No consumer will worry about latency or any other component of the physical feat of delivering internet via Starlink. Consumers are currently buying downstream bandwidth by the mbps from one or maybe two providers with no other options. Only technology oldheads who remember satphones and Hughesnet will worry, then research, then try it out for themselves. Starlink has the added benefit of being very easy to give a “consumer trial period.” It will sell itself, in the USA at least.
What do you mean remember HughesNet? People are still on it. Some of those people are young. All of them have been taught what latency is by having satellite internet.
There are plenty of non-technical consumers who have had a very unwilling education in what latency is and how it differs from speed.
EDIT: Sounds like this site's measurements for satellite internet are incorrect. Maybe HughesNet partners with fiber providers in some areas and therefore brings average speed down.
We could use objective reporting of internet speeds in finer detail. Any ideas?
Original comment:
--HughesNet's average latency is now 62ms, 58 ms jitter,--
Sure we do, we have the laws of physics. HughesNet has geostationary satellites. They orbit at ~22,200 miles up. Assuming speed of light, it's going to take ~120ms per hop. There's a minimum of 4 hops (you -> satellite -> ground station -> satellite -> you), or 480ms.
And that's assuming perfect conditions and ignoring the rest of the normal latency encountered online. It's physically impossible to have a latency lower than ~480ms with geostationary satellites.
t=0 You send out ICMP request
t=120ms it hits the satelite
t=240ms it arrives at the ground station, ICMP reply occurs
t=360ms hits satelite
t=480ms arrives back at the originating machine
The only way doing satelite-satelite at a GEO level would be useful would be
1) You're pinging a device on satellite in GEO (one you don't have LOS to)
2) You're pinging a device that's also connected via a different GEO 120 degrees away from the first
in this case you'll
t=0 You send out ICMP request
t=120ms it hits the satelite
t=240ms it arrives at the ground station 1
t=280ms arrives at ground station 2
t=400ms hits satelite #2
t=520ms arrives back at the destination
t=640ms response arrives at sat #2
t=800ms response arrives at ground station 2
t=840ms response arrives at GS#1
t=960ms response arrives at sat #1
t=1080ms response arrives back at originating machine
You could change that, but you won't save much in GEO as you're connecting to two different satelites, 120 degrees apart or 70,000km (235ms)
t=0 You send out ICMP request
t=120ms it hits the satelite
t=355ms hits satelite #2
t=475ms arrives at destination
t=595ms response arrives at sat #2
t=830ms response arrives at sat #1
t=950ms response arrives back at originating machine
Reality of course is that typical satelite latency is 800ms or more (depending how much you pay to skip buffering) rather than 480ms.
In GP's post the second satellite is the same satellite, just on the way back. The post assumes for demonstration that the ground station is the destination.
Color me skeptical that Starlink will reach its theoretical maximum capabilities, or that it is competitive with 5G. I'm fine with fiber until then. It's affordable in most of the world. The problem of broadband monopolies in the US is not going to be resolved by a new privately-owned technology. We should hold local politicians' feet to the fire, not throw money at new tech from a monorail salesman.
Regarding the 3rd world, Facebook offered India free internet with strings attached, and IIRC they rejected it. That idea wasn't too popular there or here.
Would Musk be given a pass for offering discounted or free internet to 3rd world countries, and what strings would investors suggest he attach?
> Would Musk be given a pass for offering discounted or free internet to 3rd world countries, and what strings would investors suggest he attach?
Why would he want to do that? Facebook wanted to do that, because that would drive more users to their platform. SpaceX wants to use Starlink in order to bankroll BFR[0], so what they want is to simply sell the service at a price point that generates them enough money. It's not beneficial to them to discount it, make it free, or play other shenanigans.
--
[0] - Or whatever it's called now; it'll always be the BFR in my heart.
If you have a lot of market power and you want to make as much money as possible the rational thing to do is to charge people according to their willingness to pay[1]. You can't do this perfectly but this is what is behind senior discounts and Intel's huge number of SKUs. Since people in Botswana can't profitably resell their internet in England, since Starlink can easily charge different amounts in different countries, and because the marginal cost of servicing a new region is approximately zero it would be crazy for Starlink not to do geographic pricing tiers.
If money is your only goal, that's completely rational. However, that's not how I've typically heard people champion Musk. He's been lauded as a progressive person whose aims are meant to serve humanity, not merely his wallet. Supposedly.
All human beings are complex and you can't just sort them into good people and bad people. Elon seems to be more idealist than most about the fate of humanity as a whole but also seems to be willing to work his employees to the bone, drive sharp bargains, etc to pursue that goal. He's said that he sees Starlink as a cash cow that'll fund future Mars colonies and I don't have any reason to believe that he isn't being sincere about that.
> All human beings are complex and you can't just sort them into good people and bad people.
Some would make the argument that if everyone worked with only their wallet in mind, we'd be better off. I wouldn't.
> Elon seems to be more idealist than most
Elon is not an idealistic engineer, he is a businessman. He actively misleads about products he must have been told will not meet the timelines he promotes.
> about the fate of humanity
Not sure where you get this idea. The way he talks about AGI taking over the world is creepy. He thinks it's inevitable, nevermind that we have no idea how to design anything close to AGI.
It's not like serving humanity and making money is mutally exclusive. If there are markets in developing countries he can serve cheaper than anyone else both sides profit: he makes money, the people get cheaper internet.
That said, internet already tends to be the infrastructure demand best satisfied in developing countries. Building cell phone towers is much easier than building roads, and with cheap labor and no purchasing power comes cheap internet.
> If there are markets in developing countries he can serve cheaper than anyone else both sides profit: he makes money, the people get cheaper internet.
Let's see whether it is net-neutral internet or not, then judge. My guess is it will have exclusive offers to content distributors in order to lower cost.
I think your parent also meant the developed world.
In devloping countries it's sometimes much easier to get fibre. No or very little infrastructure means no old copper wires lying around. You can go fibre instantly this way.
Fiber is affordable in cities (a market nearly impossible to service by satellite anyways). Even developed countries struggle with deploying fiber in rural environments.
Every satellite internet offering targets rural settings, ships and planes. StarLink is no different, just that they will be competitive in a wider range of such settings.
>> We should hold local politicians' feet to the fire, not throw money at new tech from a monorail salesman.
We need to do both, as Seattle has proven with its sporadic availability of gigabit fiber and the shutdown of third parties attempting to bring it here.
Yeah, I happen to live in an area that CL offers 6mbps maximum to, and it is weirdly in a minority-dense part of Seattle metropolitan. They have zero plans to expand to our area, they've stated repeatedly.
The city contract requires they service the lowest income parts of the city first, hence the extremely patchwork rollout over the last few years.
Due to the mandatory client base skew, the city is causing them to redline middle and high income areas of the city. Fiber is in Georgetown and Rainer Valley, but your SoL in Fremont's built up areas and in Sand Point.
Universal coverage should have been mandated, but its easy to end up like Kenmore (which required total Fios coverage from Verizon, now Frontier) and has seen Finn Hill go unserviced for years. Enforcement of the franchise contract is just as critical as initial negotiation.
Outside of a few select areas, most of Seattle only ever saw ADSL (or no DSL in parts of the north end) as we didn't allow large VDSL2 cabinets in the public ROW. I think this has caused Centurylink to build fiber in the city, as they've rarely overbuilt VDSL2 areas like Bellevue, Kent & Auburn.
I've noticed they seem to offer ADSL (6mbps) after 6pm or so at addresses that already have fiber, presumably their prequal database for fiber goes offline for a few hours a day.
I've called multiple times and schedule a check every three months to see if it's been updated. Been doing that for years. Won't give you my cross streets but I assure you, they have nothing for me.
They say it's available just a few blocks over. That's been the case for a long time and they can't seem to get it about 0.25 miles further. Been that way for years.
Starlink (and other similar systems that are hopeful to launch) are the real competitive threat that Google Fiber only partially lived up to being.
It threatens the existing entrenched providers everywhere. They can't hide from it. They will have no choice but to respond. They face a scenario losing tens of millions of access customers over time to this approach. And the existing options are priced so high in the US, it provides a big fat margin opportunity for Starlink & Co. to target (and ride for the benefit of paying for the buildout).
We had to go to space to route around the cable oligopoly, beyond their local crony jurisdictions.
Nope, Starlink would be crushed by tens of millions of customers (mostly concentrated in major metro areas) in the US and those customers would switch right back to cable. The capacity isn't there.
Can you show me the definitive proof that Starlink (along with its competitors, who will plausibly also launch a vast number of satellites) can't support eg 20 million customers in the US over time? I'd be very interested in that demonstration of the limits on the market.
Also can you support the premise that their customers would be concentrated in major metros, when the biggest beneficiaries will be outside of metros where broadband options are drastically worse (which is why the primary market for HughesNet the past two decades has not been concentrated in major metros).
Access providers in metros can soundly compete with Starlink. They'll lower prices, increase speeds, improve bundles, etc. Their infrastructure and customers are already in place there, they won't just ignore Starlink, they'll compete. That heavily limits the upside potential for the Starlink concept in metros. It's everwhere else, mostly lacking any real broadband, that Starlink & Co will face minimal competition and will particularly lure customers. It's why HughesNet still has over a million customers today.
I've seen a lot of people claim - with very little supporting evidence thus far - that the market is extremely limited due to capacity restraint.
While you're proving out your claim, if you don't mind given that you've got a strong handle on the market ceiling, please provide what you believe to be the maximum subscriber potential for Starlink - and the approach in general - in the US over time.
I think Starlink is for rural customers and can't compete with decent wired broadband. You were the one who seems to be putting it against cable (which exists in the metros, not in rural areas).
The MIT work linked in this thread calculates a total worldwide capacity in the ~20 Tbps range for Starlink.
I didn't see the link, but here's the presentation[0] I believe you're referring to. It estimates Starlink's max total system forward capacity at 23.7 Tbps.
> All of them have been taught what latency is by having satellite internet.
Yes, life is a great teacher.
HughesNet sends me a post card every quarter or so. Sure I live in the middle of nowhere and would seeming be likely to want it. What makes me laugh is even the old people in the area who you'd think wouldn't know how to use a computer will warn friends, neighbors and community not to use them. They may not know the term latency, but they know that the quality of calls are poor and they know uploading videos from their phones takes way too long.
I was going to ask, "Didn't HughesNet try this?" .. I didn't realize they were still around, although it does make sense for a lot of rural areas with limited broadband options.
Or RVs. Even chain stores use the service: easier to manage 1 big Hughes subscription instead of 500 or 5000 different local ISP contracts, each with their own quirks.
LTE hasn’t made as much of a replacement in Home Broadband in rural as much as you’d think largely due to tower congestion. Cell site density is so low — plus deprioritization at 30/50gb.
In addition you’ve usually got only 1-2 networks in deep rural.
competitive gamers make up a very small percentage of all internet users (or even gamers!). Like professional athletes, they will move or go to a special facility for this sort of latency requirement.
Many games are not as latency sensitive as you think - people used to play over modems!
If by "competitive" you mean anyone who plays any sort of real-time game such as RTS, FPS, or even real-time mobile game. Many consumers may take a while to figure out that latency is the reason they lose/die when it doesn't look like they should, but they will feel the effects immediately.
Granted users in rural areas are likely the most apt to benefit from this service and those users don't likely fantastic internet to begin with. However, if SpaceX markets this to areas with decent quality internet they're likely to get a bad reputation for "crappy" internet.
Yes, even those people make up a tiny fraction of the internet. Satellite internet typically will not work for those people, and there is no point in trying to sell to them.
That's the same percentage as Linux users, and just yesterday I closed my Box.com membership because the don't have a Linux client "due to low demand".
Ehhh, not sure about this. If you're a parent and your kid is begging you to get faster internet, you might listen to them a little bit. If their marketing can get those kids to do that, then that might be a successful campaign.
I agree in principle that kids can influence parental purchasing, and that marketing is geared to take advantage of that. But the effect appears to be most effective for items that are exciting and tangible objects in front of the child; think checkout aisle candy, not “boring” adult things like finance and bills.
Here’s an interesting parallel; how much say does any child have in the cell phone plan that a family uses? Probably damn close to 0. That’s a decision that the parents make on their own, and dictate the rules of usage to their children. I suspect that internet service will remain that way as well.
For candy and toys, sure. But there are a wide range of goods and services where the child has 0 say. Cars, homes, and even cell phone companies never market to kids in order to influence the parents, implying that children have no meaningful impact in these areas.
Now these services advertise to adults with children, such as being able to call your kids whenever with the unlimited cell phone plan, but that’s still advertising to the adult, not the child.
I play League of Legends and have only DSL but my ping is only 35 ms and I can't complain -- many other players have it worse than me.
What did make it impossible for me to get kills was having 30 ms of latency in my display system. I think the network latency is well-compensated (if the typical user has 50 ms of latency, just delay me by 20 ms) but the display latency kills me. When I switched to a different monitor I started playing much better in about five games.
I think the game generates a 10 MB log for a 30 minute session so the bandwidth requirement is not much.
I had the same experience playing Titanfall earlier; in that case turning on the "Game Mode" on the TV turned me from someone who couldn't win at all to somebody who could get halfway up the ladder.
Display (and input!) latency are real things, and matter quite a bit. There have been studies in the past where people measured the full button press to frame update time using high speed cameras, and not all games are equal. As for game mode on TV's, that specifically why it exists. As a passive medium a little latency doesn't matter, but when a system is responding to input, latency is noticeable.
If you're interested in how different TV's perform, segregated by their individual input ports (in cases where some ports are optimized) and modes, check out https://www.rtings.com/. I used it a couple years ago to pick out a good 4k TV as a monitor, and the very thorough information helped me pick out something that works well as a computer monitor (and also for the occasional gaming).
How does 30ms of display latency affect LoL gameplay? I don't play it, but I've seen videos of it a few times and with it not being a FPS I wonder what aspects of the game are that greatly affected with an additional 30ms of display latency.
Mind you, any mouse lag drives me up the wall. I've never measured at what point I perceive that mouse lag, but it's a fairly low number I'm sure.
I guess it depends on the game being played. When I played C&C in the 90s over modem, the latency was huge but it wasn't a major concern for that type of game.
Now that I play real time sports, and I'm no professional, it still makes a big difference in my enjoyment of the game when I'm playing at home with ethernet and top tier internet service vs playing at my parent's house on wifi and basic internet. The lag can really affect my ability to play my best and it sucks losing because you don't have a low latency connection.
That said, I'm very curious to see what the latency will be, because if it's within a reasonable range it could be acceptable. Either way, I don't think even casual gamers will have a significant impact on their sales, but it will have some if the lag is too great for some of today's games.
Try having an RDP session to fix Windows servers based in Australia with 400+ ms latency (Europe is unfortunately nowhere near Australia) on your GUI, the ordeal will leave a permanent scar on your psyche.
The speed of light is insignificant for such a short distance. Even if the signal travels as much as the circumference (40000km) of the earth at 70% lightspeed it shouldn't take more than 180ms. The lowest possible latency between Australia and Europe would be 60ms. Congested networks and network routers with suboptimal routes introduce far more latency than the distance. I regularly get between 100ms - 200ms latency between California and Germany and the latter number includes the realtime encoding/decoding of the video stream.
Internet gamers are fine with double-digit millisecond latencies, and this is what Starlink will offer. It takes less than 4ms to reach 1000km LEO orbit, so 16ms for the full round-trip (you -> sat, sat -> server, server -> sat, sat -> you). Make it 2ms ground-to-satellite and 8ms round-trip if you believe the model from this thread[0] (satellites at ~600km). Using its numbers, add 2x 2ms for each inter-satellite link you have to travel back and forth.
It still adds up to sub-100ms latencies. It ain't bad.
With the right type of architecture, low-earth satellite could offer better latency than terrestrial.
terrestrial internet goes through optic fiber in which the speed of light is maybe half the speed of light in a vacuum. Already HFTs use microwave links to shave off a few milliseconds between New York and Chicago. From Boston to Seattle the extra 600 miles up and down from the constellation would be overcome by the fast path along that route.
Note that requires switching from one satellite to the next, not the "bent pipe" architecture to your local rent seeking wireless company that StarLink and other LEO constellations plan.
I’ve read the paper that started this idea of Starlink being a low latency alternative for hft (http://nrg.cs.ucl.ac.uk/mjh/starlink-draft.pdf) There are a couple of problems with the paper. The first is that they establish a range of acceptable latency as existing between the great circle route and the fastest ping time on the internet between two points. For example they found that NY to London was 75ms round trip on the internet. However the long time standard bearer for a pure fiber path had been 64ms until the latest gen cables arrived in 2016. At that point the latency was sub 60 ... very comparable to the 55ms great circle route. But it’s important to understand that the starlink point to point route will be constantly changing — so only in one or two instances will that latency even come close to the existing terrestrial routes. More often than not the latency would trend closer to the 70ms range - well beyond anything considered competitive either in the current generation of routes much less the prior generation. It’s as if the authors simply had no idea about the hft market and what latencies might be needed (all of which has been published publicly). But there’s another problem with the study. For the solution to work, spacex would need to dedicate one of its laser links to enable communications across other satellites and dedicate that capacity to this application. And further, the study did not address the issue of optronic delays within the satellites. Most microwave manufacturers care little about hft latency tolerances and intermachine delays can be 1-2ms per hop. Assuming starlink to not to be much different (although I don’t know this to be true or not) that would definitely add even more latency to the equation. So all in all while it’s interesting and will do amazing things for consumers and enterprises, I don’t see starlink being a competitive hft solution unless the route in questions is really long and a lot of other engineering stuff can be made to work to enable it. Call me a skeptic but I don’t think we will see it.
I’m just pointing out that the slice of consumers who are a) internet gamers b) concerned about Starlink latency and c) still think Starlink is old school geosynchronous satellite. Gamers will of course worry about latency just like they optimize every other aspect of their setup, but not to the effect that SpaceX needs to mount a marketing campaign to that effect.
Yeah, and they'll care enough to actually find out what the latency is rather than reflexively shooting things down. And don't forget word of mouth. All it takes is one prominent streamer that you follow, or anyone in your friend group, saying good things about it -- and now it's on the table for you.
I don't think you're allowing for the the temerity of terrestrial services. They will conflate the poor performance of legacy satellite systems with this system. They will exaggerate the latency deficit over short paths and allow people to think this is the general case. They will do this to leverage the preexisting perception of poor satellite service performance that so obviously prevails.
Only a solid marketing campaign can deal with this problem.
Another thing is jitter. By design flying satellites are going to create jitter, and even with good latency, it can be bad for stable video streaming, video calls, etc.
That's true and bandwidth is actually something I've been wondering about: what kind of bandwidth can an inexpensive ground-based system get, reliably? Sending bits in a fiber is famously free of interference, while hundreds of miles of atmosphere is notoriously bad as a transmission medium. I haven't seen any kind of calculations around this. How many clients can hook up to one satellite concurrently, what kind of bandwidth will each get, realistically (theoretical max bandwidth is not so interesting). Without more information, my first thought has been that Star link sounds like an excellent, satellite-based internet technology ideal for connectivity in isolated places (far from urban areas, on ships and airplanes, etc) but which may not really be competitive for consumer internet access in urban areas. Hope I'm wrong.
> what kind of bandwidth can an inexpensive ground-based system get, reliably?
Theoretically 56 Kbit, but practically closer to 40 Kbit. I'm being serious! Given the choice between hugely expensive and high latency satellite Internet and dial-up, many have opted for dial-up, even in this day and age! We're not even talking that far away from urban centers, either, especially in hilly areas. A couple hour's drive outside of San Francisco is enough to get to cell-phone network dead spots, never mind LTE. There are some hyper-regional wireless ISPs using 802.12-based gear (Uniquity), but those are the exception.
Given that, (which is so extreme that I'm sure that not everybody that reads this will even believe me), any competition to HughesNet will be very welcome. Prior to SpaceX, the cost of getting a constellation of satellites into orbit was prohibitively expensive, but with this being done by SpaceX, the launches for this will be done "at cost", (corporate accounting and cost centers notwithstanding). (Which, mind you, is still several million dollars per launch in rocket fuel.)
Hopefully that means a genuine competitor to HughesNet, but that's a low bar. There is a large initial investment (of time and money), but FCC approval is an important first step towards providing this service.
The atmosphere shouldn’t be too much of a problem. It thins out very rapidly over a few km altitude, and the beam is going almost straight up. It’s nowhere near as bad as the same distance through the atmosphere at near ground level.
I am excited about the possibility and think it will (eventually) sell itself though word of mouth (how exactly do you advertise better Internet access to a market with poor Internet access? Billboards and USPS mailers?). I doubt most consumers have the background to know difference between latency and bandwidth, so hopefully this is price competitive with HughesNet!
I think that the key competitive advantage is exactly how cheap it is to acquire new users. Anyone that is off the main grid can still get decent internet without signing up to a long term landline contract. This removes one of the big barriers to entry and changes the industry as a whole.
At some point Verizon and AT&T will get the service requirements relaxed so that 5g counts as servicing a home. At that point they’ll run through and service all of the old rural installations. Hopefully the rural power companies do the same with the fiber they run inside their steel lines. There’s my guess anyway.
There is a reason for this as there are a lot of customers in on place. Satellite based internet allows you to have your infrastructure where you need it, even this requires a close control of your orbits. And then question is whether satellite internet will be a) faster b) be more reliable and c) cheaper than ground based fibre and 5G. If yes it might give these two alternatives a hard time even in metropolitan areas. If not satellite based internet will only be interesting for rural areas. Not sure if the potential higher prices off-set the smaller customer base. And in developing countries higher prices are basically off the table anyway.
Satellites still are in space, so everything is more expensive up there, getting things up, repairing things, building things for space, operating things in space...
I would say there are a lot of people who live in cities/towns instead. You need the scale for things to work. I hightly doubt this project is only aiming at people not living in cities.
VOIP. 150 ms max recommended, noticeable degradation of UX at 250ms. Many phone operators use wifi to connect when they can. If people's phone does not work well, they'll complain.
Sorry for this shallow comment but I've read the title several times now, my mind races to the same thought:
>...to sell wireless high-speed internet from space!
"I wish them wild success, it's promising to see the opportunity for disruption of our telcom mafia as wireless comms tech evolves, but selling internet from space... total horseshit! Success or not, most space internet subscriptions will be sold from a call center in India." /s
Please consider my apologies. I was taken aback by you're reply. It was not my intention to express bigotry but I thoughtlessly offered an unintelligible attempt at poor pedantic humor absent of context and comedy. You replied quickly and so I choose not to delete it.
I enjoy watching this technology as it moves to become an available option. The article is 9 sentences of an information scarce announcement, but as is often the case, I returned here to find plenty of useful information in these comments. Interestingly, ATSC 3.0 which is packet based and already overflowing with DRM now looks to be nearing a consensus. As early as 2020 (US) it could prove to satisfy enough content distribution traffic to reduce some downstream bandwidth demand and relieve congestion. SpaceX in partnership with ATT or CenturyLink for dedicated up-link over DSL would be an interesting offering.
https://www.atsc.org/wp-content/uploads/2018/01/A360-2018-Se...
