Scrubbed till tomorrow, May 16th at 7:30pm EST.

Why on earth can't people learn to use UTC? (not directing this at you, just general HN bahaviour) We agree on date standard but time has to be some mystery acronym.

> It's tomorrow at 11:11,5 GLOMQRSFFFFTIP'YËGOH`rockSWAMSWAM time.

Surely you can't expect everyone to learn these, right?

No, 7:30pm Pacific, 10:30pm eastern.

And not EST regardless.

We are in EDT not EST.

It would be hard to imagine they don't have a significant advantage in this massive new space with 4 billion or more potential customers.

Incredible stuff.

Well, maybe not at the same time? I would guess they'll be limited by bandwidth to fewer subscribers than the large ISP's.

Most people live in cities and these satellites can't serve a large number of simultaneous users. Rural and worse that rural are the best applications.

You wouldn't have a Starlink antenna each. You'd have as many Starlink antennas as was reasonable then everyone would use traditional last-mile technologies to connect to them. The available bandwidth sounds reasonably high so it should still result in acceptable speeds unless your population density is bonkers.

Could not they have a hub/receiver on a roof of a building which will then distribute the connection to its peers?

> best applications

That's debatable. Most humanitarian application maybe.

For more details: https://en.wikipedia.org/wiki/Starlink_(satellite_constellat...

In short they are planning to use a low orbit (high drag/low lifetime) to launch Ku- and Ka-band sats.

I couldn't find anything on what the consumer side of this would look like, especially since the low orbits means you can't just point a dish at a sat... at least without some fancy tracking.

The consumer side will be a phased array “pizza box” that automatically locates satellites and aims the signal, and should work on vehicles as well as stationary installations

The most important implication of the low altitude is low latency. As low as (or potentially lower than) in-ground fiber. It will be the first good satellite internet.

To the metro fiber is about 2 ms. This is 15-25 ms (I've seen conflicting numbers).

"to the metro fiber" doesn't really matter. That's just the last-mile latency.

Starlink will be bouncing your packets to earth relays. For all we know, those relays could be directly attached to AWS datacenters. "15-25ms" could be the entire end-to-end latency, not just the "last mile" latency.

Even with your pessimistic estimations that the "15-25ms" is just the last-mile of an equivalently long haul across the internet, we're not talking about some unmanageable increase in latency like geostationary satellite internet has.

Yeah, but then another 15-25ms for sat to server while your fiber is another 60ms to server.

Why is that? I assume you mean a server so far away that the number of hops (switches, etc) is higher on land than over the air? Without a large hop count difference, I don't see why air vs ground fiber would be terribly different. Air takes a direct path but with LEO satellite there's some height, and fiber takes a slightly less direct path but without height and with switches.

Light only moves at about 60% the speed of light in glass fiber. (Specifically, it goes at C / refractive index)

Why is the satellite --> server end lower latency than standard fiber?

The speed of light in a fiber is only about 2/3 of the speed of light in a vacuum. Plus satellite to satellite is line of sight vs zig-zagging around property rights-of-way in the ground.

Scrubbed till tomorrow 7:30pm Pacific, 10:30pm eastern.

What's the consumer bandwidth here?

Scrubbed until tomorrow.