Hacker News Reader

Comment by guardiangod | Hacker News Reader

>Even with that boost, the service still only delivered downstream rates of 90.55 Mbps, leaving it well short of the 144.2 Mbps median posted by the industry as a whole.

>All of this data begs the question: What happens if Starlink can’t meet the 100/20 Mbps speed obligations attached to its Rural Digital Opportunity Fund (RDOF) projects?

Am I the only one who raise my eyebrows? What's the median for wireless internet delivery? What does that has to do with RDOF's completion requirement (by 2028)? How is StarLink doing at this point comparing to the other RDOF winners in terms of deployment?

From the actual Ookla article-

>while Starlink download speeds were slower than fixed broadband for all providers combined in the U.S. (144.22 Mbps) and Canada (106.86 Mbps).

Oh 144.22 Mbps is the industry median of _wired_ broadband. Well no shi* wireless is slower. Starlink can always deploy more satellites to service the demand. Is this a hit piece on Starlink?

View on HN

SpaceX chose to target the 100/20 benchmark, and bid on that, and because of that, their success is measured against that.

The part that is a bit hit-piecey is this quote from Viasat:

“even if SpaceX were to deploy a full, 4,408-satellite Starlink system, that system would fall short in satisfying SpaceX’s RDOF commitments.”

The full Starlink satellite system doesn't have 4,408 sats. That's just the Gen 1 constellation. They have committed to completing the Gen 2 constellation, with an additional 7,518 sats, which each provide many times more bandwidth than the Gen 1 sats, by 2027, well in time for the RDOF.

They're beaming almost 100mbps from the space to surface of the earth, to us, mere consumers, and they're the worst because they're slower than some wired solutions?

I'm seriously worried about the pollution they create for astronomers, but that's seriously absurd.

Do I need to repeat? They're beaming this from space, to your home.

And not just to your home. I'm very strongly considering ponying up the money for Starlink to be able to work while camped out in the middle of nowhere.

The receiver doesn't seem to be easily portable, say on a backpacking trip or something. Perhaps that will change in the future though.

True, but I've got a pickup :)

I'm expecting an official cybertruck-bed-based antenna dish.

The RV receiver fits in a backpack, though you'll need to carry a fairly heavy battery to power it for a while.

"fits in a backpack", with the implication that it will consume the internal volume of said backpack. That does not make it a solution for parent's "backpacking trip", unless you wish to sleep on the ground in the clothes you wore. And you ate before your left. The "heavy battery" was just icing on the cake. You know some backpackers saw the handles of their toothbrush to impress other backpackers^W^W^Wsave weight, right? :-)

That's not to say Starlink or a competitor can't get there eventually. I already have pocketable/backpackable device that can shove data up and down a satellite link (Garmin InReach), it's just dog-slow, and that's likely down to the antenna and transmitter power if I were to guess. Someone with more RF smarts than I can probably cook up a foldable antenna, perhaps boost the power a bit, I dunno. But the PoC is there, it just needs iteration.

This is more of a physics problem than an engineering problem.

Garmin InReach uses an omnidirectional antenna on the Iridium network. With an omni antenna your S/N ratio is simply too low to send bits fast. If you made the device powerful enough to roast birds in flight it might work for faster uplink but then it would be dangerous to you too and the FCC would never let you use it anyway. And even this wouldn't improve your downlink speed.

The only way to improve the speed in both directions is to use a directional antenna, which points most of the uplink energy toward the satellite and concentrates the satellite's incoming energy to get the S/N ratio high in both directions. That's exactly what Starlink does.

This is made even more difficult because we're talking about LEO satellites and LEO satellites move. So the antenna has to track the satellite. Again this is what Starlink is doing.

(You can also use GEO satellites which don't move, but then latency becomes terrible. That's how the older satellite internet systems work.)

It's certainly possible to build a portable, directional, tracking LEO satellite antenna. That's exactly what the Starlink RV solution is. But it's never going to be as small as your Garmin unit.

Friend of mine does that and works from ultra-remote southwesterns deserts when so inclined.

Mine is arriving today and I'm SUPER stoked. Bye AT&T forever!

I do this. Anywhere I can drag my trailer. I have a favorite location up in the mountains that gets no signal of any kind, but now that I have Starlink I can get perfect Internet there. So instead of going up for a weekend to relax, I go up for a week at a time, work during the day, and come home on the weekend to top up supplies.

Yep, that's exactly the sort of routine I have in mind.

Longer-term, I'm hoping for that favorite location to be a piece of land I actually own, in which case my goal is to setup a semi-permanent "base" with solar, Starlink, and a yurt or somesuch.

Same, I got a setup before I went out west this year and I can now camp off grid at places I couldn't the previous years. It's great.

I'm serious that with rent costs increasing, starlink live roaming might see a real boost in live-aboard sailors

Until they have their inter-satellite links working, you and the downlink station have to both be visible to the same satellite.

It's not only the pollution. The world is "unwillingly" giving up space spots to a company that can use those satellites as technological weapons. What happens when a Chinese/India/Russia(Insert any non-Anglo "bad" country) decides to emulate them , I would happily bet my left kidney the _commentariat_ here would not be so happy.

If Russia wants to use satellites as technological weapons, I don't think they're going to ask the FCC for permission.

Would they provide a usable broadband alternative to the broomstick-up-butt treatment I get from the telcos?

Hell yes I'd take them. Oh they'd steal my data? Well, so does everyone on this site and the telco I have now.

Consider this, Mr Geopolitical power. Do you want the US with this weapon, or sit on our ass and let China deploy one now that SpaceX has shown how powerful it is?

This is the single greatest Western soft power weapon since the iPhone. And you're begrudging it? Come on.

I am latin-American Mr "I think Anglo is the world default", in our region the US has been a much more pernicious force in history than China,by several orders of magnitude. I understand we are just simple apes that love to think tribally (the logical part of our brain is shut off) but I would expect one is aware enough to understand there are other "tribes", specially when worldwide one's group does not reach double digits as % of population.

Playing the "Holier-than-thou and I think Americans are assholes" card combo by itself is not countering their point.

It is correct that Starlink is not motivated to act for your benefit, any more than for this individual. Your interlocutor has long experience with such things. So do we, really; are Starlink's incentives aligned any different from AT&T's?

A few years back, AT&T took a half $billion from US taxpayers to build out fiber to rural neighborhoods. One comes to within 1/10 mi of my house. But they were not obliged to light any of it. They did not. That fiber sits dark, more than a decade on.

This is not the US. This is a privately held corporation answerable to nobody. Whose benefit do you imagine they will exercise their power on behalf of? Yours? US voters'?

> This is not the US. This is a privately held corporation

Just like all other ISPs. As well as tech companies.

What's your point? That Starlink is bad because it isn't a government-owned entity? Or are you opposed to the idea that a non-government entity can provide a strategic advantage/soft power to the US government?

Asking because neither of those two points seem to be able to withstand much scrutiny.

My point is simply that anyone expecting Starlink will treat you better than AT&T does, and to act on behalf of the US, is very unfamiliar with both history and Elon Musk.

A hell of a lot of Tesla customers paid $thousands for self-driving, since, what, 2016? Tesla just laid off its whole self-driving department. What are the odds they will now issue refunds? Care to bet?

> A hell of a lot of Tesla customers paid $thousands for self-driving, since, what, 2016? Tesla just laid off its whole self-driving department. What are the odds they will now issue refunds? Care to bet?

Refunds for what? Every feature I bought the car with is still there. Even got the FSD beta access last month after being on the waitlist for a while. Updates still keep getting pushed, fixes and adjustments still get deployed on a regular basis. The self-driving program isn't ending, the development is still ongoing and alive, so I am not sure why I should be expecting a refund for that.

Also, get your news right. The people they recently laid off were mostly working on labeling data, they were not actual self-driving engineers. Labeling data is a pretty common and manual type of a task. I am actually surprised they've been doing it in-house for so long, as opposed to contracting it out. Not trying to justify that layoff, but claiming that it indicates the end of their self-driving program is ridiculous at best (and disingenuous at worst).

> anyone expecting Starlink will treat you better than AT&T does, and to act on behalf of the US, is very unfamiliar with both history and Elon Musk

You can make that statement about literally any company, and it is always a fool's errand to guess instead of evaluating that specific thing in question. You can say the same about Google, but I've been a happy Google Fiber/Webpass customer for the past few years. I can say nothing but good things about it, after many years of terrible experience with Comcast/Xfinity, and a few much better (but still worse than Google Fiber/Webpass) years with WaveG (WA area fiber provider).

In this specific case with Starlink, you don't need to guess either. If you live in an area where they provide service, you can sign up and compare it to any currently existing internet solutions for areas with no high-speed internet (rural and far-from-major-cities areas), which is the primary use case that Starlink was designed for.

Back to the original point about "private tech company as a soft power for the US government is a ridiculous idea." In the present day, we got Ukrainian Vice PM tweeting[0] his expression of gratitude towards the US government and Starlink for providing connectivity during their current crisis. What is this if not a proof of that soft power US can exhibit through private tech companies?

0. https://twitter.com/FedorovMykhailo/status/15327439910658170...

Either you paid extra for FSD, or you didn't. If you didn't, you won't get it. Everyone who did still doesn't have it, despite public statements ("next year, for sure") every year since 2016.

A mass layoff sends a message. Everyone left, there, certainly got it, and are updating their CVs. (I wonder what they will say they have been up to.)

You do get self-crashing, anyway: You can watch those frame by frame on YT.

And, yes, the same is true of any big corporation. But many people have announced in public that they do expect different, and need the correction. Comcast, the "Most Hated Corporation in America", is a curious choice to have substituted in for AT&T.

Anyone who imagines that Starlink did not get back way more than full value for that Ukr gesture knows nothing of PR.

> My point is simply that anyone expecting Starlink will treat you better than AT&T does

The situations are totally different. AT&T made money from LOCAL infrastructure that they had a natural and enforced monopoly on.

Starlink infrastructure is inherently global, their main goal is with minimal effort and cost to connect as my people as possible to distribute the global fixed cost. SpaceX has little intensive to involve itself as deeply in local politics. On a global scale some politics will go in their direction, others away, spending lots of money in each government is not really a very practical strategy.

The US government can force companies to do things sure, and if you have extreme security consideration that goes beyond having a VPN go threw Starlink then Starlink is not for you.

Starlink will not have a monopoly on anything, they have an essentially global outlook and they are trying to offer a uniform standard service from Stockholm to Santiago. So I do actually think its a fair assumption that costumer experience has a chance to be better.

> Tesla just laid off its whole self-driving department

You realize that is totally wrong right?

If you really think AT&T's profits are from their landline dial tone service, I don't know what to say.

> Starlink will not have a monopoly on anything

Numerous people have posted right here, contradicting you. Maybe read them?

> Numerous people have posted right here, contradicting you. Maybe read them?

They are wrong. Its simply factually wrong. Starlink will coexist with many global LEO constellations, many are already in planning.

Those other constellations do not, in fact, exist. Until they do, for most rural subscribers Starlink is the only game in town: a de facto monopoly even by strict legal definition, whether or not so governed administratively yet.

> Tesla just laid off its whole self-driving department.

I had to check since I hadn't heard of this. All the sources I checked were about 200 low-level employees who were mostly doing manual annotation of images to help improve the algorithm. Not developers themselves. Definitely nothing about laying off "the whole self-driving department".

From https://techcrunch.com/2022/06/28/tesla-layoffs-autopilot-wo... :

"Most of the workers were in moderately low-skilled, low-wage jobs, such as Autopilot data labeling, which involves determining if Tesla’s algorithm identified an object well or poorly, according to one source."

"Most of" is carrying a mighty heavy load, there.

They are closing the entire San Mateo unit. If I were among the "self-driving" staff remaining, I would be updating my résumé.

Curiously, all those dumped were "terminated for underperforming", which will make finding work elsewhere hard, unless other companies see it as a cynical ploy to dodge regulations on normal layoffs.

I don't understand the significance of the San Mateo city in this context. I don't work in this industry. Were there higher level programmers included as well, instead of work akin to Amazon Mechanical Turk?

Do you have a source for the underperforming thing? That'd be next level incompetence for them to lay off 200 people for that reason all at once. (not saying it's impossible obviously!)

Same article you linked (tx), at the end. So, you know as much as I do. Before, I only knew the headline: closed office, flushed staff.

Agree, abusing labor law like that would be pretty scummy. Is it "incompetent" if it costs less? Depends on who is getting in trouble.

US corporations are significantly controlled by US laws. These laws can do things like stopping them serving enemy countries. (The recent Russia sanctions are an example)

That is power.

I think I see your point, here: that, being a US company, the US can give it directions regardless of what the company would have chosen to do; and even in places the company is not compelled by law, it may be in their interest to stay on good terms with the feds where for a company based elsewhere it might not.

> space spots

What does that even mean? You can't see them unless you have highly speziallised equipment.

And in terms of physical spots, space is absurdly huge, space is not really a limited resource.

It seems like they're keeping a ~10 km buffer zone around each satellite which does add up. Between Starlink, OneWeb, and Kuiper a bunch of shells have already been taken.

Those are current technological limitation the same way Montana farmers in the West didn't have barbed wire.

Tracking and control technology is constantly improving. Just like as with airplane safety humanity will improve with increase use.

Starlink chose the 100mbps threshold over trying to satisfy the lower speed threshold because the higher threshold makes their business plan viable (it lets them get $$$ government funding for providing high-speed rural connectivity).

They don't get kudos for failing to meet the standard they're required to meet as a condition of the government program they need to make their business work.

They built and shipped something really cool and useful. That seems like it deserves some kudos.

> The full Starlink satellite system doesn't have 4,408 sats. That's just the Gen 1 constellation. They have committed to completing the Gen 2 constellation, with an additional 7,518 sats, which each provide many times more bandwidth than the Gen 1 sats, by 2027, well in time for the RDOF.

Starlink took govt funds and is obligated to deliver 100/20 at a minimum.

I own 2 starlinks at different locations and I can tell you 1 of those locations is consistently MUCH under that.

That being said, I am pleased with starlink at one location as it absolutely blows away the next best available option.

The regional speed differences are due to Starlink current reliance on ground stations and terrestrial backhaul. If they can get sat-sat links working and transition their backhaul onto those links that will no only eliminate the regional performance differences but likely also be faster than any terrestrial options over long distances.

I'm pretty sure you have it backwards; the current configuration of going through one satellite is always going to be faster [edit: higher throughput] than going through multiple satellite hops.

It's possible for it to be faster over long distances because light travels faster in vacuum than in glass fibre and likely less hops in space vs terrestrially where transit would need to go via several carriers to reach far destinations.

That is putting aside ground stations located in areas which themselves have poor transit which is the main problem with regional variability of performance I would imagine, though could also be congestion if they are oversubscribing in those regions. i.e too much density of terminals.

Lower latency is very, very far from the same thing as more bandwidth. Certain customers can benefit from the difference in latency. Not you, though.

Starlink will make those who would benefit (hedge funds, mainly) pay through the nose to get it. And, pay even more to keep somebody else from getting it. Probably they will charge a huge premium for 1ms faster than fiber, and 100x more for each ms better than that.

You may be certain they will not use that extra income to provide you better value.

Their stated plan is to enable the inter-satellite links for all customers. This will allow them to reach many more customers with lower operational costs. This would probably allow for more profit than just charging a few customers a premium for lower latency service.

Enabling the inter-satellite links for all customers doesn't guarantee they'll use much more expensive cold-potato routing though. It makes no sense to cross the ocean on 20 Gbps space lasers when you could use a 20 Tbps fiber cable.

Right. Any traffic that can be dumped to the downlink will. But it anyway costs nothing but RAM to stall your packets just enough to satisfy your service level agreement, keeping the channel clear for actual paying traffic.

Sending early is of course always OK. Those paying for that millisecond need it reliably every time, so not much of your traffic would need to be stalled. Maybe 10%. And maybe only while markets are open.

I 100% agree with this.

Sat-sat links only save them some landline bandwidth cost, allow them to serve airliners over oceans and deserted areas, and enable lower trans-continental latency for select super-high-paying customers (i.e. hedge funds and maybe military). They do not improve bandwidth overall.

It does (or at least can) improve a bandwidth of any give satellite due to load balancing.

Let's say you're in SF, served by a satellite that is currently over SF.

You're talking to a server in NY.

Your data has to go up to a satellite and come down from satellite to a base station.

Currently this uses 2x bandwidth for this one satellite.

But let's say SF satellite is maxed out but there currently is a satellite over Nevada desert with a spare capacity.

With inter-satellite links and proper routing, your data could go up in SF and come down in Nevada dessert, making more bandwidth available in SF.

The satellites have a full coverage of US area but there are large parts of land where there's very little use of available bandwidth.

If SpaceX builds enough base station scattered around US, they could almost 2x the bandwidth available in areas with high density of customers.

Limiting bandwidth is the downlink to terminals. Probably the birds will have edge service contracts to proxy content, saving on uplink from hubs.

They might even be equipped to broadcast identical real-time content (e.g. world cup) once for all terminals following. (Anyway I would make that work, in their place.)

Anyone who has lived with rural internet will be blown away by 90MB. This is an incredibly high bar.

They will, and they are judging by reviews.

It’s a bit nitpicky but that B needs to be a lowercase one.

It’s an incredibly high bar that SpaceX explicitly said it would meet in order to receive more government funds.

It’s not an arbitrary threshold.

> that SpaceX explicitly said it would meet

By 2028? So where exactly is the sky falling?

Sky is not falling, people just make a big deal because its SpaceX and Musk.

Well, the article makes the point that Starlink is definitely not under anything close to what it’s expected “peak load” will be, and even so, it’s not currently meeting speed requirements.

But mostly, I agree — the article seems a little over-worried.

I didn't know why people are getting bad speeds, because I'm regularly exceeding these speeds on my dish. Starlink requires self-installation and I'm willing to bet that a lot of people don't understand how to set it up properly. I will say it took me over 2 weeks to really get a good location, top some trees and set up a proper mount above the house. People post pictures of their dishes just sitting on the ground, and a lot of people don't have patience or time to get the connection working well.

Even in Central London (UK) that's fast internet. I can't get 90MB in Zone 1

through your phone line

I presume you mean 90Mb, if it is MB I am envious.

Check out the section "Technology-Neutral Service Tiers" https://www.fcc.gov/auction/904/factsheet SpaceX bid for "above baseline". They still seem to be doing much better than other auction winners though.

Ookla's speed test shows the entire network, not RDOF subscribers alone. It should be easy enough for Starlink to give RDOF subscribers the slated speeds. So yeah I think the article is click bait. Especially since Ookla showed improvement year on year for Starlink.

Setting aside the high cost of growing the constellation, I'm not sure they can just add more satellites even in theory. As I understand it Starlink has performance tradeoffs wrt to density of ground stations. Others on HN would probably know more though.

Honestly, I'm surprised that managed 90Mbps. Beaming internet from satellites iis not really a great idea because you are essentially doubling the wireless bandwidth required for a given level of service. Wireless cell service has the big advantage of only needing to use bandwidth for the link from the tower to the client while satellites require a link to the client and a link back to the ground station. There's also the fact that putting a bunch of compute into a satellite is going to be way more expensive due to weight, power and cooling in space not to mention increased cosmic ray interference.

It's highly directed radio signals, though, so it's spatially separated.

Doesn't help, really.

Sure it does. The most directed RF signal is a wire...

No. A separate wire has its own transceiver. They still have to share the transceiver among all subscribers. Beam forming saves on transmitter power and receiver S/N stats. If it were all just for you, that could improve your bandwidth. But it is not.

Lots of people would like you to think otherwise. Now you know better.

Beamforming phased array MIMO antennae certainly can transmit to and receive from multiple stations via spatially distinct beams in parallel at the same time. I don't know if Starlink satellites do so, but it is technically possible.

If you imagine each gets its own on-board transceiver, I have a bridge with your name on it.

No need for a bridge, I have signal processing products with my name on them, thanks. I'm familiar enough with the field to know how such things work.

An active multi-peer beamformer uses a single transciever complex using mathematical signal processing to transcieve multiple data channels to peers in parallel.

Some types of phased array antenna offer multiple focus points at the same time, but the electrical signals must pass through a transceiver complex for the different components to be combined and separated mathematically. In simplified terms, matrix multiplication is used to transform between parallel data channels and parellel antenna components; then the antenna geometry focuses them approximately into parallel beams.

It's not a transceiver per peer, but it behaves a bit like one for the set of focused peers at each time slot. Spatial separation by this method increases aggregate bandwidth, and therefore increases individual peer bandwidth when there is congestion.

In principle you could say each array element has its own transceiver. But the computational load goes up with the number of distinct beams, and tight enough beams to address each subscriber individually would anyway need an array orders of magnitude bigger than the actual satellite. So instead we have binning by solid angle and frequency, and also TDM and software-level addressing. Some of that must be what makes uplink data rate so much lower than downlink.

Point is just that practical limits make things rather worse than what a subscriber might naïvely hope for.

It appears that every satellite with spot beams does that.

I'm on T-Mobile's home internet. With just the modem chilling in my room, i can get about 200/20. I've bought an external antenna that I'm going to be hooking up to the modem and mounting on top of my house soon. Testing it in my front yard I was able to get 350/25. Latency/Jitter was also significantly better with that antenna.

I'm not super rural, but we have literally no other internet options, are on septic/well water, and are about 10 minutes from the nearest store by car.

I'm in a similar situation. Have TMobile and Starlink RV. Right now, I think TMobile hits faster at its best, but the normal experience is much worse. Sucks, as it is much much cheaper.

Heck, my upload with a fixed broadband provider in the US was 10Mbps! Starlink's cited performance would match that.

>What's the median for wireless internet delivery?

Who cares? If they signed up for

>100/20 Mbps speed obligations attached to its Rural Digital Opportunity Fund (RDOF) projects

they have to deliver. The article has a valid point.

They have to deliver that in specific areas by 2028 not nationwide today.

Assuming they avoided bidding on areas their network is currently contested they might already meet their obligations.

Ya, I was thinking the same thing.

This article seems like it's written by a PR/lobbying firm for BigTelCo / Satellite providers.

Not really, they are being subsidized by US tax payers to provide a service and they are not meeting the terms of the contract/agreement. None of Elon's empires would be successful without US tax payers so I think the article is merely trying to hold them accountable.

No, they haven't violated any contracts...

They have until 2028 to meet the 100/20 numbers in specific areas. Ookla's testing was not specific to those areas, and was not done in 2028.

But let's not let facts get in the way of irrational anger

It's not anger but I think the article (while probably written by a competitor or lobby) is reminding the public that Musk often makes empty promises and we should not let him continue to be subsidized by the US tax payer for an inability to deliver on time and on budget. I now see Musk also complaining about 5G interfering with Starlink. It's all a game, which is fine, but I dont want to pay for billionaires entertainment.

This is a hollow argument since none of their competitors that also are faring any better either. SpaceX is delivering close to their promise now years before they have to. It's an alarmist piece by teleco like the clock is close to timing out and things are heading south. Over time they will improve the hardware to have higher bandwidth capability, such as with their v2 satellites, with further improvements to the hardware likely later on.

> dont want to pay for billionaires entertainment

It's not "entertainment". It's actually delivering value to underserved customers, there's no need to inject politics or personal vendettas into this.

>This is a hollow argument since none of their competitors that also are faring any better either. SpaceX is delivering close to their promise now years before they have to. It's an alarmist piece by teleco like the clock is close to timing out and things are heading south. Over time they will improve the hardware to have higher bandwidth capability, such as with their v2 satellites, with further improvements to the hardware likely later on.

I hope they improve it, but Musk has a strong history of over promising and under delivering. Time will tell, I do not think this was a alarmist of fluff piece. Merely a reminder of their promises and current results.

> It's not "entertainment". It's actually delivering value to underserved customers, there's no need to inject politics or personal vendettas into this.

This is arguable. Musk depends on tax payer funding to keep SpaceX afloat to get to Mars. Nothing about politics or vendettas, but this is a billionaires joy ride. The US government also has a history of spending money to help the under served which has historically never panned out either.

Reply on news.ycombinator.com