Hacker News Reader

Comment by adtac | Hacker News Reader

And consumer internet isn't even Starlink's endgame

I'm aware that they expect one of their customers to be stock market companies where lower latency is highly important. Are you referring to that or something else?

The "stock market companies" is a fantasy promoted by random people, not SpaceX.

What SpaceX shared is that they plan to be "fiber cable" except via space. They'll carry a backbone traffic e.g. between Australia and U.S. via laser links.

Companies that currently pay for undersea fiber cable to carry traffic will pay Starlink for the same service.

Also, a backhaul for 5G operators (see https://www.telecompetitor.com/starlink-update-musk-sees-mob..., https://www.zdnet.com/article/japanese-telco-kddi-to-use-spa...).

Also airlines.

High frequency traders would look into this, but I don't think it would be a big revenue stream for SpaceX.

I used to work for a company doing microwave backhaul and they liked us as the speed of light through air is faster than fiber, and they could get more direct links. It was important enough that we had special low latency builds of our firmware with some features removed.

Eventually, with satellite to satellite laser links, and full global coverage with this efficient number of highly inclined orbit satellites, they intend to capture the entire market that is right now addressed by various very expensive motorized tracking geostationary dishes mounted on offshore vessels, plus L/S band inmarsat, iridium, and others.

Offshore oil and gas, cargo ships, Cruise ships, coastal vessels, medium-distance ferries such as in Greece, business jets, airliners, portable military applications, data links to offshore scientific applications for buoys and weather, all sorts of stuff.

I understand that’s a lot of use cases, but is it a lot of REVENUE?

Verizon and AT&T built multi billion of dollars a year of sales from cookie cutter DSL (simplistically). All these crazy edge cases…are they material? Does it make sense to shoot thousands of satellites in space for ferry boats in Greece? Financially that is.

That being said it’s really cool what a “catch all” solution we’re getting as a humanity for basically free. This feels like Google Maps / Gmail / WhatsApp where all humans get like value for free (especially in the developing world) due to one company creating a massive positive externality due to their business strategy. So cool.

Go look at the combined revenue in billions per year from all the geostationary satellite owning companies like Inmarsat, thuraya, Intelsat, ses, eutelsat, arabsat, amos, etc. Most are publicly traded ompanies. LEO properly implemented will beat the pants off it in performance and speed.

HFT already uses HF radio bands between the UK and USA which is lower latency than either submarine fiber or starlink will be. Although much more limited in kbps.

Big ass yagi uda antennas in certain locations with totally custom rf chains and modems.

Fascinating! I've heard about microwave relay towers used for HFT through an amazing series of blog posts [1], but HF is news to me.

Do you have any sources/further reading? I'd be curious what type of licensing that operates under etc.

Edit: Turns out there's an article on it on the very same blog [2]!

[1] https://sniperinmahwah.wordpress.com/2014/09/22/hft-in-my-ba...

[2] https://sniperinmahwah.wordpress.com/2018/05/07/shortwave-tr...

Experimental licenses so far or straight up illegal, they're making much more money than any FCC fines would be. The FCC enforcement bureau is actually pretty small and doesn't come after you until you really screw up someone else's pre existing licensed service.

All FCC fines are public info (called a notice of apparent liability and published in their daily data dumps), I've yet to hear of one getting fined.

I'd venture a guess they could mean using Starlink as the backbone for off-planet comms? I've heard that referenced a few times. The true end game is probably all of the above and more. If the need fits the bill and the pockets are deep enough, it'll probably be using Starlink.

I’m very ignorant about HFT, but https://www.starlink.com/ says “latency as low as 20ms in most locations”. Isn’t that many many orders of magnitude too much for HFT?

High-Frequency Trading / Arbitrage between distant markets was what I'd read.

i.e starlink can get much lower latency between say London and NY than terrestrial links.

As I understand it, HFT justified digging new tunnels between NYC and Chicago to take advantage of the shorter distance the signal would have to go. And that was transmitting at or near the speed of light. Somehow, I doubt bouncing on a roundtrip to a satellite is worth it.

> And that was transmitting at or near the speed of light. Somehow, I doubt bouncing on a roundtrip to a satellite is worth it.

That's the thing, the speed of light in fiber is actually significantly slower than the speed of light through air, which itself is slightly slower than the speed of light through a vacuum.

Over long enough distances this adds up to the point that high frequency traders already have built radio relay networks between New York and Chicago. You can't easily build radio towers across oceans, but a satellite relay network that could follow a nearly line-of-sight path is almost the same thing.

Also fiber isn't run as the crow flies, sometimes it has to take a winding path which adds even more distance to the run than you might think, on top of the reduced speed of light.

If Starlink manages to get their satellite to satellite laser link tech working, which is a MASSIVE if, they're in a very good place.

Mark Handley[1] has done a number of videos running the numbers based on published Starlink numbers, both with and without the inter-satellite laser links.

It's been a while since I watched them, but I recall that even without laser links, they can beat any of the terrestrial links over a certain distance - because the terrestrial links can't get direct-LOS.

[1] https://www.youtube.com/channel/UC-k1j7M2-hBfXeECd9YAQ_g

Based on what math? Tokyo and New-York are 10,848.68 km apart, terrestrially (though cables don't go as the crow flies), but LEO is only 2,000km off the surface. If Starlink can get connect Tokyo to New York in fewer than 3 hops, and thus over shorter distance, then you can bet that HFT firms will be falling over themselves to sign up.

It is even better than that as Starlink shells start at 550km orbit .

( It doesn’t matter as the boost from not using glass as medium for light makes it no contest anyway)

The speed of light in fibre (I.e. glass) is substantially slower speed of light than near vacuum . So yes bouncing up a few hundred km is faster for medium and long range connections .

It will not be faster than fibre for Manhattan to Princeton , but it would be faster for London - New York or London-Frankfurt

He's referring to highly speculative theories like https://discovery.ucl.ac.uk/id/eprint/10062262/

I think it’s more like cross-world latency. Where going from one side of the world to the other will be slower going via fibre optic vs through space as with starlink? I think this is reliant on the satellite-satellite latency being low and the ground to satellite being not insanely high.

Reply on news.ycombinator.com