Maybe a later version will have beefed-up phased array controls to cope, but that'll be much later, they have easier wins first.
I'm going to buy a summer house in the boonies when this becomes available (and reliable). This will finally make it possible for me to truly "work from anywhere", including the middle of nowhere, which is where I'd like to be in this day and age.
SpaceX/Starlink have done effectively no marketing spend to get those leads into their funnel. Subscriber acquisition costs (SAC) for home broadband and TV are very high, and due to churn doesn't necessarily even translate into net subscriber adds. Those numbers are good by any measure.
One representative example, Subscriber acquisition costs for Dish network:
"DISH TV SAC was $822 during the year ended December 31, 2019
compared to $759 during the same period in 2018, an increase of $63 or 8.3%." [1]
"We lost approximately 511,000 net DISH TV subscribers during the year
ended December 31, 2019 compared to the loss of approximately 1.125 million net
DISH TV subscribers during the same period in 2018. This decrease in net DISH TV
subscriber losses primarily resulted from a lower DISH TV churn rate and higher
gross new DISH TV subscriber activations." [1]
Elon’s Twitter is very much marketing. He not only has over 30M followers, but also each of his statements and companies get disproportional media coverage. Marketing isn’t only buying ads.
We incur significant upfront costs to acquire subscribers, including advertising, independent third-party retailer incentives, payments made to third-parties, equipment subsidies, installation services, and/or new customer
promotions."
Unless they are working for free any publication would cost money. starlink.com exists and is filled with marketing graphics. So they spend money, Q.E.D. .
It could still be impressive if they published actual numbers. Instead we have that $0 claim that gets repeated.
Disclaimer: They could pay people with "exposure" I heard that is the way to go with graphics artists at least.
I wanted to point out that for all of Elons tweets he certainly isn't the guy making the graphics and videos. Unless SpaceX is paying their graphic designers using exposure or outright slavery they are spending money on it.
I'm curious about how bad jitter has to be to have an effect. Does it cause problems at +/-10ms? As I understand it, the link will change from satellite to satellite as they fly overhead
With voip at least, you can use a jitter buffer to get decreased jitter at the cost of increased latency. If latency is good to start with, that's definitely an option.
I'm just not seeing how a hundredth of a second of latency variance would have a noticeable effect at all on either of those. Latency variances of much more than that are already common (and dealt with).
I will confess that I don't get it and I would appreciate thoughts on why this is a good idea instead of a FOMO kind of thing (and especially AWS Kuiper),
The total switching capacity of the entire __planned__ constellation is around that in a _single_ large switch chassis or internet core router and the individual satellites have the capacity of a beefy enterprise Wifi6 access point; however the geographic area each will service is enormous.
What I can't figure out is: what is the market? The market for an Iridium-style remote access is demonstrated to be tiny, there just is not the bandwidth or capacity to target "unserved" areas for ordinary internet.
So what's left that's big? Industrial IoT? Military? Even industrial placements (Rio Tinto, etc. mining concerns) will prefer to set up dedicated links and regional wireless to this solution, so IIOT use case is really a pre-development stage market, which is again small.
I even went looking for what the business press said (example: [1]). Reading this almost makes me think the author is in on a sort of dark joke; especially the “Fortunately, companies have new options for generating revenue from connectivity” section appears to be written by someone with a very dark sense of humor who is straight up trolling the reader.
edit: added "planned" to distinguish the long term from the immediate term [the first batch Starlink satellites didn't even do any in-space switching at all]..
Can you point me to your source on the planned constellation total switching bandwidth? I haven't read about that anywhere.
I have read that they don't expect to be able to provide high levels of service in dense urban areas. Which does make me question the business model...
I think you nailed it that the military and government will be a big customer. Iridium has ~1300ms+ latency and very low data rates, very high costs and the government is their largest customer from what I understand.
Airlines are another likely market that would be a significant amount of business (post-covid19).
If Starlink can compete against Iridium and the other existing satellite providers on price, they could at a minimum take most of that business.
Per satellite max bandwitdh, per SpaceX, at commercialization will be a max of 17gbps [1, sorry for the lame source but it's what I could find at the moment].
Depending on how they're measuring that, the actual throughout could be substantially less than that. Given that their FCC filing was 20gbps per satellite, I think that's the raw, uncorrected bit rate, not the corrected bitrate which will be lower.
A coworker and I did a back-of-envelope guess based on their power budget (switching capacity in bandwidth and pps and power are basically directly tied, ditto capacity and spectrum) and that's about what we arrived at as their upper bound (imho, not even half of that will be achievable). Note that this is a double-counted metric (bps total is bps_in+bps_out) for actual traffic volume.
So let's go with that.
The current SpaceX fantasy business model is to build out a constellation of 30,000 satellites, but near term not even close to that. Being very generous, let's credit them with an incredible 10k satellites, pretend they are not power constrained and use the full 20gbps allotted to them, and you get a total of 200k gbps total capacity for the constellation, uncorrected, assuming ample power, or on a good day 125Tbps for real data.
Using a somewhat older core switch, the Juniper QFX10016, can do line rate switching for 480 100GE ports; 48Tbps in a single 21RU chassis from 2017. [2] These are commonly deployed in pops. The next generation will be doing that for 400GE ports in the same form factor, 192Tbps. A Cisco 8818, which is shipping right now, does 260Tbps per chassis in a 33RU form factor.
The point is, anyone claiming we could serve even a large metro with a single one of these switches, let alone planet earth, would be laughed at, yet here we are. A given nation-scale network will have hundreds of routers with these kinds of capacities.
I also don't think Starlink will get anywhere near the estimates I'm using here.
I think something to consider here is that there is an iterative development process taking place with the satellites. Given the stated lifespan of the satellites is 3-4 years, they likely expect to see additional innovation over time and are gambling that the network will be improved over time, and may one day actually be capable of serving urban areas.
This seems like a wise approach and appears to work well for both SpaceX and Tesla.
The most viable use case in the near term is clearly to serve remote and rural areas which is admittedly a small market. There is a chance you might see governments subsidize service to rural areas that are underserved. I also imagine there will be a number of marine and aviation customers - though that's likely even smaller than the rural market.
But yeah, I'm not one of the people who will claim it will serve urban markets any time soon... That's rediculous - as rediculous as saying Tesla will achieve level 5 autonomous driving this decade
I don't think the iterative nature actually helps SpaceX. Whatever use case they're targeting, and like I said, I don't know, and would like to know, isn't likely to be made more viable by bandwidth that changes substantially by the hour, and the total cost of getting a satellite into orbit makes generational turnover doubly challenging.
I just do not get the business model. It makes no sense for commercial internet access, even in rural areas. I think perhaps temporary access in undeveloped areas, where the total capex is lower than that needed for a temporary LTE deployment, for example, might be it, but that is not a big market.
I guess we'll see if they can pull a rabbit out of a hat. A lot of folks didn't think battery electric cars were going to be possible either. And... Landing freaking rockets on barges in the ocean? I'm no fan boy, but I'm also not going to be a naysayer on this one.
Starlink lets SpaceX make use of cores with long launch records that others may not be willing to risk their payloads on, and SpaceX can also use those launches to test other risky maneuvers. So... Perhaps there's some other value in Starlink that we're not privy to.
IMO, the market is rural folks who don’t have fast internet or any other options. Less than 6mbps down is normal in rural Ohio, for example. And there are no other wired options, and existing satellite options aren’t much better.
These people can even be just a mile or less outside of a town with internet options, but none of the telecom companies have upgraded capacity in the countryside for years. Obviously, a modern wired connection would be way better, but ISPs are just not willing to invest in the upgraded infrastructure where nearly no one lives. You could lay a mile of fiber to reach only one or two households. (Or less.)
Satellite makes a lot of sense here because it can cover the whole area without having to build a ton of infrastructure relative to the population.
> In addition to getting the satellites in orbit, SpaceX will need to build a vast system of ground stations and affordable user terminals if it is going to connect consumers directly to its network.
Why are ground stations required? I thought you were connecting from your dish to the satellite and therefore could be anywhere with a view of the sky.
There is currently no satellite to satellite communication with the first iteration of the technology. This means you need ground stations to receive the traffic coming from the user terminals. This can either hooked up directly to the internet backbone (ideal), or bounce to another satellite closer to the backbone connection.
Basically, user computer <--> user wifi or router <--> user terminal <--> satellite <--> ground station <--> fiber <--> peering to other internet providers.
There are two kinds of ground stations: the one at your home/business (the modem/"pizza-box antenna") and the ground stations that connect the satellites to the trunk lines/backbone.
The ground stations will have very fast fiber connections to the greater internet for the satellite network. The user terminals will connect to dishes / antenna at peoples houses, like directv
The ground stations tie the satellite network into the internet. Once they have the inter-satellite laser links operational, they will be able to get by with many fewer ground stations.
Eventually you'll want to talk to a computer that isn't directly connected to starlink. The ground stations are necessary to bridge that gap (and run the constellation)
I know Wyoming law enforcement rely on apps that can function in a disconnected environment since I heard network coverage through out many parts of the state is really poor. This would also help a lot of park rangers.
I would love to have this as a backup internet option. I live in a town that has Google Fiber (literally 50 feet away from me), but I don't have access. I'm oncall and cannot be without internet at almost any time, and my internet goes down much more than I'm comfortable with. I hope Starlink is an option soon!
That is what I currently do. I use >~1TB a month in bandwidth though, so it isn't practical to tether everything so I end up only tethering my laptop and even that is still pretty expensive on Google Fi ($10/gb until I am throttled).
Fortunately my internet has only had problems maybe 4 or 5 times in the last year. It just feels irresponsible of me to rely on data that eventually gets throttled to be almost unusable when I need it for my job. Maybe I will get a backup AT&T internet connection and expense it at some point :(.
5 million times, say, $50 means $250 Million a month or 3 billion a year. Not sure if they can get people to switch from optic cable to satellite and while the world is large, very few outside rich countries can afford to pay $50 a month.
> Not sure if they can get people to switch from optic cable to satellite
I think the market for this often would not even have a cable option. They’re stuck on old DSL at like 6mbps down or less for $90/mo with bundled landline, or if they’re lucky, the cable company ran a line by their house. If starlink was offering just 20mbps down for $50, everyone I know in the country would switch immediately. I definitely don’t think starlink will compete with city fiber or cable connections. Heck, mine is $30 for 250mbps down. No way satellite is going to compete with that.
In the city, and even small countryside cities, this isn’t a big deal, but rural communities don’t have choice or speed, at least in the US! That’s not a big market on a small scale, but added up across the US, it would be a lot of people.
But you don't have to sell it at $50 everywhere. The marginal cost of an additional subscriber is very low, particularly when the inter satellite link is enabled. If people can only afford $10 per month that is better than having no revenue at all in a particular area.
There's nothing stopping others from doing the same thing, and indeed others are planning to. SpaceX is leading the pack, but eventually there will be competition.
Coming from San Francisco, I didn't realize how many people face this reality of poor connectivity. I don't care much about who is doing this as much as I care that someone is doing it at all. Most people don't realize how much of an internet access disparity exists. Especially as more education moves towards digital, internet access is a fundamental utility.