Hacker News Reader

Comment by kjksf | Hacker News Reader

Starlink enters the chat.

Most people don't know but Starlink's ambition is to capture only a small part of retail market (a person buying a starlink terminal) but a big chunk of backbone traffic.

That's why latest version of the satellites have lasers to beam data between satellites so that they don't need to go back and forth between satellites and base stations.

That business will be direct competitor to under sea cables.

Fun fact: light travels a bit faster in space than in an optic cable.

Microsoft has already did partnership with Spacex for Azure: https://news.microsoft.com/source/features/digital-transform...

View on HN

Starlink laser capacity is 100Gbps. Undersea cables often carry more than 100Tbps.

> more than 100Tbps

Might be much more than that once twisting light for lossless compression becomes mainstream.

I will probably come at a premium, but I think it's pretty much given that governments and the military will have a huge interest in this, if only for the fact that you can't cut a laser beam in space. At least not without anyone seeing it - unlike undersee cables. '

Governments and the military already have their own space internet; I doubt the US would want to go into an agreement with a commercial entity owned and controlled by someone with dubious allegiances. They will require heaps of legal changes and certifications first, including making sure Musk does not have control over it.

The military has already engaged in such an agreement, known as Starshield.

https://en.wikipedia.org/wiki/Starlink#Military_capabilities

> SpaceX also designs, builds, and launches customized military satellites based on variants of the Starlink satellite bus, with the largest publicly known customer being the Space Development Agency (SDA).

> In 2019, tests by the United States Air Force Research Laboratory (AFRL) demonstrated a 610 Mbit/s data link through Starlink to a Beechcraft C-12 Huron aircraft in flight. Additionally, in late 2019, the United States Air Force successfully tested a connection with Starlink on an AC-130 Gunship. In 2020, the Air Force utilized Starlink in support of its Advanced Battlefield management system during a live-fire exercise. They demonstrated Starlink connected to a "variety of air and terrestrial assets" including the Boeing KC-135 Stratotanker.

The military wants far more bandwidth than is available with their current setup, for everything from drone swarms to far-flung observation bases wanting to check their email.

The military could change the specs and require the sats be equipped with kinetic interceptors. Strategic Defense Initiative mk2. Oh, wait, they already have. It's now called SDA.

https://en.m.wikipedia.org/wiki/Space_Development_Agency

The militraries space internet is very slow and high latency. Think 1 broadband connection shared with 100,000 troops...

They aint watching youtube over that.

Starlink is always 1 missile away from being turned off though. You don’t need to disable the network, you don’t need cause a Kessler syndrome, you just need to convince 1 billionaire that servicing a region is going to be met with force.

>Starlink is always 1 missile away from being turned off though

One missile going into space will trigger every single ICBM early warning system on the planet. And since Starlink is not one system but thousands, trying to disable their backbone functionality would seem like a nuclear first strike. Yes, some countries could probably do it. But you can't do it without being seen, so the risk and possible ramifications are infinitely higher than for undersee cables.

One missile going into space will trigger every single ICBM warning system on the planet.

Most of the world's great powers don't need to fire a missile into space to take out a satellite. They've already lifted that capability into orbit.

I don't believe any of those powers would take out satellites willy nilly. I think it would take open direct military conflict. At the same time though, I didn't believe any of them would 'cross the Rubicon' with respect to undersea pipelines and internet cables either without direct open military conflict.

So, yeah.

Gonna be fun times in the future I guess?

> They've already lifted that capability into orbit.

source ?

> One missile going into space will trigger every single ICBM early warning system on the planet.

Rockets go into space all the time without triggering accidental nuclear war. https://en.wikipedia.org/wiki/Kosmos_1408 IIRC was blown up in an ASAT weapons test without any advance warning; nukes headed to DC/Moscow have a very different flight path than one looking to hit a Starlink node.

Of course they don't trigger nuclear war, that wasn't the point. The point is that they trigger detection systems. We know exactly what happened to Kosmos 1408, including all actors and events, down to every minute. We even have a Wikipedia article listing all those details that you were able to dig up easily. Compare that to the undersee cables that got cut, where we will probably never know exactly what happened. If we cought someone red handed while destroying critical infrastructure, that would be a pretty good casus belli.

> If we cought someone red handed while destroying critical infrastructure, that would be a pretty good casus belli.

We're not going to go to war because someone openly took out a Starlink node to send Musk a message. Not unless we were already going to do it for other reasons.

I find that hard to believe.

It's not that it's a message to Elon; that would be a message of air supremacy to the entire world. There is no way that wouldn't provoke a pretty drastic reaction of some sort from groups that seek to hold the title of 'owners of the sky'.

Case in point: Russians have taken down US drones in recent months. https://www.cnn.com/2023/03/16/politics/us-military-video-dr...

No war. We send a grumpy note to their ambassador and just consider it a minor acceptable consequence of obliterating significant chunks of their armed forces in a proxy war elsewhere.

These US drones were operating in a theatre of war. Things would turn sour very quickly if they shot down civilian critical infrastructure far away from active warzones.

It'll take a lot more than 1 missile to disable Starlink.

I suppose 1 missile aimed at Hawthorne California might do it.

You don't need to disable the system in a technical sense, you just need to persuade management/ownership that it might be a good idea to take the hint.

> Starlink is always 1 missile away from being turned off though

I think one Elon Musk can make more damage to the service than one missile.

It's a distributed, self-healing satellite constellation: it'd take thousands of very expensive [a] missiles to dismantle it. That's one of the features that makes it so attractive for militaries.

See, for example, Chinese military thinking in response to Starlink. They're fully aware they currently have no way to disable it (kinetically). Supposedly, they're putting a priority on figuring ways to do that, in a war.

https://www.scmp.com/news/china/science/article/3178939/chin... (nb. this is Chinese media–based in Hong Kong)

- "With more than 2,300 satellites – and counting – in orbit, Starlink is generally believed to be indestructible because the system can maintain proper functioning after losing some satellites."

- "The unprecedented scale, complexity and flexibility of Starlink would force the Chinese military to develop new anti-satellite capabilities, according to Ren and his colleagues."

[a] Here's what the US used for its anti-LEO-satellite test. Wikipedia says these cost about $10 million each.

https://en.wikipedia.org/wiki/RIM-161_Standard_Missile_3

Such an engineer take. There’s an xkcd about this that involves a wrench.

Let me explain, “the Russians shot one of our satellites out of orbit and say they will continue if we keep providing internet to …, let’s stop” is much more likely than “1 down 2300 to go! Whoop”

I doubt Starlink has the bandwidth needed to beat out an underwater cable, it's good as a backup option though.

I'm curious to know what is more expensive to roll out capacity for.

starlink launches are expensive but so is doing work under the sea.

Starlink uses wireless connections, be them laser or radio. That just cannot scale in the same way that fiber can. It's an apples and oranges comparison.

Quick note that: yes, but free space laser is a bit different than wireless because bandwidth can be extremely high and adjacent transmitter interference is essentially zero.

It still has to get to/from space at some point on each end.

Couldn't you scale horizontally, e.g. with thousands of satellite base stations scattered across a continent? Then each one could occupy a separate, non-overlapping spatial channel.

Perhaps you could split up and distribute your network backbone over something like that?

That'd require scaling up either the number of satellites or their size (talking to thousands of base stations at once = more power, more antennae) dramatically.

It is possible with micromirror devices to have a single lens on a satellite to send thousands of beams in different directions.

That opens up the idea of N satellites having N^2 bandwidth... Which scales much better than optical fibers, where in general to have double the amount of bandwidth, you have double the cable laying effort.

Current starlink satellites do not have this ability - they can connect to max 3 other satellites I believe, which is pretty much the minimum for a fully connected network.

>> in general to have double the amount of bandwidth, you have double the cable laying effort.

Getting double the bandwidth just means laying a slightly thicker cable, and I do mean slightly. Adding a bundle of extra fiber when laying a cable adds maybe a millimeter to the cable width. The cost of the fiber is almost irrelevant to the cost of the other layers and the effort of placing it. Most cables are therefore laid with plenty of extra "dark" fiber for later expansion/redundancy.

Why can't space optics scale as well as undersea fiber optics?

Because there's a non-space part of the route you have to deal with.

Lasers + clouds = unfun.

Aren't the lasers in space? I don't think the satellites beam lasers to/from earth.

Until you put all of Sweden (and the people they want to talk to, like Estonia) in space, that data's going to have to come down at some point. Fast inter-satellite data doesn't help you if it has to queue to get in/out of space.

Because the "beam" in a fiber is tiny and can be controlled, while the beam between two sats is actually very large and runs in free space. Most of the transmitted energy will always overshoot the receiver, resulting in significant crosstalk issues once multiple sats are in sight of each other. You cannot have two beams running the same path on the same frequency, not in space. But new fiber can be laid down infinitely along identical paths without any degradation.

Both are expensive, for sure, but the undersea cables will be used and paid for by major internet providers, whereas finding customers for Starlink is still proving to be tricky.

On the other hand, submarine cables can work for decades; Starlink satellites each last only very few years.

Considering Musk’s history of far-right tendencies, I hope that countries and companies alike are thinking very carefully whether they want critical infrastructure to be backed by this guy.

What do you mean by "far-right" tendencies?

https://twitter.com/elonmusk/status/1601894132573605888

https://apnews.com/article/elon-musk-twitter-inc-technology-...

https://www.rollingstone.com/culture/culture-features/elon-m...

Further examples abound.

Eh, seems like guilt by association. Plus the 3 A.M. Fauci tweet is objectively funny.

https://twitter.com/TheDailyShow/status/1627669410629701636

"We will coup whoever we want! Deal with it" is unfriendly diplomacy from the perspective of small countries not allied to the US.

I am a bit surprised that this has to be elaborated here (as stories like the following are posted here fairly often), but here you go:

- musk openly sharing a German nazi party stance and equating migration with “invasion” [1]

- he allows hate speech to proliferate on twitter [2]

- he regularly shares conspiracy theories that are common in right wing circles, eg. casted doubt on the fact that the shooter of Allen, Texas was a white nationalist (even though he literally had swastikas tattooed) [3]

[1] https://www.theguardian.com/world/2023/sep/30/german-governm...

[2] https://phys.org/news/2023-04-analysis-speech-significantly-...

[3] https://www.bbc.com/news/world-us-canada-65539698

>Considering Musk’s history of far-right tendencies.

I don't remember any far right tendencies, but I sure do remember him cosying up to the Chinese communist party. Yes, it doesn't even take one missile to disable Starlink in a warzone. All it takes is an order from his Chinese overlords threatening they'll take away his factory in China and hell do what they ask.

I think him taking down your starlink access when you are using it is a much more risky part for countries. Already happened to Ukraine as well.

> Already happened to Ukraine as well.

Could you point to more information on this? According to CNN the Starlink network was never active over Crimea, so there was nothing for Elon to “take down” Ukraine was asking for something more than what Starlink could do with its current infrastructure. Starlink was already providing much needed internet access for the rest of Ukraine though.

https://amp.cnn.com/cnn/2023/09/07/politics/elon-musk-biogra...

Musk’s recent involvement in Ukraine’s use of starlink makes me deeply hesitant to rely on SpaceX for communications, especially when Russia is involved as is likely the case here.

A service that is like starlink not run by Musk could be the way forward, of course.

Amazon's trying: https://en.wikipedia.org/wiki/Kuiper_Systems

SpaceX's (not so) secret sauce is the cheap launches, though. That's going to be hard to beat for at least 5-10 years.

That story was widespread, and ultimately false, with the correction never reported as loudly.

SpaceX had always had the cells over Crimea disabled, Ukraine was not aware. Ukraine asked for the service to be enabled there for their drone, this went against SpaceX's policies and the US's policies, so they refused.

At that point, the US had not given Ukraine any guided weaponry that had enough range to hit Crimea, Ukraine was looking to use Starlink as a guidance system, this would've created all sorts of issues and since Ukraine had asked for the cell to be unblocked IIRC a day beforehand, there wasn't enough time to even consider working through the various risks it'd expose SpaceX to.

Even if it didn't happen, it highlights the fact that it might be possible, right?

Sure, no country or company that needs reliable communications should rely on the stability of any single point, similar to how the DoD insists on dissimilar redundancy in national security launch providers, expecting the options to have supply chains with very little overlap such that issues with one don't affect the other.

There have been problems with suddenly-lost connectivity in the past as well, because the Ukrainians advanced far faster onto Russian-occupied territory than SpaceX and the US government expected them to be able to.

I think the larger point is that access is controlled by a single point of failure, and that single point of failure is exceedingly unreliable these days.

We would need an organization, not an individual, in control and that organization would have to be truly dedicated to net neutrality. Not dedicated to net neutrality when it serves them and their friends, or net neutrality as an empty marketing slogan, but actual real net neutrality.

(Plus, your "correction" doesn't even come slide to covering all the issues that have been experienced, and the correction itself needs a few layers of correction, but that's all beside the main point.)

Could be, but I'm also uncomfortable with using civilian tech / utilities like this for warfare; if a government wants sattelite internet, they can launch their own. And I presume they have.

> Fun fact: light travels a bit faster in space than in an optic cable.

Doesn't it have to travel farther, tho? Does that cancel it out?

Starlink is only an extra ~10% of the Earth radius at orbit height, but optic cables slow light by ~30%. So it's actually counterintuitive but "could" be just as fast. (Depending on how far you need to send data, it will regress for intra nation but will theoretically improve long distance communication).

What's the theoretical max bandwidth in LEO's near-vacuum? Could we use a wide 1GHz-1THz channel for directional beams to hit 1Tbps, given a decent SNR?

Cosmic dust and H atoms seem like non-issues for 1000km links? Would power or thermal reqs be an issue here?

I understand it's only about satellite-to-satellite, but still it's interesting.

You're missing the fact that Starlink doesn't take your traffic to the destination. It takes to a nearby downlink station. Then it goes on those same optic cables to the destination.

It can take your data to the destination via inter-satellite laser links. That's how they offer service over water and near/at the poles, where downlink stations are not in line of sight for the satellites.

"can take your data" is not the same as thing as "takes the shortest path". There is no way that Starlink is using precious inter-satellite capacity when your data can go direct to a downlink station and transit on dirt cheap fiber

No one's saying they are, the point was that they're capable of it.

Seems that they operate in various altitudes but taking your approximation of 10% of the earth radius, that is 10% + 10% as the signal needs to travel to and from the satellite to the ground, that increases the total distance travelled

In certain cases low orbit constellations with inter-satellite links (like Starlink) can beat underwater cable for latency:

https://news.satnews.com/2020/10/14/leo-speed-when-milliseco...

Fiberoptic cables aren't perfectly straight either, and light in space is about 30% faster than light in fiber. Considering the link up&down to space adds about 1100km, the break-even point is at around cross-continental distances.

The speed difference is big enough that high-frequency traders have been using microwave relay networks for years now. Turns out that a relay chain of point-to-point radio links (which also operate at light-in-space speeds) has enough of an advantage over regular fiber links to be worth the hassle.

You might be right in pointing out that for latency-sensitive low-bandwidth applications (HFT basically) starlink trunks over long distances should beat undersea optical.

As others have pointed out though, starlink as it currently stands isn’t going to corner the market on general internet backbone traffic.

Backbone traffic will never occur over starlink. The bandwidth is tiny. The latency is high.

Starlink barely works when you are on a moving boat or away from the coast. Believe me, I've tried.

With the appropriate dish?

Possibly. If you find a solution, please forward it to me and I will share it with the nautical community.

Unfortunately, Starlink is not yet available in neither Sweden, Finland or Estonia: https://satellitemap.space/?norad=48119

Too much North.

Why didn't you link to the official map that shows availability in the Nordics? https://www.starlink.com/map

Bit of a shame this is the top comment, it has almost nothing to do with the article?

https://news.ycombinator.com/newsguidelines.html

> Please don't comment about the voting on comments. It never does any good, and it makes boring reading.

What the top comment is is a factor of time anyway, check again later; commenting on the position a comment has doesn't add anything to the discussion either.

Reply on news.ycombinator.com