Plus, inferior/high-latency is fine if it's global. I'm guessing people won't generally be trying to play Call of Duty on these universal links, but rather getting essential services like wikipedia, email, etc.

The latency is like 11ms. It's not going to substantially worsen my COD ping.

No, the latency for traditional satellite internet is between 600ms to 1000ms (one full second). Cite: http://arstechnica.com/information-technology/2013/02/satell...

The whole point of the original comment is that these are LEO satellites, rather than the geosynchronous satellites in use for previous internet service. These won't be traditional; they'll be about a thirtieth as far away.

There were mentions of both technologies and I was replying to devindotcom who was talking about how high latency satellites are still useful. Here's a link to that comment: https://news.ycombinator.com/item?id=8920059

That is why I was careful to specify traditional satellite internet services. It appears the distinction was lost.

For geostationary orbit, yeah it's high latency. That's like >22k miles up. Low-earth orbit is a few hundred miles. The difference in latency is huge.

The highest cost is that you will lose sync with the satellite, since they will be rotating over the horizon frequently. If you can launch enough satellites, then you can have multiple in view at any given time and thus keep a continuous link.

The distance between geosynchronous orbit and low earth orbit is huge. One is ~22,236 miles and the other is 99 miles to 2,000 miles.

That article is about satellite internet using geostationary satellites. Those satellites are as much as 200 times farther away than satellites in LEO.

22,236 miles up, opposed to LEO which starts at about 100 miles up.

Yes I am aware. I am discussing the replies above me, not the article.

That's because geostationary orbit is at 22,000 miles. These will be more like Irridium, which is at 485 miles. You can go lower, but it's more expensive because of extra fuel costs.

Down thread there are a lot of numbers thrown about, some slightly wrong or only in miles, so let me summarise them! The values are taken mostly from wikipedia, with some data double checked against [1]

Low Earth Orbit (LEO)

altitude: 160 to 2,000 kilometers (99 to 1,200 miles)

orbital period: ~88 to ~127 minutes

latency: 20 to 40 msecs

Geosynchronous Earth Orbit (GEO)

altitude: 42,164 km (26,199 mi)

orbital period: 1 day (approximately 23 hours 56 minutes and 4 seconds)

latency: 240 milliseconds

[1] http://iml.jou.ufl.edu/projects/Fall99/Coffey/INDEX.HTM

So if the orbital period is about an hour and a half to two hours, does that mean that you can expect to hold on to a single satellite's signal for about 40-50 minutes? Longer? I wonder how much power a cell phone will have to use to track a satellite that's moving through the sky - if the tracking will dominate power usage, or sending and receiving the signal will and we won't see much change.

I wonder how this will be different (if at all) from GEO satellites when it's cloudy. Could you presumably find a satellite near the horizon to use to bypass the clouds?

The magic words are 'phased array antennae'.

Much less - these satellites are low, so you're not going to have line-of-sight for anything near half the orbital period. Not sure about the precise numbers, but I know Iridium was really challenged by the requirement for clients to roam between satellites on the order of once a minute. As they demonstrated, though, it's a solvable problem.

Satcom data rates and reliability are definitely going to be inferior to fixed lines and I assume terrestrial wireless.

But crappy sat network is better than no network. If I can build a log cabin 1000 miles from anywhere and access HN, or log into HN from my sailboat offshore, I will finally be able to never be productive from anywhere.

Yeah it should open out a whole new segment to consumer markets.

Free(but very slow) ad supported internet every where on earth.

Connection to the cloud for a million different kind of Internet-Of-Things class devices. I think this is pretty big as these devices consume very little data, but generally have issues getting connectively at all in some the more remote areas these things tend to get deployed. Think ocean current mappers, artic ice melting detectors, or a multitude of geo-sensors that would love to have a connection in extremely remote areas.

Satellite-based emergency beacon in every phone?

Slow but usable internet for remote villages. Pushing news, weather, education materials. Perhaps combined with solar power One Laptop Per Child class devices.

He's already stated that it won't be free and that the consumer base stations (using phased array antennas) will run from $100 to $300.

Free is subjective. They won't be free, but Google could subsidize stations and add ads.

IoT companies can create through own hardware and pay bulk rates.

Cellphone companies certainly have the option to get on the boat too, if the hardware necessary to communicate to satellites is small enough to go in a smart phone, as have the option to make an emergency call(or most likely SMS/Email) nearly anywhere in the world is a pretty killer feature.

Yes, obviously. The same is true of high latency networks, but that wasn't the point of the grandparent.

Depends what market you're targeting. Google's "Project Loon" (data delivery from antennas stationed on high-altitude blimps) was explicitly targeted at third-world areas without much on-the-ground infrastructure.