Starlink is a big deal | Hacker News Reader

Starlink is a big deal

caseyhandmer.wordpress.com

134 pointsmartythemaniak111 comments

This article seems to be trying to get to a conclusion without a lot of support.

>This works out to be around $100k per satellite, more than 1000 times cheaper than a conventional comsat launch

A standard comsat satellite can potentially cover 1/3 of the globe with hundreds of Gbps of bandwidth. Starlink covers a tiny area with much lower bandwidth. Strength in numbers.

> 100MB per second using advanced coding such as 4096QAM

QAM isn't used in satcom, and certainly nowhere near 4096. Why the assumption?

> Each Starlink satellite includes all the complicated electronic switching gear found linking optical fibers together

Nope. The first batch likely had a problem and won't have inter-satellite links (per the hint from shotwell)

> All up, 2500 channels each supporting 58 Gbps is a staggering quantity of data, roughly 145 Tbps

The MIT study showed simulations that was about 20x less than that, so I'm not sure where these numbers are coming from without doing a proper simulation.

The piece that was left out, and possibly the most important, is user terminal cost. It will certainly be more expensive than a GEO antenna, and you need to recover that cost from the user. So far, there have been no practical, cheap, high-performing phased array that I know of.

> So far, there have been no practical, cheap, high-performing phased array that I know of.

Do you thing that is more due to technical limitations or more due to relatively small markets for such antenna that perhaps has not led to economies of scale?

I think technology limitations. Directtv/dish would have used it already if it was cheap enough since it's easier to install.

They would only have done it if an off-the-shelf solution was already available.

StarLink is going global and the target price for their terminals is $USD200 or so.

That's pure speculation. There's nothing special about starlink for buying power. Anyone would have done it if a cheap solution/off the shelf was available.

The point is not about buying power, it’s about potential income versus up front development cost.

That's the exact same thing every company in every business has to solve. What makes starlink unique in this case?

The up front development cost is slightly lower than a few years ago, and the potential income is astronomical.

The antenna cost has been confirmed to be targeting a price of 300$ per unit.

>The antenna cost has been confirmed to be targeting a price of 300$ per unit.

Like the Model 3 was "targeting" $35,000 before incentives.

Right, so targeting is one thing, and $300 will certainly be a good price point. But we'll have to see if they hit it. Many, many others have tried and failed.

What are physical limitations? With phased aperture array we have 1) individual antennas 2) individual electronics for each antenna. Which of them is expensive?

To the best of my knowledge, they can make cheap phased array antennas that have poor performance. Poor performance can cut substantially into the overall satellite capacity, so the only way to make up for that is to charge more money. Then your service is less attractive, and people won't sign up. So you really need really well performing, and inexpensive.

> will certainly be more expensive than a GEO antenna

I don't think this is necessarily true. Iridium antenna and RF circuitry is much cheaper than GEO because it doesn't have to receive a weak signal from far away.

That's not really what drives cost in this case. That's a very small part of it, but the phased array aspect alone is expensive based on frequency, which they're using the same ones as GEO.

The first batch didn’t have inter-satellite comms onboard due to the mirrors being too robust and presenting a risk of surviving reentry.

How can this be cost competitive against cell towers connected to fiber backbones? This seems like the obvious future, as towers are already being built, upgraded, and replaced due to the explosion in data demand for mobile phones.

For Starlink, launching satellites are both incredibly expensive and hard to maintain. And likely they will have much lower bandwidth than cell towers, and the uplink hardware will be much more expensive than cellular hardware.

Right... I have traveled all over Vietnam, Cambodia and Laos by motorbike. Been in the deepest areas of nowhere, 100's of KM from anyone or anything. I had solid 4G connections everywhere except the deepest valleys, because the military has installed towers literally everywhere.

It's not about wealth it's about population density. Think not about dense areas like Vietnam but sparsely populated places such as Australia, Canada, Russia, etc.

Of course the problem with this is that there's relatively not many people who live in the middle of nowhere because most people want to live near civilization. So I am doubtful that starlink will be an economic success.

None of the 3 countries I mentioned are densely populated, especially in the areas I'm talking about, and especially Laos, who get all their telcom infrastructure from Vietnam and China.

If you aren't on the coast, there's a whole lot of empty space in Vietnam. Same goes for Laos and places not bordering the Mekong.

I lived in Laos back in the late 70's/early 80's. One of the big treats we got was 30 minutes of phone call time back home to the states on Christmas. Barring that, it was 3-month-round-trip mail via diplomatic pouch.

Now I can hop on to any number of free webcams in Vientiane and check out the happenings where I used to live just for the heck of it, no questions asked, no money required.

Sometimes it still shocks me how much the world has changed in the past few decades!

I bet you have some amazing stories. Have you made it back at all?

Vientiane is all grown up and becoming a bit of a popular expat area. Vangvieng is super popular with backpackers and Luang Prabang is a bit of a tourist trap, but is still tiny and really beautiful. I spent about a week just hanging out in Nong Khiaw... tiny little town on the river, super relaxing.

The infrastructure is changing a lot too. Roads are becoming real roads. Chinese are moving in and building roads and hydro dams all through the country. Bought up all the land from the locals. Really a total eco disaster that nobody is talking about.

Oh, lots of stories, for sure. Being one of the very few foreigners around during the water festival meant is was 100% impossible to get home without getting utterly soaked, in the friendliest possible way.

And I remember clearly there was only one English sign anywhere in the Vientiane Airport, and it was a warning before the metal detector that said 'No Guns, No Knives, No Durian' in many different languages. More than one flight to Bangkok had someone casually cleaning their rifle to pass the time, and knives were shrugged at. They Meant It when it came to trying to take a durian on the plane, though!

I would very much like to return and see how things have changed. Things haven't aligned to make it possible for me, yet, but it's on the list.

Here's a small gallery from the National Day parade in 1980, if you're interested: https://imgur.com/a/GRmSl

Nice, thanks for sharing! =)

You should try rural southern U.S. Good luck finding a 4G connection.

If only the US military controlled the US telcos like the Vietnamese military does... oh wait. ;-)

Google suggests 150K minimum for a cell tower. Another search suggests that a Starlink satellite costs one million dollars, based on the assumption of a Falcon 9 launch and does not include any improvement in costs.

So, it's an order of magnitude increase in costs if we assume that a cell tower have the same capabilities and do the same thing as a starlink satellite, which it does not, since a cell tower is just a structure you can lease to host equipment on.

I wonder how much is the capital expenditure is for infrastructure spending versus the number of cell towers?

According to google, there are 307,626 towers in the United States. Another source gave me 55.71 billion expenditure for calendar year 2018 which also includes maintenance. Divide the expenditure by towers gives me 181,096 USD spent per tower.

So, 350,000 dollars versus 1,000,000 dollars. Hmm.

That cell tower only covers a small geographical area, and you have to physically maintain it along with paying property taxes. Also, that cell tower can't peer with other cell towers to route trans continental packets.

Not to mention how much will DOD subsidize their own sub-constellation for SkyNet to prime the cashflow? I'm guessing at least $50 billion.

A starlink satellite only covers a small area. That's why they need 4,000 just to provide bare minimum service.

Google says there are ~300,000 cell towers in the United States. Napkin math says the satellites appear to cover about an order of magnitude more ground.

But more accurately, it’s a question of how many birds are in the sky above you at a given moment, and how narrowly you want to focus their antenna based on the desired bandwidth availability per square mile. It’s the same idea with place cell sites closer together at lower power, except I imagine the satellite constellation design can’t be tailored around specific dense environments, and rather will provide a relatively constant global density, or constant bandwidth densities at certain latitudes.

Cell towers aren't a great comparison. They tend to heavily lump them in dense areas, and you have many companies who've built infrastructure in the same place to get to that number. SpaceX doesn't have as much control over covering more dense areas, short of launching more satellites.

Wireless backhaul is not uncommon between (among?) cell towers. Probably a cell tower won't have as many other towers in LoS, on the other hand, equipment on towers is easier to debug.

But cell towers require leasing land. This page says the average land lease is $45,000/year: https://blog.sior.com/cell-tower-leases-add-big-value-with-l...

Starlink satellites are expected to have a 5 year lifespan, so maybe tack on another $250,000 to the cell tower for a cost comparison.

According to the article, cell phones in cities are NOT a use case for Starlink. (I don't know if the problem is power or antenna size.)

Also, Starlink satellites have to forward traffic to/from ground stations that are close to the area the Starlink satellite is serving. I don't see the costs of those stations accounted for anywhere here.

Push on through the hand-wavey bit at the beginning and there's some treatment of your question. haven't finished it yet but it's a good read so far...

SpaceX want to make it. Incumbent ISPs don't, that's the difference.

What about iridium? They didn't want to make it?

I don't think they wanted to make such massive constellation like SpaceX, so I assume they have way more limited capacity. And their satellites are much bigger.

Sorry, my point was that they really wanted the tech to work financially as well, and it's the closest to starlink that we have today. Iridium remains one of those stories for how things can be far more expensive than planned, despite the difficult tech actually working.

The article explains the case of Iridium. Basically, Iridium is severely constrained by not having very cheap satellite launch capabilities.

Iridium focuses on applications that require low power use and a small antenna, but don't need much bandwidth. (Phones, telemetry, emergency text messages, ...) So indeed, they didn't want to make an internet backbone.

What about Teledesic?

More ISP competition would be a good thing, especially outside cities. But it's hard to see how Starlink's costs, even with cheap SpaceX launches, could ultimately be lower than land-based ISPs where last-mile fibre is already installed. I guess Starlink might be successful in markets where last-mile fibre is not available. So it seems like a good thing but Starlink making a lot of money will depend on land-based ISPs being useless.

I know ISP will use the Starlink network and then resell to consumers but I really hope over time there is some commodity hardware that people can use to directly access the network. It is very unlikely and would probably never happen but I can't help but think if people can use this somehow to bypass Government censorship like in Iran/China or use it to just get online in the face of complete internet suspension like in Kashmir.

Government will ban the use of such technologies.

And they will deploy technological means to ensure you're in compliance.

Perhaps.

I’m a huge fan of Tesla and SpaceX but I’m not sure Starlink is a “big deal” the same way that Starship is a Big Deal.

Starlink isn’t a paradigm shift unless the receiver can fit in the cell phone form factor.

In the meantime, Starlink is a great way for SpaceX to print money using the same infrastructure and expertise they needed to build anyway.

SpaceX needs launches to make the Starship viable economically. Starlink is a way to get those launches and may be a good business in itself to finance Mars exploration.

Smartphones will directly benefit from Starlink as suddenly is becomes much easier and cheaper to deploy cell towers.

Starlink is the key to near-global cellular coverage.

It's a Big Deal for folks who are behind Internet censorship, isn't it? No way for the various country firewalls to prevent these devices from working in their countries, is there?

Quite the opposite. Your equipment will be promptly detected and you'll be sent to gulag...

Most dictatorships forbid using satcom.

Elon has said that they'll allow Starlink partners to censor when they resell the service in their local market. Probably the only way that they'd be able to sell their service in that market.

A lot of people in rural areas — close to major metro areas in the US — have terribly limited internet options. Traditional satellite internet is a decent option if it works on your property, but there’s a substantial latency. Fixed wireless requires line-of-sight to a tower, and cell providers don’t necessarily have the desire to get towers in these areas in a meaningful number. Starlink could bring connectivity to a lot of people.

I would love to have a suitcase-sized wireless hotspot that I can bring with me on road trips. There are a lot of places in the US that I drive where there is no 4G, or coverage is very spotty at best.

Also, I work at home, and the main reason I don't work while traveling is that I can't depend on reliable internet. With a portable, reliable internet connection, I could absolutely see myself being more mobile, perhaps working from other countries more often.

Also, more options for internet connection at my house would be awesome since it would likely drop prices for internet in many areas.

I'll get excited about fast broadband when it's actually available and delivered to me and I do the speed test and it's as fast as promised.

There's so much talk about broadband and its potential and so little actually delivered.

Show me the speed test. Everything else is just vaporware.

We already have satellite internet. It is 100x further from earth and using ooooold technology, compared to the Starlink fleet.

The scale of satellites planned for Starlink is nearly 2 orders of magnitude more than all satellites ever launched. Current count of everything humanity has sent up is less than 2000, 40,000 is planned for Starlink.

The change in viable throughput and what it means to a globally accessible, high-speed network is worth discussing.

As I say, show me a product I can buy that delivers on the promise, then it's interesting.

This is completely false. Which new technology do you think starlink satellite have that others don't? They're actually significantly less sophisticated than many older satellites. The only difference is they're planning to launch such a large number of them, but that has nothing to do with the payload technology.

There are plenty of others that will/do give good throughout globally by that time, or sooner.

Quantity has a quality all of its own

Are the laser links used elsewhere yet? I remember reading about various people being interested in them, but this is the first deployment I’ve heard of.

There have been many filings in the past to use them, but only iridium came to fruition.

I’m most excited about the prospect of things like this delivering the truth via uncensored internet to places like China and North Korea. Am I naive to think there is no way they could stop this and that the real internet would be available world-wide?

China could easily require filters on any receivers sold in China and ban the use of uncensored receivers.

Yes, that's naive. You need landing rights in every single country you plan to put a signal down in. There's nothing special about starlink; plenty of other companies have satellites that could cover those countries, but are not allowed to due to landing rights.

Here's some decent info:

https://www.accesspartnership.com/navigating-through-the-reg...

Landing rights aren’t particularly relevant for NK, since it’s not like they can fine or imprison SpaceX. China is more relevant since they could retaliate at other US owned property or at specifically Tesla.

Nk isn't even an issue since they won't let you buy the antenna. If you did, you'd likely go to jail.

Black market exists, but NK isn’t far away from ASAT capability.

Black market for something that is clearly visible on your dwelling and can be seen by satellite is extremely risky. Black market for other things like movies make sense, because nobody knows you have them.

You'd need to flood those countries with cheap receiver hardware without it being intercepted by the government, and then provide connectivity to those receivers to local users over a decent size area - larger than your usual WiFi hotspot I reckon.

If there existed a local tech community with the chops and determination to successfully distribute the open internet delivered through this cheap receiver hardware to local users - all in an environment of very limited privacy - I'd be really surprised.

Maybe in China... not so much NK.

This is a great article. I don't think, though, that author adequately estimates counter-monopolistic actions. Those actions, I think, will happen if SpaceX even approaches the position which this article describes.

> because only SpaceX had the vision to spend a decade struggling to break the government-military monopoly on space launch.

This is completely wrong.

Will Starlink used to power Tesla Internet?

this was just posted the other day i believe

nope that was "SpaceX Starship is a very big deal", same author, same series https://caseyhandmer.wordpress.com/2019/10/29/the-spacex-sta...

> There are only three trillion dollar industries in existence: energy, high speed transport, and communications.

I would suggest that housing is a 4th trillion dollar industry.

And agriculture.

https://croplife.org/news/agriculture-a-2-4-trillion-industr...

and drugs

And defense.

and healthcare

And financ e

And food.

travel

Travel is, at least partially, transportation. But I'd add education.

and minting currency

https://en.wikipedia.org/wiki/Trillion-dollar_coin

Maybe slightly off topic, but Musk’s companies are likely to be the biggest human decentralizing force in history.

Leaving aside that Spacex is trying to decentralise at a planetary level, the solar roofs and batteries allow you to live anywhere in the world without an electric grid (water is generally easy to get with a borewell), Teslas will drive you anywhere (mostly) by themselves, and Starlink will give you high speed Internet. Given the way lifestyles are changing, I don’t actually think people of the future will cluster in cities.

Musk’s companies are likely to be the biggest human decentralizing force in history.

He's an interesting figure, but it's a little early to be eulogising him this way.

Right, because the reason i live in a city is the easy access to electricity, water, and high-speed internet access!

I don't know if you were being sarcastic or not but for me the reason that I live in a city absolutely is easy access to those.

Then considering moving to cheaper rural places where in 2019 all three are easily available.

>but Musk’s companies are likely to be the biggest human decentralizing force in history.

Are you completely unaware that Musk and his family are being sued for self-dealing and fraud? Are you not paying attention to what's happening to cash bleeding companies?

>the solar roofs

The solar roofs that the CFO at the time they were demonstrated publicly didn't think were real, or that the global sales manager for Tesla energy didn't know if they sold a single one during his tenure? Those are changing the world?

SpaceX can't go to Mars. Teslas don't drive themselves. The Solar Roof doesn't exist. Starlink has yet to prove viable. Neuralink and The Boring Company are jokes. These are money-burning "ideas" to get to the next funding round.

It's easy to be a nay-sayer, when the subject of your criticism is taking on long-term difficult problems.

I wonder whether, four years ago, you'd have written "SpaceX can't land or reuse rockets - that's a pipe dream"? Or more recently "Tesla is a basket-case that can't return a profit and will go bust"?

* SpaceX can't go to Mars... yet; but that's really hard, so they're working on it and are closer to it than anyone else

* Teslas don't drive themselves yet; but that's really hard, so they're working on it and are showing constant improvements to their software in pursuit of that end

* Solar Roof v3 probably does exist, and might be a viable product - let's see

* Starlink has yet to be in a position to be proven viable; but that's a question of time

No doubt Musk over-promises and under-delivers (to the extent that it's a meme at this point) but he has already delivered hugely in some areas, and I suspect he'll prove you wrong time and time again over coming years.

Waymo and others appear to be far ahead of Tesla in the self driving realm.

Indeed; once again, Musk is making it hard in the shorter-term for himself/Tesla by going a route based on a totally different vision (i.e. eschewing LIDAR) to the rest of the industry.

Time will tell whether he'll win out in the long term; will this be another landing and reusing rockets moment, or will his approach to self-driving cars be a relative failure?

Seeing as he has to know that the cost of Lidar will continue to fall precipitously, I have to believe the only reason he is actually going all in on cameras only is because he has to be able to continue selling the fantasy that the Tesla you buy today will be fully autonomous with nothing more than a software update.

Tesla and SpaceX are already both massive successes at this point, which wouldn't change even if they both disappeared tomorrow. They've already proven themselves.

Leaving aside the terrible environmental consequences that lowering population density results in, there's two main reasons that keep people living in cities: work and service availability. Tesla is doing impressive work, but they won't have an effect in how people cluster into cities.

That's completely backward? Its centralized human settlement that's cause environmental collapse over history.

That’s a surprising thing to read, do you have any references?

While I would accept that centralised human settlement has been a contributing factor to certain diseases, I had always been told that environmental collapses have generally been caused by either natural disasters or wide-scale rather than centralised overpopulation.

Collapse is usually caused by overpopulation and mismanagement of resources. Decentralize the people and the collapse just comes sooner.

The total human impact is way way bigger when population is sparsely distributed. Cities enable people to live close together, ensuring more land is kept free of negative human influence. Think about it: if the population is the same, the needs are the same. Food, water, waste treatment, transport, luxuries. At the very least, you need the same of everything, but if the density isn't big enough there's no economy of scale that can kick in.

I remember reading about how some US suburbs are effectively bankrupt, because they can't maintain their infrastructures due to not enough people living there. Imagine it, a high-class neighborhood that is forced to replace asphalt roads with gravel roads.

Ironically I live in a house on what seems to be a fairly affluent street [0] in Berlin, yet the road outside the house is unpaved gravel.

[0] seems to be, but I don’t know if the appearance is deceptive as I’m a foreigner who only speaks the local language to the level of simple newspaper stories.

I choose to live in cities because I want to be where the people are. And I'm not much of a "people person."

Nothing beats being minutes away from face to face interaction.

If you cover great distances per minute - say, 600 km/h airplane does 10 kilometers per minute, and 50 kilometers per 5 minutes - then you can live in a sparse (average 50 km between people) environment.

"There are only three trillion dollar industries in existence: energy, high speed transport, and communications"

I really love Casey's writeups, but I can think of quite a few more trillion-dollar industries than just those 3 :)

What are some?

Is there a definitive list of the largest industries in the world, when we think in an abstract sense (i.e. energy, transportation, communications)? Barring how challenging it would be to calculate, of course.

Or maybe a list that aims to encompass all major industries via abstraction? i.e. WhatsApp and Starlink could be layers in the “communications” bucket.

According to some random article[1] I duckduckgoed, agriculture is a $2.4T industry.

[1]: https://croplife.org/news/agriculture-a-2-4-trillion-industr...

As a non-native speaker, I have to ask: shouldn't that be "duckduckwent"? ;)

Try DuckDuckGoosed.

Start at Maslow basic needs.

Food, shelter, security, warmth (energy)

Energy, Transportion and Communication are three basic necessities. Here are some more:

Food ($3-4T), Housing ($16T), Healthcare ($11T), Education (?), Clothing/Apparel ($1T)

Mining is just shy of $1T.

Banking: https://www2.deloitte.com/content/dam/Deloitte/global/Docume...

Sure, but these happen to be the ones Musk is supposedly disrupting, so it fits with the mindset of Musk cultists.

Reply on news.ycombinator.com