Hacker News new | past | comments | ask | show | jobs | submit login
Starlink review, four months in (jeffgeerling.com)
500 points by geerlingguy 3 days ago | hide | past | favorite | 373 comments





It's sort of mentioned, but not emphasized in this review that connection dropouts happen every few minutes for a few seconds. That makes starlink fine for any kind of asynchronous content like web browsing, torrenting or video streaming, but unusable for video calls, stream hosting, voip, or online gaming. It's implied that this is due to the trees obstructing a full view of the sky, but I have actually heard these connection dropouts are just about universal due to the constellation not having enough infill. Just a warning that for most of us we are still several hundred satellites short and some connection handoff updates away from this being a useful internet connection.

I have a property where Starlink would be perfect and I would pay triple the price to be able to do zoom calls over the connection.


Honestly when I first got the dish, and had it in an open field, that was the case—but now the momentary dropouts between satellites are less than 1-2 seconds.

They had 1000 or so satellites when I first started testing, and there are now something like 1600 or so. Most of the time, I don't even notice when it switches satellites.

If you were doing some more real-time work or extremely latency-sensitive operations, then yes, you need to stick to a different type of connection. But it's really seamless now, compared to even a few months ago.

Most of the software I used either showed no sign of the dropout, or at worst would freeze a frame or show a loading indicator for a brief moment before getting back to normal.

Online multiplayer gaming and/or streaming are the main areas where I'd have to not recommend Starlink for now.


I'm thinking of using this for a backup connection if FIOS fails... we'd switch our office over to starlink... currently we switch to comcast and it's basicaly unsable... so wonder if starlink would be good for ssh connections etc...

I work from home on Starlink. My SSH sessions work just fine. When I first got the service, it was dropping quite a lot, but even then SSH worked pretty well. But these days, if there are any drops, I can't tell. The improvements in the last few months have been massive. I am even about to drop my land line, because WiFi calling on my cell phone has been nearly perfect. It works better than my cell phone does when I am in a good service area.

mosh instead of ssh helps for unreliable connections

When I connect to remote servers over SSH, I usually start a screen (another alternative is tmux) session and then perform my administrative and diagnostics actions in that session. This way, even if I get disconnected, I can reconnect and join that running session. There are other benefits to a screen multiplexer - such as multiple remote users being able to attach to the same session, a joy when a distributed team needs to diagnose an issue or watch some actions together.

You may also want to look into mosh [1] as it will handle dropouts and continue where ssh would just fail on its own. Also has local echo for points of high latency.

Was a real life saver on some of the terrible sat connections of yesteryear.

1. https://mosh.org/


Mosh is great, but if you don't need the predictive terminal emulation, then you can try Eternal Terminal, which is easier to get running due to not needed an UDP port open.

How does it compare to autossh?

autossh just re-reestablishes your ssh connection. It doesn't tolerate packet loss very well or have local echo like mosh.

mosh does require UDP ports, though, which might be problematic with some firewalls or proxies.


Combine with mosh and you can even keep editing while the connection blips!

yeah i use screen for everything too... still there are times when a connection dropping is a bit scary...

LTE makes a lot more sense if you have a decent signal. The cost is lower and speeds are comparable in many areas if you have good hardware.

If you live somewhere with FiOS, Starlink is going to be a non-starter. FiOS requires high density. High density kills Starlink connections.

I suppose Starlink can support a small, negligible, percentage of customers in an urban (or even suburban) area. (Same total number as a rural area.) So, if you think they are going to cutoff registrations as oppose to oversubscribe when they hit that point you may want to sign up right away.


I'm not sure if you are using FiOS as a synonym for FTTH. If so, you might be interested to know I live somewhere with FTTH that is reasonably low-density (~25-30 people per sq km). I think Starlink will be a very serious consideration here.

When I say FiOS, I'm thinking gigabit+ FTTH/FTTD (I'm not 100% sure of the difference). Given that you quoted people per sq km (about 70-84 people per sq. mile), you're probably not in the US. In the US, it's hard to get fiber outside of areas with the high density I'm quoting (at least in my experience).

But yes, that density seems fine for Starlink (although maybe it isn't in the end, who knows yet.)


My understanding is that StarLink is aiming for 100M un-guaranteed in beta. So that's considerably below gigabit.

But I also know that most households with FTTH around here use considerably less than 50M.

So perhaps you meant that gigabit internet kills Starlink. FTTH has some advantages over Starlink. But Starlink also has advantages over FTTH. If you need more than 100M, Starlink is not in the game yet. But if you want some leverage with the local monopoly and have typical usage...

Or, if you snowbird (common case here) and want the same provider on the border of Canada and the border of Mexico...


The total aggregate bandwidth Starlink will have when their entire constellation of satellites is deployed and the number of satellites that will be visible at a given location at a given time place an upper limit on the number of customers they can have even with overselling bandwidth.

In the US their plans are to be able to support something like 5 million customers. That's less than 5% of the number of customers Comcast has.

It really is by design meant for people in areas that do not have terrestrial broadband.


This is exactly right!

I think that 5M is a high number (at least with the current number of satellites and their architecture), specially if they do not apply data caps.

I would say that 1M customers in the US is a more reasonable figure. If they hit these number, I expect the performance to degrade significantly.


I thought something like 20% of America did not have terrestrial broadband. At 5 million this is a giant change, but not as universal as I originally thought.

I wasn't referring to the speed (although obviously fiber has higher speed and reliability, and that will likely never change). I was referring to the fact that it seems like if you have the population density (in the US, at least in my knowledge) to justify a local ISP rolling out fiber to homes, you have a population density that will oversubscribe even the completed Starlink constellation (locally). Starlink just is never going to work in NYC or SF or areas with significantly less density there. Although it may work for a lucky few who manage to enroll before the area is full and locked down.

Some low-density parts of the US have fiber as well thanks to programs like https://www.usda.gov/reconnect

Sure, but it's some for a reason. It's a generality. So is "if you're shopping in Goodwill, you're not driving a Ferrari". Obviously, the recently broke, extremely thrifty, people looking for a collectable or to reclaim something accidentally donated are exceptions.

Maybe that program has covered more than enough people to make it worth talking about, but I didn't see anything like that.


I’d fix the Comcast problem before switching over to something that is pretty much guaranteed to have drop offs at this point in time.

There are also devices that will give you a cellular LTE backup like the U-LTE from UniFi. I’ve heard the plans are a bit pricey.

https://store.ui.com/collections/unifi-accessories/products/...


Yeah we have mifi’s as backup as well but usually only a few people can be on those. But also thinking about dedicated fiber… but honestly none of these seem as cool to me right now as bunch of satellites

For ssh you could use tmux or screen in case you get disconnected

I have five different Wi-Fi APs on my property. The handoff between them results in roughly 40-60ms dropout. That still interrupts audio calls. 1-2 seconds is huge.

Yeah, I am on lots of video and audio calls for work. 1-2 second drops while doing a client presentation are absolutely not ok.

I’m a multiplayer developer and I’m waiting to get a Starlink to test on (I have tested a few times on my moms). I want to makes games tolerate of the latency and instability of Starlink. Mobile networks deal with similar issues so I think it’s possible to make games with Starlink in mind.

I believe these kinds of latency and dropped packet issues be easily simulated on a normal PC without the need for a Starlink connection

Most tools for simulating latency and packet loss locally are a pretty poor model of the kinds of behavior encountered on real networks. In particular, latency due to bufferbloat rather than speed of light delays is extremely important to simulate, but not well-supported by many tools.

I do simulate these conditions, but there’s no replacement for real world condition testing.

exactly. Just use charles proxy (or mitmproxy) and ask it to drop packets/throttle the connection. No need to wait for starlink.

I’m not sure if it’s an endeavour worth taking honestly.. Starlink is still in beta, and once they can use starship to launch the satellites (hopefully somewhere next year) the launch capability will increase to 400 satellites compared to 60 on falcon 9. At that point they will be limited just by the satellites build throughput, so they will reach the 11k coverage very quickly.

I’m already getting a Starlink and I have one available to test on at my parents house so I think it’s worth it. I’m also evaluating it as a way for me to move closer to my family. I’m stuck in Portland because I require really good internet as I’m a 100% remote game dev working on multiplayer games. And the only option in Albany is really bad and expensive Comcast service.

What about an SSH connection? Do you have to reconnect when it switches satellites?

No, the connection seems to be paired up through the ground station, so I wouldn't get disconnected via SSH. Mosh may be a better option if you want to make it feel rock solid though.

> They had 1000 or so satellites when I first started testing, and there are now something like 1600 or so. Most of the time, I don't even notice when it switches satellites.

Enjoy the early adopter moment. Even if they keep increasing the numbers, they will probably move those new satellites in a much wider net to cover more subscribers the second they must show a profit.


They are planning to increase the number of satellites by an order of magnitude. They’re launching extremely fast with Falcon 9, and prepping Starship for launch as well, which has 5-10 times the payload. May see Starlink launches on Starship join Falcon9 within a year or so.

So the opposite is true. They’re likely to massively increase the number of satellites.

If you want to argue the per user bandwidth might be different than for early users, that’s somewhat more plausible. But the number of satellites will increase. They can’t actually significantly change the inclination of the satellites once launched as it takes an insane amount of propellant, and even for solar electric thrusters, so your concern about them moving the satellites to other orbits is very unlikely.


No. They are increasing satellites by an order of magnitude, but users by about 20x. It's going to get worse

If they have 1600 satellites in orbit but are adding 40,000 more satellites (with more capability), then it’ll get better with 20x users, not worse. But anyway, the question was about satellite visibility, not bandwidth per user. The person I was replying to was claiming there will be fewer visible satellites as they’d be moving to different inclinations, which just isn’t feasible let alone likely.

> Even if they keep increasing the numbers, they will probably move those new satellites in a much wider net

The only way to make a wider net whilst simultaneously adding satellites would be to raise the altitude of the satellites, which I am fairly certain is not possible for the existing satellites.


>but not emphasized in this review that connection dropouts happen every few minutes for a few seconds

I don't see this at all. I have constant uptime monitoring, and connection drops are now a minute or two per week. We use it for VoIP, there is no cellular coverage at all where I have it deployed.

Edit: Also, I mentioned this in my fuller main comment but this is around the 45th parallel in New England, and around 1500' (500m) above sea level. This location is also within approximately 50-70 miles ground level of two separate Starlink ground station installations. The nature of Starlink is that there is much more of a geographic component than most people are used to in a WAN link, so it's probably important when talking experience to specify rough area of the world one is in. Once the network is completely built up that may not matter much anymore, but at this point there are definite coverage density differences, and with the current bent-pipe usage ground stations matter too more. Anyone interested in getting an idea of current planetary station and sat deployments might find this site interesting:

https://satellitemap.space/


As a user of Starlink for more than 4 months - the quality has improved. While you say it is unusable for video calls, i think that is way overstated and it completely depends on where you are trying to connect.

Compared to the other options which were atrocious (10 MB down max, 3 MB Up max, weather changes everything) - the hiccup you get maybe every 10 minutes for 10 seconds - is annoying but not a deal breaker for VOIP calls. If you are doing client side calls maybe a deal breaker - team calls manageable but annoying. Also I do calls with our Australian team (and were North America based) and they are on cable internet and they get hiccuped in the same amount. So actually I would say that Starlink is on par if not better than their connection.

If you are comparing the internet to city quality cable then yeah not comparable - but thats not what they are targeting. They are bringing remote areas online.


To be honest, one-on-one Facetime calls and Zoom are almost perfect now, with few bits where it would pause and come back. Group calls were even less of a problem, because we're all used to one or two people having connection issues, and it's easy to work around that.

Yeah, different call quality requirements I think. I am mostly on client calls and I have a ready alternative to be on fiber. If I head out to my place in the country and have crappy call quality as a result, that does not go over so well.

That was the feeling a few months ago. But in the last 2 months, we have used zoom and other applications with very little drop outs. I have various TV stations running for hours at a time at high resolutions, and at the most there is a freeze frame for a split second once or twice every few hours. Frankly, I have had more issues with all the other internet connections we still maintain, than with Starlink. (Zoom was always dropping out with the others.) For example we have: (Slow) high speed DSL, (slow) high speed lte. In all those the internet download speeds are very variable, start fast (5-12) dropping to 0.5-1 and going up and down over time. Starlink maintains over 20 down, going up sometimes to over 30. We are in the countryside in Canada, so true high speed doesn't exist. For now, Starlink is the most dependable high speed option. (What will happen when more subscribers will join all on the one satellite?)

Zoom handles intermittent dropouts better than anything else I have seen and I don't quite understand the whole mechanism.

When you lose a connection for a few seconds, maybe even 10 seconds, the video will pause, but when your connection reconnects, it continues where it paused. So you don't miss anything. I believe at some point when the speaker stops talking, like waiting for a response, it will jump cut their video to a more live feed again. So that you don't get too far behind.

I'm curious if anyone here knows about how this works and if it is common practice in live video chat?


WebRTC adjusts playback speed to accommodate variable network latency. Here's a paper which describes it a bit.

https://www.isca-speech.org/archive/PQS_2016/abstracts/19.ht...


Yes! I had noticed this - cool to read more about it.

As Geerling points out, he has substantial obstructions. I have no obstructions, and see a few seconds of downtime per day. My wife and I regularly have multi-hour zoom calls with no problems.

Obstructions are a problem, but users with no alternative are much more motivated to locate the dish appropriately.

Starlink is not for people who have gigabit wired connections. For those of us who were lucky to get a hotspot to work long enough to use our 15Gb cap, it is a godsend.


It works with video calls, I’ve done it often.

Occasional blips occur, but the call isn’t dropped (and this is with some tree obstruction).

So it’s worth just getting it if that’s what you want it for. It’s probably 10x better than your existing connection.


My mom has been using Starlink in Albany, OR for 2 months. It started spotty, but now works better than her other Comcast connection. The only issue is that every day at 7ish they lose connection for about 5 minutes. She’s said it’s not been a huge issue and plans to cancel her Comcast at the end of her contract. She works remote and Starlink has been great for video calls and video streaming. She’s getting 30-40 down and 20-30 ms latency; Comcast is 20 down and 25 ms latency.

I use it for zoom daily. Yesterday was the first day in awhile where I had difficulty completing a call. The handoffs now last only about a second or two. Previously they’d be 15 or so seconds but that hasn’t happened for over a month. It is my daily driver though I do have back up DSL just in case.

> It's implied that this is due to the trees obstructing a full view of the sky, but I have actually heard these connection dropouts are just about universal due to the constellation not having enough infill.

If you go to https://satellitemap.space/# and enter in your GPS location in the settings (45, -90 for a rural northern Wisconsin as an example), and you can see the satellites that that location has visibility of.

And there are times when there's nothing in that area of the sky.


https://starlink.sx/ can generate a coverage prediction chart for you

> "It's sort of mentioned, but not emphasized in this review that connection dropouts happen every few minutes for a few seconds."

He mentions that this is due to the Starlink dish's view of the sky being partially obscured by trees.

This will presumably improve as the number of Starlink satellites grows, as it will be more likely that there will be an unobstructed satellite in view at any moment, and less frequent switching between satellites.


And he also mentions that if this were his only Internet, he would trim the tree branches to get a better view of the sky. That's not what I'd do though; I'd put the satellite dish on a radio mast. Anyone who's ever done any kind of radio work knows that antenna height is everything, and in his specific case it's important for a different reason: avoiding line-of-sight obstructions.


I mounted a point-to-point link in a tree at either end.

1. Trees sway. It did matter for the PtP link in my case.

2. Installation and maintenance took time and it was dangerous.

3. Leaves and branches grow. Obstruction and misalignment over time were problems.

Advantages: it did work, and it was cheap and fun to do. I probably could not have installed a tall enough mast (too ugly to be accepted).


Repending on height of the mast, it could be more expensive than the rest of your setup combined

trimming trees is a recurring activity, it might be expensive over time depending on type of trees and access etc

Yeah holy shit. The dude has obstructions and then writes a review about it? It's like putting bad gas in your car, then writing a review about how bad it was... Admittedly getting rid of the obstructions was a TON of work. It took me a few weeks of moving my dish around, and fiddling around with different mounts and options. I even topped a tree. But once I got rid of the obstructions, the service improved dramatically. Now with the launches that we have had of new satellites and other updates, I don't see hardly any drops at all. The most common annoyance with mine right now is the router/dish crashing. This happens maybe once every two weeks. Not much worse than the shitty Comcast-supplied modem. My next step is to use my own router and I think that'll take care of a lot of it.

Partial obstructions are not going to be rare for customers. While not a best-case review, this is arguably a more authentic review.

I get the crashes every week or so too and I use my own router. I think it's the firmware in the dish.

Had a team mate on Starlink and they would drop from our zoom call essentially every single time.

Maybe your meetings are just boring

How long is that ago?

Dunno, I did a ~20min FaceTime call that was essentially flawless. That call ended with me being very impressed by the stability and bandwidth of the connection. Far better than the satellite service at the location previously. Maybe it was just pure luck? Not sure, but the consistent 90-100mbit downstream and 2-digit ping is pretty impressive.

> It's sort of mentioned, but not emphasized in this review that connection dropouts happen every few minutes for a few seconds. That makes starlink fine for any kind of asynchronous content like web browsing, torrenting or video streaming, but unusable for video calls, stream hosting, voip, or online gaming

Yep, been our experience as well. We've got a few of us who wanted to trade from our cottages and its just unusable if you need a continuous signal for more than 10 minutes at a time.

That doesn't mean its useless, just that its not usable if you want to do voip, trading, video calls etc.

Hopefully they'll figure out what causes drop ever few minutes at some point. But currently given how expensive it and the hardware are its a very disappointing product.

I guess we're just spoiled now a days with the 1Gbps wired internet that most city homes have access to.


If you’re getting drops every few minutes your view is obstructed. That’s about the time it takes an LEO sat to cross the sky so your dish probably can’t see a critical region in the orbit.

When these dropouts occur, is the IP address stable?

Yes. We have the drops also. But they are happening less, and for only a blink. In the worst case, we experience a freeze frame. Most of the time can barely see the freeze.

I think it's important to remember that "connections" in networking aren't actually connections. Everything is sent one packet at a time over networks that aren't presumed to be stable.

That means the entire system is resilient. A 2-second pause in connectivity usually won't mean the app dies. It means the application presumes it dropped a few packets, which it did.

Now, if you're playing a very fast-paced multi-player game, a 2-second lag at the wrong moment can spell disaster. But most video call programs can easily handle a 2-second blip in connectivity, annoying though it is for participants- I see those all the time on my non-Satellite based internet.


Obviously latency is important in synchronous use cases like video calls, however I wonder if a delay for slower one on one discussions would be all that jarring for users

For the remote areas this is intended for, it is already a 10x or more improvement. Great internet service is better than perfectly reliable slow internet.

The dropouts are not universal. Results in the Starlink subreddit make it quite clear that you’re either obstructed or maybe have bad dish.

It may be that anyone who is willing and able to do so has already thought of it, but on Linux you could multiplex the connection with LTE/3G/dialup and probably get pretty good results.

The outages have been getting better recently. They are supposed to go away entirely once the first constellation is fully complete. If you don't have obstructions, that is.

I would expect that won't stay a problem for long, they're still pouring those things up there.

I hope that the problem really is due to lack of infill because that means it'll be temporary.

> It's implied that this is due to the trees obstructing a full view of the sky

Why is the antenna on the ground and not up on a tall mast?


It certainly can be on a mast; I put mine on a (short) steel tube mast, and it's been a great improvement on service that was already a vast improvement over Hughesnet-provided satellite.

He mounted the dish on his roof. There are still some tall trees in the way of the dish.

Here's his YouTube video of the installation:

https://www.youtube.com/watch?v=ynsCVOz7jv4


Mine is mounted near the top of my roof's ridge. The problem is there are 8 trees (five of them more than 75') around my one-story house.

> That makes starlink fine for any kind of asynchronous content like web browsing

This sounds terrible for web browsing. Last thing I want is to know is to submit a form and then have my connection drop out in the middle. Imagine being in the middle of filling out an application or opening an account or verifying your identity or something like that.


If the connection drops for less than 5-10 seconds (this happens even on my Cable Internet sometimes), it's no problem. Most timeouts and TCP connections are okay with complete dropouts for 30 or 60 seconds (sometimes longer), as long as your local LAN doesn't drop your network connection.

Doesn't that depend on the nature of the "drop"? When it's just lack of signal then sure, but when I see connections break even momentarily (on cable...), there's often some sort of feedback (I think sometimes it's a connection reset?) that causes the browser to just error immediately, and then I have to reload the page... despite still being on the same network with the same IPs and such.

That is because of the way your connection is being dropped/what your router is doing. It’s sending reset messages back to your client telling it your connection is dead, instead of trying to resend your packets - which then succeed when the connection is back up.

That's a possibility in theory; I haven't root-caused this to know what's happening in every case (again: not every drop is the same). Regardless, I'm just talking about what the average end-user might see; I don't really care where the blame goes.

I've had Starlink for about 6 months and it is a massive improvement on upload speed at 10-25 Mb. Download speed is a mixed bag wildly oscillating from 5 to 100 Mb and back. It's okay for downloading things but it's terrible for any sort of video conferencing. There are brief dropouts on average every 6 minutes or so and my obstruction map is better than the author's. My neighbor up the street got slightly better service by mounting his dishy on a 20-ft antenna pole above his house.

Local ground service is 20 Mb download and 2 Mb upload. And that's just barely sufficient for watching streaming video and video conferencing. Gigabit service is but a mile and a half away but no one is going to pay to lay the fiber into our neighborhood. So the last mile and a half is copper from 20 years ago. I think that's going to require political will to fix and I don't think that political will exists right now nor will it in the near future. We could have paid $5,000 per house to lay it ourself but our own neighborhood couldn't come to consensus on that. Now imagine that at a national level.

So if they just deliver 100/100 within a year or two, this is an epic win IMO and I will cancel ground service. And if they don't someone else will so I'm not worried. But it took Teslas to spark the electric vehicle industry. Now there's a lot of choice. I wouldn't be surprised if something similar happens here.


> Gigabit service is but a mile and a half away but no one is going to pay to lay the fiber into our neighborhood. So the last mile and a half is copper from 20 years ago. I think that's going to require political will to fix

I imagine once Starlink is an actual choice, the telecoms will install wireless at the end of the fiber and offer you faster, more reliable service than Starlink.

They don't do it now because they get your money without having to do anything at all.


Starlink isn't driving the telecoms to install wireless. The telecoms were already planning this before Starlink.

T-Mobile Home Internet is already available to 30M households in the US (out of around 130M households so around 20-25% of US households). T-Mobile is looking to have 7-8M subscribers within the next 5 years which would make them the 4th largest ISP (behind Xfinity, Spectrum, and AT&T). Verizon is looking to cover 50M households by the end of 2024 which is years away, but shows that 5G home internet is coming.

Wired home internet companies aren't avoiding installing a wireless link at the end of their fiber out of spite. It's a combination of who has wireless spectrum and the technology/capacity available. If you're talking about Xfinity or Spectrum, they don't have the wireless spectrum to offer that. If you're talking about Verizon/AT&T/T-Mobile, they're working on it, but it takes time for the technology and spectrum to be there to provide the capacity people expect for a home internet connection. The recently concluded C-Band auction means that wireless carriers are going to have more spectrum available to provide more capacity (and they spent nearly $100B getting it). 5G NR provides more speed and capacity.

If it was just out of spite, Verizon/AT&T/T-Mobile would have been offering wireless home internet for years in areas where they had no wired network. They weren't getting the money in places where they didn't own the local telco (which for T-Mobile is everywhere and for Verizon/AT&T is most places). Even when they owned the local telco, most people would be buying cable internet.

The problem is that home internet isn't easy. A wireless customer probably uses 10GB of data on average. Streaming HD Netflix is 2-3GB per hour. Home internet usage is usually an order of magnitude higher (and can be even higher than that). Basically, you need to increase your network capacity by at least 10x if you're going to be signing up home internet customers. With new technologies and spectrum, that's what wireless carriers are doing over the next 1-5 years.

I think that terrestrial wireless will be big in the future, but it's not because of Starlink putting pressure on telecoms. It's because their networks are going to be seeing massive capacity upgrades over the next few years that will enable it. Verizon/AT&T/T-Mobile would have loved to offer wireless home internet years ago, but the technology and capacity simply wasn't there. I mean, they did offer home internet years ago, but it often cost hundreds of dollars a month and was only available in really rural areas (not just places that might hate 20Mbps service). But new tech and capacity gains are allowing them to offer new service. T-Mobile is first out of the gate because it got new spectrum earlier, but Verizon and AT&T will be following in the coming years.


I don't believe that TMobile home internet is available to 30 million households. The only numbers i can find say 30 million people. Some announcements from T-Mobile say 20 million households.

But it appears to only be available to people in mid sized cities. In other words, people that already have somewhat acceptable internet service options.


I’m in a rural area (the nearest town’s population is about 1,000) and I know someone that has T-mobile’s home internet, fwiw.

T-Mobile's home internet is cellular. 4G/5G. And I've heard it's pretty good from a friend that runs a WISP.

I personally don't hate 20 Mb service and I don't think it is the problem except for streaming video.

The problem is the metastasis of ridiculous dynamic content that is driving up bandwidth requirements and latency and delivering a reduced quality Internet experience in return, not because dynamic content is intrinsically bad, but because of how it's being utilized.

But in both situations I've lived through recently, the fiber has been just out of reach along main roads for some time now. And I know a bunch of people in similar situations on side streets where AT&T and XFinity declined to extend service after building the infrastructure along main roads to do exactly that.

Anecdote: during 2019's power shut offs because of high winds, my house lost power for 3 days, but my neighbor 25' from me did not lose power. Seems like a similar situation in some ways. Both of our houses were on hills served by separate power lines less than a quarter mile from a main road. The only qualitative difference I can think of is that the last 200 ft to my house is below ground.


I live in a community, where Verizon FIOS (later sold to Frontier) put down fiber in the ground, and then didn’t connect houses! So this fiber is there for 10 years now, unconnected…

I imagine the more immediate threat is 5G internet. But currently the reception in this neighborhood is terrible for T-Mobile (despite their coverage map insisting otherwise) yet one of my neighbors claims he can get 50 Mb download with an AT&T hotspot. I have approximately zero faith in our current ground-based providers. They are the PG&E of broadband IMO.

Upgrading to my first 5G phone recently made voice service work at my house but did nothing for data.


With an external antenna aimed correctly I'm sure you'd get a much better cell signal. That may be what your neighbor has.

Keep in mind T-Mobile and AT&T are entirely different providers. AT&T coverage may be better where you live.


There are too many places that don't get 5G in the first place.

Yet unlike ground-based service in my experience, the situation is improving with time. It took 15 years for my old place to go from 2.4 Mb/300 kb to bidirectional Gb and that only happened because AT&T was feeling generous briefly in 2019 w/r to expanding service (in that case, 400' of fiber to my cluster of houses). They stopped expanding shortly thereafter because so many people were in exactly the same situation that they were swamped with requests.

I get around 48-56Mb down tethered to my 5G T-mobile phone. (in the north Seattle area)

Same latitude, lucky to get 2 Mb in either direction. But down in Roseburg, Oredgon I got 135 Mb/s download sitting in a parking lot so it's not my phone.

Edit: I just walked through the neighborhood repeatedly invoking speed test and I can get 50 megabits at various points. But I can't get it anywhere on my own property. There's no real rhyme or reason that I can see as to where it's good and to where it's bad.


> And if they don't someone else will so I'm not worried... I wouldn't be surprised if something similar happens here.

I'm a little worried about it. Other than Amazon's Project Kuiper, I don't really see who else can compete.

Unlike EVs there's not an existing industry that's doing _mostly_ the same thing, that just needs to start offering a new line. Running a LEO megaconstellation is extremely different than putting up a handful of massive geostationary satellites. I just don't see the existing satellite internet players being able to compete.

Frankly, it's a massive investment of resources to get started (which is true for auto manufacturers as well), but doing it economically requires launch costs that only SpaceX is currently able to provide.

Nobody else wants to build a competitor to Starlink by paying SpaceX. Which is why I think Amazon's Kuiper is the best shot to compete—but it requires Blue Origin's New Glenn to come online to really get the economics working. I know Amazon has tapped ULA as a launch provider in the interim, so they'll be able to start getting satellites up and running soon, but the economics seem...painful if they're going to use expendable rockets for the bulk of the constellation.

It's definitely possible that we see a flourishing of offerings, but I could also see a world where in 5 years time there's really only still Starlink.


>Nobody else wants to build a competitor to Starlink by paying SpaceX

I'm not sure what you mean here. Outside of, say, military usage, every single telecom company is a competitor to Starlink. Internet is internet.

This commercial product is turning out to be what the "naysayers" thought it would be; brilliant if you're in a place with terrible internet, but noncompetitive in any kind of urban environment.

Again, imho, the huge thing here is the ability to get a quick internet set up anywhere in the world that can't easily be taken down (...military).


> I'm not sure what you mean here. Outside of, say, military usage, every single telecom company is a competitor to Starlink. Internet is internet.

With regard to satellite Internet mega-constellations, it is true thus far. The two other major constellations (OneWeb and Kuiper) have avoided SpaceX as a launch provider, choosing competitors instead – OneWeb has gone with Arianespace Soyuz and Virgin; Kuiper has chosen ULA (and is assumably going to choose Blue Origin too once they are ready). SpaceX has said they are happy to launch competitor constellations, but it makes sense that its competitors aren't happy to fund their competition. It is very unlikely that Arianespace or ULA can actually beat SpaceX on price, so avoiding funding your competition is the only logical explanation for OneWeb and Kuiper's decisions.

SpaceX is launching other satellite telecommunications systems, that obviously also compete with Starlink to some extent, but less head-on. For example, they are contracted to launch one of the ViaSat-3 satellites. ViaSat-3 is a satellite Internet constellation, so in that sense does compete, but it only has 3 satellites in geostationary orbit, so it isn't really going after the same direct market as Starlink. (ViaSat appears to be getting scared of Starlink, as evidenced by their anti-Starlink applications to FCC and threats of a lawsuit; given that, I wonder how much longer they'll be willing to buy launch services from SpaceX.)


Is OneWeb still in the game? I saw that they had been financial issues and were sold to the UK and I kind of assumed that they had ended their constellation goals.

It’d be great if they were still working on it, though


They are still in the game, but not quite the same game.

They have done 8 launches so far, including 4 this year, with another 5 scheduled this year and then more next year. At the moment they have 248 operational satellites in their constellation. Their goal is 648 satellites, which is gong to take them (approximately) 11 more launches.

But 648, is quite small in scale compared to Starlink's plan for an initial constellation of 12,000, with an additional 30,000 currently going through the approval process. Starlink already has more working satellites in orbit than OneWeb's planned full constellation size.

Given that, it is difficult for OneWeb to compete with Starlink head-on, so they are basically ceding to Starlink the B2C space and going after B2B instead. I think with UK government and UK military they are going to have some success, since the UK will likely prefer OneWeb given that it is a UK-headquartered company with partial UK government ownership.


Well, that’s great! I’m glad they’re still competing in the space and have a good foot-hold.

I think competition in this space is important in the long run.


> I'm not sure what you mean here. Outside of, say, military usage, every single telecom company is a competitor to Starlink. Internet is internet.

That’s certainly true, but there are real physical limitations that make it hard to compete with Starlink without a LEO constellation.

Sure, you could do ground based infrastructure build outs, but the population density means it’s not profitable or you have to charge really high rates to make up the fixed costs. Other satellite providers can compete in the rural areas, but Geo and LEO are have important differences. The first is latency (internet is internet but 60ms internet is not the same as 600ms internet).

When I made the statement I did, I was mostly referring to satellite internet providers, because land based providers have largely abandoned or drastically underserved this market segment.

I don’t see a GEO constellation being able to provide a similar internet offering to Starlink ever, due to light delay. I don’t think ground based solutions will pop up to start competing for these low density market, because they’ve already been written off as unprofitable even before there was serious competition for the market.

> This commercial product is turning out to be what the "naysayers" thought it would be; brilliant if you're in a place with terrible internet, but noncompetitive in any kind of urban environment.

Well that’s probably because the naysayers were predicting the exact same thing as the supporters. Spaced has described Starlink consistently as being designed to service rural and suburban areas that are underserved by existing providers. The goals never were to compete with providers in super high density urban markets.

I’ve been a huge supporter of Starlink since the first announcement, and it’s shaping up pretty much how I imagined which also lines up with your description.


If you want to build a competitor to Starlink that does the same thing and is ideal for terrible internet areas, you need to put up satellites, and SpaceX is the only company that you can pay to do so economically.

A bit of a tangent, but is there any information about how many constellations are physically possible? Obviously there's a lot of space in LEO, but it's still limited. Given that each provider would need their own very dense constellation of satellites in different orbits, things would eventually start getting crowded.

I don't think we'll get there any time soon (or ever), but kind of an interesting thought experiment.


I haven't done the math and I'm certainly not an expert, but my lay understanding is that the number of constellations is much more limited by spectrum frequency availability, rather than the space constraints in LEO.

That is, if we had an very large number of companies wanting to create satellite internet constellations, we would run out of useful EM spectrum before we would run out of LEO orbits.

I'm not positive about that, though!


Yeah, that occurred to me as well, though I don't know much about how the satellites communicate with eachother. I imagine each constellation would need its own band, but not sure how wide it needs to be? Intra-constellation communication could feasibly be done with lasers, in which case, all they may need to worry about is the Space-to-Ground leg. So maybe it's not too bad? I don't really know, but certainly interested in learning more...

You’re underestimating the amount of space in space (sorry, couldn’t resist). Volume increases exponentially as diameter increases.

I don’t think anyone would worry about 1800 or even 180000 things spread over the entire surface of the earth. We have billions of cars, and they mostly avoid running into each other. Even 100 miles up? Far, far more space - and a lot less ‘stuff’ to worry about too.

They could pack a million satellites up there and basic traffic management would be more than enough to avoid collisions.


We will see.

But it's definitely don't comparable to cars. Say you want safety margin of mere 1 second, that means reserving 8 kilometers free corridor in front of every satellite. At all times. That isn't a small volume.


Why one second and 8km? Satellites are not manually maneuvered.

Airplanes are only required to keep 1,000 to 2,000 feet vertical separation. They travel at much slower speed than satellites (550 mph vs. 17,200 mph), but they are also much bigger. They tend to travel in narrow corridors, as opposed to being spread out like satellites.

NASA and SpaceX have started working together to address the collision issues with their fleet:

https://www.nasa.gov/press-release/nasa-spacex-sign-joint-sp...


They are not manually maneuvered but there's still unpredictable inteplay of solar wind and tenuous atmosphere to correct. And some sats will fail, the environment is harsh. So I believe some safety separation will be a must, used 1 second as a rule of thumb and the orbital speed is 7.8 km/s at the altitude.

I suspect more safety time will be required. Say a micrometeoroid impact happens causing one satellite to go off track. Now the controller computer needs some time to recompute whole swarm trajectories to minimize debris impacts. Even sats that are currently on the other side of planet might need to start thrusters immediately, as it takes only 10 minutes on average to cover the distance.


Let’s do some math and see. If we assume the earth is a sphere (I know it isn’t, but it’s mostly close), and using 4(pi)r^2, it’s about 511 million km2 in area on the surface (counting oceans, etc). Double checking, it’s actually about 507 million km2, so decently close.

4 * 3.1459 * (6378^2)

Starlink is orbiting at 550km altitude (approx). At that altitude, we’re looking at…

4 * 3.1459 * (6928^2) = 603,977,365 km2

The thing to keep in mind about orbits is the defining factor is your eccentricity and velocity. Faster velocity == higher orbit, assuming it’s a round/circular orbit, that velocity will also be consistent.

As long as everyone is going in the same direction and trying to maintain a circular orbit, you can stack orbits ‘on top’ of each other or put satellites quite close to each other just like planes do with flight levels. So even that number is misleading, as you could 10x that if you had decent traffic control.

Even without that, you’re talking a density of 1 satellite per 600 square kilometers of space with a million satellites for a given ‘flight level’.

The velocity matters a lot less than the relative velocity to other things you might be nearby - think of it like two cars going in lanes next to each other on the same ‘direction’ of a freeway, vs a free for all demolition derby with no traffic control, lanes, etc.

If everyone is co-ordinated you can have problem free high density. If random stuff is happening, including people potentially trying to hit you - it’s going to be a mess. If everyone just does whatever and never co-ordinates, yeah someone is eventually going to run into someone else - but unless there are a LOT of satellites, the odds are low.

With it being LEO, debris and dead satellites are also going to be a aerobraked pretty quickly.


If it's all so easy peasy and every satellite has zillions of square kilometers at disposal, why do we have collisions and required maneuvering already? And "dead satellites are also going to be a aerobraked pretty quickly" - at 550km this means months. That's definitely not quickly enough.

1) That happens INCREDIBLY rarely right now

2) Many of these incidents are from people are also putting things in Lagrange points and Geosynchronous orbits, and similar 'crowded' areas of space - unlike these constellations. Think jostling for space in downtown Manhattan, vs being fine having a place in Kansas somewhere.

3) Right now it's mostly a demolition derby out there, with only minimal traffic handling/control and the occasional 'haha, I got you' intentional satellite explosion/missile test.

As noted, even with all those factors, it is still incredibly rare. Only a tiny handful of problems have been tracked to orbital collisions with anything man made, and that is with us putting all sorts of things (including 'off the books' military top secret satellites, nuclear reactors, random bits of micro satellites, etc.) up there since the 50's.

Months is pretty quick for this kind of thing - and remember, unless it gets energy from some kind of serious-delta-v impact (or other delta-V adding type of event like a high energy explosion), it will lose velocity and DROP in altitude if the orbit remains circular. If the orbit doesn't remain circular, especially at LEO, it will have to dip significantly into the atmosphere to get to a higher altitude at any point in it's orbit, meaning even faster degradation and burn-up.

The reality is that it's huge up there, and if we do even basic thinking and planning on this to not do something obviously dumb (like tangential, perpendicular, or opposite direction orbits all over the place at the same level), it isn't likely to be a significant problem; just like it hasn't been a real (as in significant probability event) problem so far.

Cosmic rays, high potential static build-up, magnetic storms, micrometeorites, etc. are just as big or more of a problem. They rarely cancel a mission or damage anything notable, but it isn't zero risk there either.


Yeah, that's certainly a good point.

The analogy with cars is a bit tricky though as we have well defined rules that help us pack them very closely together on the surface. For example: roads, lanes, controlled intersections. Those things aren't quite the same when talking about orbits because cars and satellites aren't spread out the same way.

The other commenter made the point about some safety margin, which obviously has to be a part of the equation. The speed and error of orbits would play into whatever that number comes out to be.

Regardless of there being lots of space, there's still some limit, even if it's 1,000 constellations.

Some variables I don't quite know:

1. Acceptable altitude range for constellations

2. Typical positioning error of a satellite

3. Average failure rate and failure mode of a satellite

   - ex. does a dead satellite automatically deorbit itself or do we need to wait for the orbit to decay? I'd imagine decay at that altitude is fairly quick. Would a decaying satellite in a higher orbit need to be accounted for?
4. Required satellite density in orbit and the number of unique orbits, their arrangement, etc.

I'm sure there are other factors, too...


>Volume increases exponentially as diameter increases.

Cubically.


Thank you for the correction! That is indeed correct.

But it is not the same thing, due to the speed that satelites travel.

Satellites aren't going to swerve, though.

The limit is far more how tight you can beamform the signals and how much usable spectrum there is: this is the fundamental thing limiting the density and hence number of users starlink can have.

HughesNet and ViaSat have both been around for a while but their rates and their service are terrible. But unlike Starlink, their service is reliably and predictably crappy so there's that I guess. There's nothing stopping them from improving their own services.

> There's nothing stopping them from improving their own services.

Except the laws of physics I guess. Geosynchronous orbit is about 36,000 km out so best case scenario if you were going straight up to a satellite in such an orbit and back down even at the speed of light in a vacuum that’s a lower bound of a quarter of a second of latency. So unless these legacy satellite internet companies launch LEO constellations of their own, which there’s no indication they’re capable of doing, it seems pretty hopeless for them.


I don't know when the project started, but ViaSat has been working toward constellation service of their own for at least 6 years: https://www.viasat.com/space-innovation/satellite-fleet/vias...

ViaSat-3 is a geostationary constellation of 3 satellites, which I don't really consider to be a competitor to Starlink due to its physical inability to ever reduce the latency lower than 500ms

> There's nothing stopping them from improving their own services.

Well, they can’t improve the latency of their service due to light delay. If they want to compete on that front, they need to move to a LEO constellation.

Neither HughesNet nor ViaSat has started working on LEO constellations (publicly anyway), and these things take years. If they want to get into LEO constellations, they’ve ceded a 5 year head start to SpaceX.

I’m not optimistic that either of them will be able to remain competitive with Starlink over the next decade.


Yeah, it'll be interesting to see, although Viasat-3 will be 1/4th the cost of Starlink (on a $/Mbps/month) basis and Viasat-4 will be about 1/15th. Even launching for "free", it's gonna be pretty tough for SpaceX to meet those numbers. The GEO satellites also require simpler and cheaper ground infrastructure, both on the ISP and on the customer side.

Oh, interesting, I hadn’t seen those numbers. Where did you see the ViaSat 3/4 pricing? All I could find was the current offerings (which are more expensive), but I didn’t look hard.

I’d love to dig into the other offerings that the GEO players plan to provide. My dad was a Day 1 sign up for the Starlink beta, because it’s dramatically better than any offerings in his area (including existing Satellite), but I’d definitely like to share any better upcoming options with him.

I think latency might be a deal breaker for him, because they do a fair amount of video calling and don’t like the audio lag


Ah, those numbers are from the Viasat earnings calls and some rough estimates. Each Viasat-3 satellite will have about 1 Tbps capacity and have a lifetime of 15 years. I don't know about cost, but Viasat-2 was in the neighborhood of $300e6, so, I assumed the same for Viasat-3. Viasat-4 is claimed to have a capacity of 3-5 Tbps and I assume a similar lifetime.

Now, these numbers are all well and good, but they don't include the cost of the ground system and the terminals, and I don't know what those will be. But, I would expect them to be cheaper, most crucially because GEO gives you more flexibility in where you place ground stations (vs LEO) and once you aim the dish, you pretty much never have to move it (LEO dishes spend their whole life slewing to match the satellites; it turns out that requires some parts that aren't so cheap.)

Anyway, I don't know what Starlink or Viasat's pricing will be in a few years, but I think Viasat will have a competitive edge in terms of overall cost.

Latency is definitely an issue for some users and some applications. If you watch what Hughes has been saying for years, it's that the future is hybrid, where a terminal uses both LEO and GEO, for what they're best suited for (i.e., interactive traffic over LEO, big, cheap bulk bits over GEO.) If you Google a bit, you can see the Viasat press releases alluding to this same capability. I don't know if/when that tech will make it to residential service, but I would expect that if there's unmet market demand, they'll get it out there.


> But it took Teslas to spark the electric vehicle industry. Now there's a lot of choice. I wouldn't be surprised if something similar happens here.

I really hope not. There are enough satellites out there, there are already too many, space pollution is real. Maybe we can find a better solution to this particular problem.


I think these LEO constellations are actually the least concerning part of space pollution.

There are many more LEO satellites, true, but their orbital placement is such that the satellite's orbit naturally decays in a few years. Which means if a satellite becomes uncontrollable (or is destroyed in some way), the debris will clear in a relatively short period of time.

I'm actually much more concerned about the much smaller number of satellites in medium and geostationary orbits, where the decay time is decades or centuries.

Failed satellites or debris in these orbits will take a very long time to clear, and strikes me as a much larger concern than the larger number of LEO sats


This doesn't make much sense. MEO is huge and GEO is only "relatively" small because it's narrow along two of three dimensions we care about†, which has no influence on any debris from a collision and thus it would most likely leave GEO altogether.

Would a volcanic eruption in New York be very bad? Yeah, I guess it would, but, that's not going to happen. Whereas California has several volcanoes that - while we've no reason to expect them to erupt this year - certainly can't be ruled out for our lifetimes, so, makes sense to manage that risk not worry about New York.

† A Starlink, or a GPS bird, has a "ball of yarn" orbit, it doesn't really matter which part of the planet it's over at any particular time so long as we can predict where it'll be for the near future. But the whole point of a Geostationary satellite is its apparent fixed location in the sky from a point on the ground. To do that it needs to orbit at the same rate the Earth spins, limiting the orbital radius to a tight band - and it also needs to orbit over the equator, the result is all GEO birds are in more or less identical orbits, just offset in time.


We're looking at two different axes. You're considering "how much of the orbit is occupied". I'm looking at "how long does a dead or destroyed satellite continue to occupy space".

If we have a collision that causes a significant amount of debris in a medium earth orbit, that debris will continue to exist for a very long time, so a wide portion of that orbit will be unusable or dangerous.

> Would a volcanic eruption in New York be very bad? Yeah, I guess it would, but, that's not going to happen.

That's fair, but I think your analogy might fall down on the comparative risks of damaged or destroyed satellites in LEO/MEO compared to the comparative risks of volcanic activity in CA/NY.

Yes, it's true that there is a larger risk of collisions in LEO with the number of satellites operating there. And it's true with the larger number of satellites, there are more risks of a satellite losing control. But that doesn't mean GEO and MEO satellites are without risk. Just last year there was a significant risk that a GEO satellite had the potential to explode due to a failing battery (https://spacenews.com/directv-fears-explosion-risk-from-sate...).

I'm mostly interested in hedging against worst case scenarios, and the worst case scenario for a LEO constellation is much less problematic than the worst case scenarios for MEO constellations.


What is "too many"

> So if they just deliver 100/100 within a year or two, this is an epic win

It's unlikely that they (Starlink or someone else) will offer symmetrical speeds. It's not that they're looking to be mean to you. It's that uplink is harder and people use a lot more downlink.

Even if they dedicated as much wireless spectrum to downlink and uplink, uplink would likely be slower. We see this on traditional cell networks. They dedicate as much spectrum to uplink as downlink, but their cell tower is able to better transmit than your equipment and so the downlink becomes faster. Even with so many more users downloading than uploading (and causing congestion), the downlink is usually faster.

Newer wireless networks aren't going to dedicate equal spectrum to downlink and uplink (using time-division instead of frequency division to separate downlink and uplink). Again, this isn't to be mean to you. It's just a reality that people use a lot more downlink bandwidth than uplink. Starlink isn't immune from that reality.

Elon Musk has already said that they should be able to hit 500,000 customers, but that scaling to millions of customers will be difficult. They're going to have to cap how many people get service in an area and/or put in network management to make sure that some users don't use up all the bandwidth from others nearby.

They're also going to need to focus on downlink capacity to serve what users need. That doesn't mean unusable uplink. As you noted, 10-25Mbps uplink is an important improvement. But wireless internet options (including Starlink) will need to balance that with the downlink capacity users need.

> We could have paid $5,000 per house to lay it ourself but our own neighborhood couldn't come to consensus on that

Over 10 years, that's $42/mo. Over 20 years, $21/mo. That's non-trivial if the solution to your internet woes might just be a couple years away. It's probably one of the big reasons why wired companies won't want to be spending money expanding networks in suburban/rural areas. Let's say that you invest in a network expansion and you expect to make it back over the next 20 years. Then 3 years into your investment, Starlink, T-Mobile, and Verizon are all offering home internet service to your customers. Sure, your fiber network might be "better", but that will only attract some users. Others might get a package deal from their wireless carrier giving them a better price. Now you go from having 90% of households to 50% of households and the investment that you made probably won't work out. For most people, 100Mbps is plenty. Sure, some people love the low-ping times of fiber and love gigabit speeds. People like us here on HN. For most people, they want to be able to use Netflix/YouTube/Facebook/etc. and there's going to be a lot of competition for that market.

> Now imagine that at a national level

Realistically, this already exists on the national level in that we spend billions subsidizing rural connections. Starlink is receiving lots of government money to provide rural internet. I think a big question is whether Starlink is looking to grow well beyond what the government will subsidize and whether government subsidies will flow to other companies more. I'm sure that AT&T/T-Mobile/Verizon are all looking at what rural internet subsidies might come their way as they launch rural home internet.

We do have some political will to fix the situation, but it's a very expensive situation to fix for a lot of rural areas in a wired way. Should people in cities subsidize suburban/rural lifestyles? As a country, we pour money into roads, low fuel prices (even as climate change ravages the planet), rural telecommunications, etc. If every home in your area is on 2 acres of land, it's going to cost more to wire up the place, it's going to cost more to get roads everywhere, it's going to use more fuel to get from place to place.

We do spend a lot making rural internet happen. It's just an expensive proposition. Heck, Starlink is very expensive at a $550 startup cost + $100/mo. That isn't cheap competition to wired internet - and that's after large government subsidies that might end up being $2,000 per user. Starlink has received $900M in federal money and Elon Musk is hoping to serve 500,000 users so that would be $1,800 per user from the government. That's not the $5,000 your service provider wanted to extend fiber, but it's still a lot of money. Plus, Starlink is likely to be getting more federal money in the future (and they might end up serving a couple million users).

There is political will and we've spent incredible amounts of money over many decades and continue to spend even more. It's just hard to serve many rural areas. If one is in an area with 5 people per square mile, that's a lot of wire for very few users. Wireless/satellite might make the most sense since installing one thing could serve hundreds or thousands of users. Even 20-50 people per square mile can be a lot of work to wire up.

While wireless home internet is in its infancy right now, I'd expect it to get a lot better over the next 5 years. As you noted, your neighbor installed a 20-foot mount to get better reception. T-Mobile Home Internet customers are rigging up directional antennas mounted on the outside of their homes to get better speeds. Given that people install satellite dishes for TV, it seems very reasonable that we'll see wireless antennas installed to offer internet service. Again, when something is in its infancy, there are less options and it's less fully realized. But that will change over time.

I think the next 5 years will be an exciting time for home internet. I don't think that Starlink is going to be doing most of the exciting stuff and I don't think we'll see symmetrical connections, but I think we'll see great stuff that will bring better connections to people who need it and will bring competition to the marketplace to prevent monopoly providers from taking advantage of their customers.


>Over 10 years, that's $42/mo. Over 20 years, $21/mo. That's non-trivial if the solution to your internet woes might just be a couple years away.

Except... The value of each house would probably increase by $10K-$15K or so:

"Controlling for speed, homes in CBGs where fiber is available have a price that is about 1.3 percent more than similar homes without fiber."

https://realtorparty.realtor/community-outreach/rural-outrea...

My new neighbors are increasingly tech TLAs as the former generation sells their homes off for 2-3x what they paid 20+ years ago. Median house price in my neighborhood is ~$1.1M ATM.


From that abstract:

> Applying this strategy to a micro data set from England between 1995 and 2010 we find a significantly positive effect, but diminishing returns to speed.

England isn’t filled with million dollar homes.


I think it is difficult to overestimate how much other rural internet providers have over-promised for their federal $$. I know that in my area (northwestern Montana), a variety of ISP's have received rural internet funds, and it seems like a fair bit of money has been spent on very little additional high-speed coverage. There are many towns from 350 - 2000 population that are under-served, even with a lot of federal subsidy (I'm told the local telco removed internet capacity from our town of 400 to provide it to a nearby school system. I know that if I pay $5K to get a telephone wire to my house, I can have dial-tone, but not internet, not because of distance, but because of capacity limits.) Perhaps when those ISP's applied for their grants, they simply underestimated the costs.

Regardless, we do have Starlink, and it is transformative.


Southwest Montana here: your situation sounds much worse than ours. I did have to build my own Wisp to get service to my home but in town there is decent HFC and fiber loop if you can afford it. I'm curious if you know for sure that you have regional backhaul capacity limits. I ask because I hear stories from local people here, including folks who should be better informed because they work in city planning, about limited connectivity. Their picture of the connectivity situation is wildly innacurate, in part because providers don't disclose the details of their installed plant.

My understanding is that the local telco (centurylink) would just as soon not be here. It certainly seems unwilling to make the investment necessary to provide more connectivity. Some smaller communities have limited fiber (Babb, St Mary) but there is none in East Glacier and Browning.

I see. So those towns are still microwave-fed? Around here I don't think we have any towns left that are microwave-fed. Gardiner had backhaul fiber installed a few years back. That was the last connectivity hole I'm aware of in this area.

> So the last mile and a half is copper from 20 years ago

In the UK, BTs 'fibre' rollout is almost exclusively FTTC and then copper phone lines after that. My parents have a phone line from 35 years ago, and the cabinet is maybe half a mile away, but they can get 50 Mbit down (I can't remember upload, maybe around 15 Mbit). The provider rolled this out a lot quicker than ADSL (we never got ADSL2), so I assume it's technically not that complicated.


Using copper is weird.

The New Zealand government invested ~ £750 million into fibre to the premises infrastructure, in partnership with other companies. About 80% of NZ homes have access, with say half of those actually using it, but percentage is growing. Usable plans are about £40 per month. In my city you can’t sign up to a cable connection, even if the house is already wired up for it. I belive you even have trouble signing up for copper, because the infrastructure companies just don’t want to support it. The main competitor left is mobile, although I would love to see StarLink be available here.

https://en.wikipedia.org/wiki/Ultra-Fast_Broadband


Yup, I used to have 1Gbps in Auckland with fiber right up into the house, moved to central London and I could get...a 10Mbps plan...that only delivered 4. Fortunately out in the suburbs now I can get up to 330.

There are a lot more people in the UK though, housing is much denser and people are very resistant to change; it's easier to rip up the ground in NZ bc our country has only existed for 150 years, in the UK it's a lot harder with all sorts of caveats/consents.

Same as covid: Jacinda did an excellent job, but at the same time as a realistic kiwi, it's very true that NZ has a very small, spread out population and the benefit of being small and extremely southerly, international travel just isn't as prevalent there as say, Europe.


About 15% of that Openreach rollout is Fibre to the Premises. Overall about a quarter of UK homes can get FTTP/FTTH either from Openreach or another provider. A lot more in Northern Ireland, somewhat less in Scotland.

However actual usage is quite different. If your Internet seems fine, why would you buy a more expensive service that claims to be "faster"? In my city Toob are trying to aggressively acquire customers for FTTP with a price they're presumably losing money on, but I don't expect that to last.

The technology for a setup like your parents is either VDSL or G.fast. Near to their old (perhaps slightly battered) green BT cabinet is a newer one, the newer one has fibre to it, and a DSLAM, the DSLAM ties to the old cabinet with copper, and so only that "maybe half a mile away" distance is covered by the DSL technology, from there it's fibre.


I think nearly 100% chance we'll get at least ~1T in infra spending on the 'hard' stuff which includes broadband. But that cost makes me wonder is it worth it when we could instead support efforts like Starlink?

Or build a government run version. though while lots of good reasons to - e.g. low rates, free access for kids etc - letting Gov own even more of our internet is probably not great.


20mb is totally fine for hd streaming and video conferencing. if that isn’t working well you aren’t getting a steady 20

I'm typing this message from Starlink. For me it's absolutely transformative; 10x the bandwidth I can get from any other source and very reliable.

Except for outages related to obstructions. That's a real problem and the author's situation is not good. There's ways to work around it on your property; a taller mount, a tree install, cutting some trees. But ultimately Dishy needs a clear view to the north and there's no getting around it.

I have some smaller obstructions for my install and it was a little annoying but fine. But in the past week or two it's gotten way better: my packet loss went from 2% to 0.6%. Details here: https://nelsonslog.wordpress.com/2021/07/20/starlink-improve...


> But ultimately Dishy needs a clear view to the north and there's no getting around it.

I don't know much about radio or Starlink's signals, but is this a situation where a strategically-placed radio reflector would help? Assuming those are a thing. So like, a reflector mounted on both sides of a large tree. Are the signal beams too narrow for this to matter?


The signal needs to be carefully aimed, so unless you get software that is able to take the reflector into account, it's just making things worse.

huh. perhaps I'm exhibiting my total ignorance here, but why do they have to point North?

I know that our Satellite TV dish when I was a kid had to point to a specific angle southwards, to match the geostationary position, but I'd not expect that with Starlink, unless you were in the Falklands or Antarctica, or something?


It's a purposeful choice to avoid interfering with radio bandwidth allocations to existing geostationary satellites. As I understand it, Starlink dishes aren't allowed to send signals to a portion of the sky around the equator where the geostationary satellites are located.

Starlink would probably work even better if they didn't have to deal with this restriction, but Starlink might not have been allowed to exist if they didn't design it to work this way.


Huh. I had no idea, but makes a lot of sense, thinking about it.

Thanks!


First to clarify; Starlink points north in the northern hemisphere.

Your satellite TV dish is talking to something 36,000km up in geosynchronous orbit, around the equator. That's to the south of you.

Starlink are in 550km orbits moving very fast around the planet in a fairly inclined orbit. As another commenter has said, the apparent effect is the cluster tends to "hang out" in the north. It's complicated, a good visualization should help explain it. I don't have one at my fingertips.


Mm, i appreciate the basics there, but if you look at the spread of satellites against latitude (eg https://satellitemap.space/ ), it's pretty consistent so without any outside reason, I'd expect the Starlink dish to simply point up across most of the world, away from the poles.

coder543's answer elsewhere here seems to furnish the remainder of my confusion.


Maybe this picture helps? https://www.nasaspaceflight.com/wp-content/uploads/2020/06/s...

The radius at a northern latitude is smaller than at the equator, so the satellites are more tightly clustered.

I think coder543's answer is wrong, or at least has the causality backwards. Those geosync orbits are very, very far away from the LEO orbits of Starlink.


I'm assuming that at 550 miles up, the beneath-the-horizon range will be quite broad, but as equally, I'd assume further-off satellites to be slower and more conflicted, given the density and interference presumably increases.

I just don't see how - sans external requirement - pointing up isn't more efficient.

- ed. Ok, sorry, I get it. Aim in any direction and you have access to more satellites, even if further away. Pointing up isn't 180º access. Reception is probably less than 90º, I guess? My bad, I'm dumb.


It's OK, orbits are confusing. We all know in our hearts the earth is flat.

North is where all the satellites cluster as they reach the peak latitude of their orbit period: https://cdn.geekwire.com/wp-content/uploads/2019/02/190208-s...

My Starlink experience has gotten a lot better recently. In the past few months.

I'm in rural Arkansas, near the southern edge of the rollout still I believe. I have maintained 3 ISPs this whole time. I have an EM160R LTE modem that will do 5x carrier aggregation and pulls around 240-250 mbit from my local AT&T tower. I also have T-Mobile's 4g home internet (5g works here on my phone, but they won't give me the home internet for whatever reason) which pulls 100-115 mbit. Starlink itself is somewhere between 180 and 240 down, but only 15 up. On the ATT line I can get 40-60 mbit upload, and that's one of the main reasons I keep things set up this way.

I'm about to try disabling the wan port for T-Mobile to see what it's like without that ISP. I don't do any connection bonding - straight up round robin load balancing with no stickyness, and with the amount of servers and services that use multiple TCP streams I can see 300+mbit downloads often. Pings range from 30 (when using Starlink) to 90. (when using one of the LTE connections)

I no longer game enough to comment on it. My kids play Roblox and PS4 online games and don't whine about it, so I think it's sufficient.

I don't really do Zoom meetings. MS Teams is what we use. I don't use the camera very often, but the calls will pause and drop and the people I work with have coined this as, "being Starlinked". Usually a few seconds and rarely does it take an actual redial to reconnect anymore. Just a dead period.


If you get 250Mb from AT&T why bother with the others?

Reliability. I’m also trying to not use too much data on each provider. Afraid of getting kicked off.

Cell service is metered, even if they say it isn't.

I'm in Ontario, Canada, 46.5 degrees latitude and typing this on Starlink.

Even with the occasional dropouts Starlink is 10-100x better than any other option that we have here (the only options are LTE, or other satellites, like xplornet).

Even though we're only a few minutes drive from a municipality of 160,000 people and on a major highway, there, is no wired connection, and doesn't really seem likely that a wired connection will ever happen. Since moving here 7 years ago the pricing/data rates for the LTE data packages available have doubled in price. Literally doubled.

With Covid we had two adults working from home, and two kids home schooling, on a slow LTE connection with a total bandwidth of 100GB up/down. Even things like windows updates required planning and rationing of the internet.

The state of connectivity in Canada is so abysmal. At this point I hope Starlink matures enough to add a voice service.


100W 24x7? That's quite a lot, right?

This adds some color perhaps to the argument that this is for underserviced regions -- they don't mean third-world or impoverished even though it sounds like that, at least when I heard people defending Starlink.


It's a little less than a modern efficient fridge (2-3 kWh/day). I used to have 100W light bulbs in the house, so it's not a crazy amount of power, but it's most significant to anyone planning on using Starlink 'off-grid', since it's a lot more than just a little 4G or 5G hotspot, or a standard DSL or Cable modem and router.

A modern efficient fridge uses way less than 1 kWh/day. The one I bought in 2018 uses 0.44 kWh/day. 2-3 kWh/day is awful.

Three random examples:

https://www.appliancesdirect.co.uk/p/kge36awca/bosch-kge36aw...

https://www.appliancesdirect.co.uk/p/ffu3dx1/hotpoint-ffu3dx...

https://www.appliancesdirect.co.uk/p/htf-540dp7/haier-htf540...


I should be more specific: an efficient 'USA-megahome-sized' fridge. Our typical fridge is often 25-35 cubic feet (850 litre)... that's about double the size of the small fridges used in many parts of the world.

That's not an average size even in the USA and you could probably still do it on less than 1 kWh.

Of course, once you're doing the 'USA-megahome-sized' thing, it's pointless to bother with energy efficient fridges. And many people don't!


Anyone know if Starlink is happy turning off and on again throughout the day?

I know we are all used to 24 hour internet, but if I were energy-conscious or off-grid, I might want to turn it on for 3-4 half-hour sessions during the day.


I tested it a few times (putting in stow mode, turning off while I was rebuilding network rack, etc.), and it always picked back up within 2-4 minutes.

I know a few people who have Starlink on an outlet timer or WiFi switched outlet, and only have it run during the day. It seems to be okay with that.


Yes, 2.4kw in a day is a lot. I was hoping to switch to it from tethered cell service for my off-grid cabin. (I need 24/7 for security cameras), but it looks like I won't be able to without a substantial solar and battery addition.

It would cost me about £11/month to run. For context, I currently pay ~£80/month for electricity (UK average is ~£60[1]). Not prohibitive, but it's a pretty sizeable increase. When you consider I can get 4G connectivity[2] for £22/month, with significantly less power consumption,, it doesn't seem that attractive unless you have no other options.

[1] Table 2.2.1 https://www.gov.uk/government/statistical-data-sets/annual-d...

[2] Claiming typical down speeds of 50Mbps-100Mbps


100W is a couple of light bulbs. It's not nothing, but it's not all that much. It'd be challenging off-grid, but not a concern anywhere with electrical infrastructure.

100W is a lot. It is a couple of very old incandescent bulbs or ten 60W equivalent LED bulbs.

No, it’s ONE incandescent bulb. And much less than the bulbs that were needed to light a fairly large room. We used to have a lot of 100W lights until not many years ago, honestly I don’t see the problem in spending that amount of energy if you have no other means to access a fast internet connection.

perspective i guess - i used to use 2 100watt bulbs to light a single room.

For perspective, these will cost around $0.3 a day in electricity, when in the room you spend most of your time in.

And yes, that's easily over a hundred dollar a year.


It is a lot but it’s also outside so it won’t heat up your house and in most places it only adds up to about $10 a month or less on your power bill

A more durable reference comparison: 100W is similar to a human. An adult human, not exercising heavily, but not asleep, maybe reading a book, or talking to a friend, something in that ballpark.

Also, if you aren't off-grid, you're apparently not so far from civilisation that previous utility suppliers couldn't be bothered to provide service to you. Maybe if this generation's utility suppliers got their act together you wouldn't need Starlink anyway.


At 25c per kWhr, that's 60 cents a day, ~ $18/mo.

To be clear, electrical rates are much different in different parts of the country. At my house in the suburbs outside St. Louis, MO, the cost is about $9/month. Not nothing, but not too significant compared to the total cost of the service.

That's actually non-trivial. That makes the monthly cost nearly 20% higher.

If it's your only (or best) option for decent internet, that cost probably won't consern you at all.

People who are off grid might grumble at having to provision more solar, battery and inverter capacity.

But for people who have other Internet options, $18/month might influence their decision.


25c per kWh is pretty high, I think that's high even for CA. Most states are in the 12-13 cents per kWh range [1].

1. https://www.electricchoice.com/electricity-prices-by-state/


Electriciy in CA is ridiculously expensive. $0.27/kWh average rate is common in CA. Going up to $0.40+ if you get into the higher tiers.

But even with insane generation rates, most of the electric bill is the power company service fee, not the generation fee.

For example, out of a $200/mo bill, maybe $50 is electric generation, the rest is the service fee. PG&E is insanely badly run.


Many places are tiered, so if you go above a certain kW/h the rate doubles or triples. Cali and parts of NY state are like that (never lived anywhere else tho).

8c per kWh here where I'm using it here in rural Arkansas.

Is this power requirement just because the hardware hasn't had a chance for years of iteration or is it a "hard" requirement that no amount of product iteration can fix?

In other words is it 100W all the time because of physics or is it just "sub optimal" hardware that in theory can be mitigated through smarter protocols and fancier hardware?


For the foreseeable future it's mostly physics.

The "all the time"-part is of course something that can be worked on even without changing the hardware. A simple switch on the outlet to turn it off during the night or if everybody is at work/school and thus doesn't need it would reduce power consumption a lot already.

"Fancier hardware" could also mean a timer for the outlet switch or a "smart home"-solution. This could mitigate connection time by turning the router back on before you arrive at home/wake-up.

There's a lot of potential for not having to use 100W all the time without hardware changes.


I'd say 80% physics, 20% poor hardware/design.

Talking to a satellite in space through the atmosphere has a lot of attenuation. That's for the physics.

For comparison, wifi is 0.03W for 100 meters, mobile phones are 2W for 35km range. A satellite using 100W is not out of the ordinary.

There's a balance to decide on, power vs bandwidth, noting that doubling power has little effect on signal (log2). They could probably half the power for little difference in operation, but a notable difference on the bill.

It's exactly the sort of thing that should be adjustable, there could be a setting to adjust power/bandwidth/quality. Both for customers and for support.

edit: the video mentions up to 300 Mbps and up to 200W power consumption during snow (extra power to melt the snow). They have a lot of margin to reduce power while still being able to watch YouTube 4k.


Presuming that a large part of those 100 W go to the power amps for uplink transmissions, it seems that it should be possible to make at least that part scale with how much outgoing data the terminal is actually sending. But that may require new hardware, and it may conflict with the ambition of reducing cost per unit.

It totally can make sense in third world improved regions, just not in homes (and/or possibly not on all the time). You could pretty easily run a small business off of 100 mb down, 50 up, and in that setting, you can probably power it with a small solar panel and just turn it off at night.

A "small" solar panel? A 100W solar panel is 2 x 3 ft (.60m x .90m). You'd need multiple panels even if you run it only during the day, because fixed solar doesn't just go from 0% to 100% once the sun rises.

Solar technology, including the footprint is advancing rapidly. I believe there are panels that can do much higher wattage around that size. Your point stands though. “Small” is a relative term.

> Solar technology, including the footprint is advancing rapidly.

Is it, though? The price drops have been significant, yes, but the efficiency of COTS panels hasn't seen any dramatic improvements in the past 20 years. Silicon panels have a theoretical limit of 29% efficiency and current models are close to 20%. There's not much room for improvement left there.


You would need 1 standard 400 watt solar panel ($250) plus batteries and inverter ($???) to power starlink 24 hours a day in this mythical zero power area that you all seem to be worried about.

In which fantasy land would that be the case?

A 400Wp panel won't be able to provide 400W most of the time to begin with. Unless you're in the middle of a desert near the equator and have a sun-tracking installation, weather and varying daylight hours are a thing.

Under realistic conditions, a single 400W panel gets you between half and two thirds of the required power (e.g. up to ~1.5kWh/day, depending on location).

Just a quick reminder: 400Wp means the panel produces 400W of electricity when brand new (it'll degrade over time), at 23°C ambient temperature and an incident angle of 0° (i.e. sun directly over it). None of that is generally the case 8h a day on average, which is why your estimation won't work.

Depending on where you live, optimal sun hours vary between 3.5h/day and 6.5/day (continental US) or 2.9/h/day to 5.9h/day (most of Europe) [0].

You'd also need to plan for big batteries, since seasonal differences may be huge depending on latitude.

[0] https://solargis.info/imaps/


That is what batteries are for. and you don't need many to do the job.

A reliable 100W solar installation is indeed non-trivial especially at higher latitudes. Source: typing this on a packets routed via a solar powered relay site on a mountain. Our site only needs ~20W.

A single 400 watt solar panel and batteries could power this 24/7, no problem.

No. It can't.

My office has 5kW of panel hung, 10kWh of battery, and my Starlink terminal is on the house system (grid tied) because an extra 2.4kWh/day isn't workable with my system in the winter, short of a LOT of generator time - I'm severely power limited during inversions, and the Starlink dish more than doubles my office's "idle power draw" (it's the property network hub, one of our internet connections, inverter idle, sleeping computers, etc).

Yesterday, the 15.9kWh system on the house produced 78kWh for a "sun factor" of about 4.9, so a 400W panel would produce about 2 kWh. Whoops. That's not 2.4.

In the dead of winter, the same system can produce 2.5 kWh - yes, 2.5 kWh on a 15.9kW nameplate system. A 400W panel won't even power on the charge controller in those conditions.

To reliably power a 100W load, 24/7, in most areas, requires probably 1500-2000W of panel and 20+kWh of battery - or someone willing to light a generator, which is the far cheaper option. But "a 400W panel and batteries" (implied as a trivial thing to set up) definitely won't. It won't even run it 24/7 in peak sun most of the year.

... and that's before it tries to melt the snow off. From the blog post:

> During the heavy snowfall, Dishy quickly spiked up to 125W, peaking at 175W towards the end of the snowstorm.

Solar panels don't produce much covered in snow, either.


You are probably technically correct, but most people don't need 24hr internet access, so that one 400 watt panel would provide high speed internet for let's say 12 hours a day and that's plenty for this odd scenario of living in a place with no power but still needing high speed internet.

The claim was 24/7 powering with "a 400W panel and some batteries" - which is simply false.

I deal with the realities of off-grid power in my office on a daily basis, and a vast majority of what's written about solar and batteries by people who don't have experience with them is simply wrong. I try to correct it where I can.

With a 400W panel and a few kWh of batteries, you could reasonably accomplish 14-16 hours of access during peak sun in the summer, 8-12 hours in spring and fall, and 0-3 hours in winter, except for 5-6 hours on sunny winter days.

Although I've heard that the newer Dishys use somewhat less power (50-70W), which does improve things.


Ah the joys of hiking 1000ft up a mountain in subzero temps through 4' drifts to clear ice off the panels. Although, for the most part I've found that provided you pitch the panels at 50 degrees or so, snow tends to slide off. The battery system has to be able to carry the load through 48h of zero panel output at least to let snow clear. In 20 years I've only had to physically clear ice off the panels three times.

48h is less than 5 kWh of batteries, one quarter of what the parent post proposed as the absolute minimum.

It will depend somewhat on the area, and you can trade panel area for battery pack capacity to a reasonable extent - but you either need a lot of panel area to power you on dark days, or you need a lot of battery to ride through those periods. Out here, at least, we'll have a week where my 5kW of panel on my office produces less than 1kWh/day, often rather significantly less.

The problem is that after 48h, then... what do you have? You need to have either enough panel area to both run loads and charge the batteries, often in dim conditions, or you need a lot more battery.

Doing reliable off grid power takes a lot more than most people who haven't dealt with it assume, especially if you're dealing with serious winters and don't want to light a generator.

I'm basing my estimates off what it would require for me to run 2.4kWh/day, year round, without generator, on my office system - and it takes a LOT. I have 5kW of panel, 10kWh of battery, a typical winter day load of around 3kWh (base loads + minimal compute loads for a work day), and I still require a couple gallons a year of gas to run this reliably. Depending on the winter, I go through 3-5 gallons of propane for heat as well, though I've been trying to move to kerosene - more efficient, as I can run a lantern as a combined heat/light plant on my desk, and heat my hands, which are the only real constraint on cold temperature work in there.

I know those numbers sound somewhat nonsensical for a 100W load, but I literally make a living in an off-grid office...


How on Earth you would need 20+kWh of batteries for a system that even run 24/7 uses 2.4 kWh? And absolutely doesn’t make sense to run it 24/7, even when no one is using it, if you are running it off grid.

Doubtful. Most of the US does not get 6 hours of full/peak sun (2400w/6 = 400w)

Gotta remember the double whammy here. Starlink is mostly rolled out at higher latitudes where solar power is less effective. Still, solar is cheaper than utility power most of the time. Might need 1.5 of those panels. Maybe I'm crazy, but I wouldn't consider off-grid unless I had like a 5KWh array at a minimum. Probably with wood burning stove as primary heating source if I really couldn't expand that solar for electric heating due to cost.

With heat inverter pump tech. taking over fridges, water heaters, AC, etc. and making them so much more efficient, it can give a lot of headroom on energy requirements.


Only if you live in the Sahara. Panels are rated for power they produce at 1,000W/m2 irradiance, in UK average is 100W, that panel will produce 40W on average.

That seems absurdly low. My 250w panels get within 90% of rated output and I'm in Seattle.

I think they're averaging throughout the year, using the number of "Mean Annual Sunshine Hours" * "Percent Possible Sunshine", which would account for average cloud cover and such.

I'm sure that for a few hours during the summer, the solar irradiance in London is probably 90% of what it is at the equator. That doesn't help most of the year.


Real-time visualization of the Starlink constellation:

https://satellitemap.space/


That's interesting to see with knowing that just a fraction of the planned constellation is there. That's a lot of dots already. I'm guessing that the few visible string of dots are more recent launches that haven't quite reached their final positions yet. That's also interesting to see how long it takes the train to not be a train any longer, while at the same time showing how frequently new launches have been occurring.

I believe the constellation is about 90-95% complete now. The first shell at least. There's plans for more shells but that's for redundancy, not coverage.

I was going to comment about how surprised I was that they almost finished the first shell without deploying laser links but apparently they have: https://arstechnica.com/information-technology/2021/01/space...

I think the laser interlinks are proving more difficult than SpaceX anticipated. I had expected them much earlier in the deployment than now, and it seems they are currently only planning them for polar orbits.

I think they need quite a lot of satellites in polar orbits to get coverage at the poles, and they have barely launched any yet. They are close to 100% coverage of the rest of the world already though.

They're doing polar launches but compared to the number of customers in sub-70 latitudes it's a pretty small market.

Well, 9/10 is still a fraction, just not as small as I thought it was. ;-) I must have fallen behind on launches, not realizing they were this far along.

That is a lot redundancy. Current satellite count is around 1,600, and FCC approved plans through phase 2 would bring that up to ~12,000

Is that really how many Starlink satellites are up there? Foo. I had no idea.

What are the dense lines of satellites?

At least for the short lines, they're launched in batches from one point, so they take some time to spread out to their final orbits. These are neat to view after a launch.

Some of the comments here are kind of ridiculous. They are complaining (or just pointing out) issues that are EXPECTED at this stage of the beta.

Starlink has like 1700 of the planned 42,000 satellites in orbit now. Of course there are going to be temporarily blips in service plus the random longer dropouts during system upgrades.

Even with the current issues, the service is revolutionary in the remote areas that it's intended to service. Why Geerling thinks it's appropriate to compare beta starlink to his home cable/fiber service is beyond me. It's totally fair to review the current state of starlink, but to then conclude that "I don't love it" because it's not as good as his cable service is just plain dumb.

Why did you even begin the review with the expectation that it could be better than your land service in it's current beta form? You're not even supposed to be on the starlink service if you have great landline bandwidth and starlink should block you from their service as you're stealing bandwidth from people who don't have access to high speed internet.

One person even said "I hope they can figure out why it drops occasionally" as if some of the smartest people on the earth don't know exactly why it drops out. It drops because the satellite mesh network is only 4% complete!


I mean, Starlink is available as a closed beta commercial product. There's nothing wrong with comparing it in its current form to cable service in its current form, especially if you're going to pay $$$ for it.

The point is for the reader to figure out if they should try Starlink right now (if eligible) or if they should wait for some of these issues to be resolved.

To put it another way: would you complain that reviewers were judging Google Glass unfairly, because Google had grandiose plans for it in the future (that ultimately never happened)? Or would you recognize that Google made the decision to sell Google Glass in its current form, and thus accepted that it would be judged against its competitors?


It's totally fair to review the current state of a beta service. It could be informative for people who are unsure if they want to try the beta-service.

But to compare it as an equal, in it's beta form, against a service that it's not even built to compete against is just plain dumb. If you already have access to high speed internet, starlink it not intended for you.

His final comment about "Liking, but not loving starlink" implies that he's comparing starlink to cable internet as equal competitors. They are not and they are not intended to be and that's even considering the fact that starlink is in beta.

This guy should not even be allowed on starlink (in the long run) because he already has access to high speed internet.

This whole comment section is full of stupidity like:

"isn't 100watts a lot of power?" <-- typed from gaming computer with a 1000watt power supply "what about space garbage, isn't space garbage bad" <-- as if LEO garbage won't just decay back to earth "will starlink be like FSD and maybe never get delivered" <-- as if they are related issues, they are not "I don't like Elon" <- because reasons, but also irrelevant "starlink bad because worky less good when obstructed" <-- too dumb for words "i'm worried about the starlink monopoly" <-- you should be banned from the internet


> "isn't 100watts a lot of power?" <-- typed from gaming computer with a 1000watt power supply

This feels like a bit of an unfair dismissal. I'm using a laptop that rarely goes above 30W (and is off half off the time), so I'm not sure you can make that assumption. A 1000W power supply also doesn't mean it's using that continually - a GPU can consume >300W under load, but drop to around 10W when it's idle. Further, 100W 24/7 would add around 12-13% to my power bill (which is already above average) - it is a noticeable amount of power.


It was a single light bulb in a decent size room until not so long ago. And some rooms needed 2 or more and it was absolutely normal.

> "isn't 100watts a lot of power?" <-- typed from gaming computer with a 1000watt power supply

A gaming computer with a 1000 watt power supply will reastically have a 50W idle load. It's only if the CPU and GPU are fully utilized (like in a gaming scenario) that it might draw several hundreds of Watts.


> This guy should not even be allowed on starlink (in the long run) because he already has access to high speed internet.

I don't plan on owning it in the long run—I'm going to be giving the dish to my cousin who's on a farm with slow rural DSL once Starlink is available in her area.

Unfortunately right now Starlink's available in suburban St. Louis but not in most of the rural communities around it :P


> Starlink has like 1700 of the planned 42,000 satellites in orbit now.

I'm not sure if anyone believes that 42k number. They are launching ~60 satellites at a time - that would mean ~700 launches. There is no way that will be economical for the relative handful of people (500k? 1-2M?) who could realistically be interested in this.

Not to mention, the lifespan of these satellites in orbit is tiny, just a few years. They would have to be constantly launching new satellites to keep up.

The current state is probably more or less the best Starlink will ever offer - as more people will join the network, coming closer to Musk's 500k number, bandwidth will significantly diminish, even if the number of satellites is maybe doubled.

And if federal funds dry up, I expect the whole venture will quickly go bankrupt, or remain alive with a handful of satellites and a huge price spike.


Hold onto your hat then: in the US alone there are at least 20M people without high-speed internet access [1]. Around 45% of the entire world's population [2] still doesn't have internet access - at all, not just broadband.

They've said starship, when operational, will be able to deliver 400 starlink satellites to orbit at a time. The plan doesn't sound far fetched.

[1] https://archive.is/do3Qp

[2] https://en.unesco.org/news/new-report-global-broadband-acces...


The vast majority of these people will not be able to afford a 1500$ satellite dish (that's how much it costs to build the thing, even though they sell it for 500$ for now) + 100$/month internet.

No but their community or local government can and will. Internet access has monumental positive impacts for an economy and $1500/$100month is such a small amount of money for a city or communtiy resource for the expected return. No, it's not going to be on every mud hut in the world. But starlink will definitely provide internet to millions who didn't have it before, and do a great deal of service for lifting some of those people out of poverty.

And why should one assume the cost once deployed in the millions will equal the cost while in beta with 60k users? There's another comment here already mentioning their prediction of a $300 cost for the dish in the future, you can only expect the subscription to go down as well.

Plus, the places where Starlink is the most useful - rural areas - already have sky high costs. I just looked up what service looks like in Brazil: ~R$400/month (US$76) for 20mbps, with a 80GB data cap. When you have to buy the dish it costs over R$10k (~US$2000). This is the best you can get.


According to them, the satellites have a maximum life span in orbit of 5 years. They can launch 60 satellites at a time, at a cost of 50M$ per launch. So they need to spend ~1.4 billion dollars every five years just to keep the constellation at its current size, ignoring the cost of the satellites, ground stations etc. If they have 500k customers, that means that each customer must pay ~50$/month just to pay for the cost of keeping the satellites in orbit.

Add ground station costs, internet bandwidth, satellite production, dish production, customer service, personnel costs, R&D and you'll quickly see that the current prices are much below what is required to be profitable. Even if they have 5M customers (very hard to do in 5 years time), I wouldn't be hopeful for their prospects of not going under.

Of course, they may well survive on government handouts.

Edit to add: the bottom 45% who don't currently have access to high speed internet are unlikely to afford even a 300$ dish and an expensive data plan. The price of current plans is one of the main reasons why they don't have access in the first place.


$50M is the launch cost for SpaceX customers. Musk has said in an interview that the marginal cost of a reused Falcon 9 launch is $15 million. That means even the tiny subscriber base they have at the moment is already paying a significant chunk of the launches.

That would make your numbers $420m every five years, or $84 million a year - slighly more than your average telecom CEO compensation.

But the goal for starship is $2m per launch. Let's assume that is bullshit and go for $10M instead, and only 300 satellites per launch. That's already down to ~$10M/year to maintain the entire constellation. Now you have room for all the other costs you mention. If they reach the stated goals, it will cost peanuts... you can see where this is going. Starship will be a ridiculous leap in launch costs if it comes to fruition.

The max capacity for the network, once at 42k satellites, is estimated to be around 50M people worldwide. Plus, you could have a downlink shared, say, among 50 people and spread the cost - 1-10mbps is still miles beyond 'no internet access'.

You might think the plan is crazy, but the math checks out. If this whole thing works, they could charge $10/month and be in the green - the satellites themselves will be the cheapest part of the whole system. No rent, no right of way, no construction, no maintenance work, free energy and extremely low material costs.


> Musk has said in an interview that the marginal cost of a reused Falcon 9 launch is $15 million.

Musk has said many things in interviews that have little to do with reality. Even if that number is true launch costs are more than the rocket itself. Even if we accepted the number, to make economic sense, they would have to get more value from 60 satellites than the $35M of profit they would make on each launch if they were launching commercialy.

> But the goal for starship is $2m per launch. Let's assume that is bullshit and go for $10M instead, and only 300 satellites per launch.

Prices for rockets similar to Starship are in the $150-200M range, so that is more the ballpark I expect from Starship, if it does fly. Perhaps it can go down to as low as 75-100M range, half the current cost of similar rockets! $10M is fantasy.

Not to mention, they are already being accused of breaking environmental regulations and other problems, so it wouldn't be surprising if launch costs actually go up if they are to maintain compliance.


> Prices for rockets similar to Starship are in the $150-200M range, so that is more the ballpark I expect from Starship

Why would they even develop this new vehicle if it turns out more expensive than Falcon 9, or Falcon 9 Heavy? That would not make any sense. Plus those rockets are not reusable.

They have already achieved a reduction in price/kg by at least 10x with Falcon, see page 11 here: https://ntrs.nasa.gov/api/citations/20170009968/downloads/20...

> $10M is fantasy

$2M is the fantasy, I already added 400% to account for skepticism. Starship is not a conventional launch vehicle, it's designed for full reusability with a turnaround time measured in hours (vs the current 14-50 days for a Falcon 9 refurb), like an airplane. Land, refuel, safety checks, fly again.


> Why would they even develop this new vehicle if it turns out more expensive than Falcon 9, or Falcon 9 Heavy? That would not make any sense. Plus those rockets are not reusable.

They are making a bigger rocket and making it reusable. The bigger the rocket, the more expensive the launch. For Starship the price per ton should be smaller than for Falcon 9 and Falcon Heavy, but that doesn't mean it should be expected to be cheaper in absolute terms than much smaller rockets.

> $2M is the fantasy, I already added 400% to account for skepticism.

You're still talking about a price that would be a hundred times better per ton than any other rocket in history. Allow me to be extremely skeptical, please. Musk's arbitrary promises do not set a benchmark for me - he has been known to lie through his teeth about things like hyperloop, FSD, Neuralink, Starship earth to earth, and timelines in general.

And on the same note, what Starship is theoretically designed to do and what it will actually do when it is actually deployed may be significantly different things.


There is no other fully reusable rocket similar to starship (or even different, it doesn’t simply exist). Please stop inventing numbers and lying. And for your information starship already fled and landed and there is an orbital launch attempt likely in the next 2-4 months.

I'm talking about the costs for launching a rocket of similar size. And note that for Falcon 9, the difference in price for new rockets VS reused boosters is 62M -> 50M, not an almost hundred fold decrease.

Reply, because I can't edit: so far, only the second stage stage of Starship has been flown. The SuperHeavy booster has not yet been flown at all,nevermind successfully landing, and ditto for the two-stage system.

To add on the costs (yeah I have nothing better to do)

- ground station: must be negligible compared to everything else. They are literally planting giant routers out in a farm, need to pay rent, power and cabling to an exchange. Zero other infrastructure costs

- internet bandwidth: despite Netflix's attempts, ISPs have peering agreements that are usually free, as does Starlink

- satellite production: nobody knows, but they have said the payload is cheaper than the launch, so somewhere between $250-500k each at start

- customer service: online only, seems like the dish is self-diagnosing, will likely end up being much lower than traditional broadband


> - ground station: must be negligible compared to everything else. They are literally planting giant routers out in a farm, need to pay rent, power and cabling to an exchange. Zero other infrastructure costs

Fair enough, though you're neglecting the costs for building this all over the world, if they want to get access to any size of market.

> - internet bandwidth: despite Netflix's attempts, ISPs have peering agreements that are usually free, as does Starlink

Actually, according to Cloudlflare[0], you still need to buy a significant amount of traffic through transit agreements - cloud flare pays for 40% of its traffic in Europe (at a price they don't disclose), 60% in the USA (same price as in Europe), 40% in Asia (at 7 times the European price), 10% in Africa (at 14 times European price), 40% in South America (at 17 times European price), and 50% in Oceania (at 17 times European price). Only the Middle East region is 100% peered for them.

So the cost to actually deliver internet is definitely not going to be 0.

> - satellite production: nobody knows, but they have said the payload is cheaper than the launch, so somewhere between $250-500k each at start

So that is another 15M$ per launch at their declared aspirational cost (or an extra 100M$ per launch if you believe the Starship numbers). Just maintaing the entire constellation today would cost ~100M$/year, assuming almost all satellites reach their 5 year life span (current failure rate is 6% after one year).

> - customer service: online only, seems like the dish is self-diagnosing, will likely end up being much lower than traditional broadband

You can't sell online only customer support for access to the Internet if you want to reach any kind of realistic market. Furthermore, customer service more broadly includes delivery, sales, marketing, training materials; and this will be required all over the world, since they must target customers in remote places with little or no acces to the internet today.

[0] https://blog.cloudflare.com/bandwidth-costs-around-the-world...


Fair points. Though I’m not sure cloudflare is a good reference - they are a content provider and will pay for a competitive advantage (latency). The post mentions 50% of their costs come from 6% of traffic, they could simply choose to not pay for transit and let these be rerouted elsewhere.

Once Starlink has the satellite “mesh” operational they would probably become a tier-1 ISP and have zero transit costs. Maybe even sell traffic to other ISPs.


Sure, that's a good point, and the Cloudflare article is from 2016, so the market may have advanced as well.

On the other hand, unless they somehow get the laser communication working, which is still a research problem despite their optimism, they have way too high latency to act as a tier 1 ISP - all traffic currently goes consumer - 1 satellite - ground station - Internet.


Your math is way off. 50M subs with the coverage over land and total bandwidth will put their provisioning rate (Bps/sub) way below even the Geo satellite providers. They need to charge a lot more or provide slower service by having more people.

For the full network:

    42000 sats
    20gbps each
    ~1k users
    2gbps @ 100:1 oversub
    1gbps @ 50:1 oversub
And that’s assuming next generations will not improve downlink speeds. It does assume satellites would be fully utilized which is wrong unless we populate the oceans, but that’s the theoretical max.

That's the entire point. They're mostly over the ocean/areas without high-paying customers 70-90% of the time. Because of that, it would be more like (in the best case): 42k sats@20G * 0.2 * 5 million subs target = 33Mbps/sub. However, that's not even close to realistic either because the high-paying subs are in the US and parts of Europe only, and you can't keep piling 42k satellites in the same plane. So in reality, that will be cut significantly from that.

No, there are so many wrong statements in your comment that I don’t even know from where to start. The current cost for 60 satellites on falcon 9 is around 10M$ if I remember correctly. From next year they will likely be able to send 400 satellites for the same price with starship and the aspirational goal is to have a cost of 2M$ per launch. It implies a cost of ~55M$ to send 11000 satellites. Even if they will never reach that aspirational goal it would probably cost them ~300M$ for 11k satellites and ~1B$ for 42k satellites at 10M$ per launch. It’s perfectly feasible, you just can’t understand how starship is such a game changer.

> The current cost for 60 satellites on falcon 9 is around 10M$ if I remember correctly.

Apparently Musk is claiming 15M$. The public numbers for a launch are 3.3 times that, so even if that is pure profit, it's awfully good profit compared to what Starlink is bringing in.

> From next year they will likely be able to send 400 satellites for the same price with starship and the aspirational goal is to have a cost of 2M$ per launch.

This is based on Musk time - the likelihood of that happening in this time-line is not particularly good, given the track record of his promises. Time will tell.

The claimed 2M$ price is laughably 'aspirational'. Until I see it, I don't believe a word of it.

> It’s perfectly feasible, you just can’t understand how starship is such a game changer.

Let's wait for it to actually fly to space and back until we declare it such a game changer, ok? Musk's promises have a habit of never happening, or happening much later than claimed.


If they are able to get Strarship and the Super Heavy Booster working, which at this point in time I think is very probable, those economics will change a good bit as the cost of launch will come down a lot.

There is absolutely no basis for the cost to be that low. Nobody believes that. Not even Elon, since he's commented on it several times.

The current approved proposal is 11k satellites, the 42k satellites still doesn’t have approval if I remember correctly. Likely from next year they will be able to send 400 satellites with a single launch using starship, that will make the 42k constellation perfectly feasible.

Starship is still in a research phase, so any clear capacity or timeline are impossible to give. The economics of 42k satellites re-launched every 5 years don't work anyway, not for the size of market they can realistically capture (remember that their main market are remote, population sparse regions - which tend to be poor).

I can bet you anything you want that they will not have anywhere near 42k satellites in orbit in the next 10 years. Perhaps in time, over a few decades, as they spend time building up a customer base, they could reach that, assuming Starship delivers on its many promises and if the service proves popular.


Yeah, they will never get to 42k, it's just marketing hype. Once they have one shell with sufficient coverage, adding new shells just tanks the economics of the constellation because of the uneven distribution of demand.

If they are charging full price for it then it is reasonable to complain. It doesn't matter whether they call it "beta" or whatever else.

Applications are open for YC Winter 2022

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: