Hacker News Reader

Comment by walrus01 | Hacker News Reader

Regarding spacex in this, my starlink terminal recently hit a milestone of 0.00% ICMP loss to terrestrial ISP things in the pacific northwest. Averaged over 5, 6 hour periods. 20 icmp pings every 60s, basically a default smokeping install set to once a minute.

Previously it was running somewhere around 0.15% to 0.60% which was still totally usable... if you put a similar test setup on a noisy/oversubscribed coaxial cablemodem DOCSIS3 segment in a randomly chosen house in a suburb somewhere, you might also see an average of 0.50% loss in many cities.

For anyone who has seen packet loss, jitter and transfer quotas on deeply oversubscribed consumer geostationary services, this is amazing.

The next big thing will be to see if they can really get satellite-to-satellite laser links working so that starlink service area and coverage is not limited by having multiple moving LEO satellites in line of sight of both the CPE and a regional spacex earth station at the same time. for instance if people wanted to put a working starlink terminal somewhere really difficult to reach like socotra, yemen.

View on HN

I'll go as far as to say that Starlink is now so good i honestly believe it's going to change the other industries and not just its own.

As an example I've been looking at properties outside of suburbia since testing a friends Starlink terminal. I know i can get proper broadband on par with cable anywhere. It's changed my personal long term future plans. I think it's going to change a lot of things.

Isn’t starlink internet connection expensive atm? $99 per month plus one time h/w costs?

Not just expensive. IIRC, the receiving dish and associated gear burns 120W or something like that.

Living with more space though, you can offset that with solar pretty easily.

That's still only around $100 a year in electricity assuming 10 cents/kWh.

In Germany, regular prices scratch 30 cents/kWh, so you're looking at another 25€ a month just to run the dish.

A regular unlimited Gigabit Uplink runs 40€-80€ a month, with hardware included and a power usage in the range of 10W.

But, unfortunately, anywhere you have a Garden large enough to host starlink, you probably won't have anything near that speed available, so it's still pretty competitive.

It's not a replacement for suburban internet. It's a way to pay just 50% more (in Australian pricing) for something that offers way more options.

That's not terrible. I pay something like $140/mo for 1gbps (in reality I've never seen speeds >300mbps) with no data cap. $100 for ~100mbps with no data cap is reasonable to me.

You know you can like get them to fix that right?

How? In the fine print for most of these contracts it says something like “up to 1Gb, actual rates may vary based on network load” or something similar.

how?

Call them up, say you are not getting the speeds, then when the tech shows up, simply don't let him leave until you can verify a speed test at that rate.

If you get a good tech they will help you, you may get a dud, in which case you simply have to call them up and work the system again.

It's totally doable, I've done it dozens of times at every dwelling I've inhabited. I'm currently on their 1GB plan, and I do get 1.4Gbps down (they recently upped their provisioning rate, and with a newer modem with 1 2GB link on it, with a suitable router [I run the UDM Pro] it works).

about the same as comcast in the US...

In Italy an average Gbps fiber connection costs about 25€ (about 28$) per month which is insanely cheap compared to other countries.

This also means that, at least for us, 99$ (88€) per month for such a thing is a lot more at the end of the year (+756€). I hope they’ll eventually adjust that price for different markets because at the moment it’s not even competitive in my area.

it's not meant to be competitive in areas that have existing reliable and competitively priced terrestrial wire line infrastructure (fiber, docsis3 or docsis3.1 on coax cable, short reach VDSL2 or g.fast on old copper phone lines, short reach WISPs capable of 150Mbps+). it's for areas that have none of the previously mentioned last mile access technologies available to each house.

That’s not expensive.

I get 1gbps for ~$100/mo in the bay area, but if I lived a few miles further from the city I’d get 10mbps DSL over wet string regardless of how much I’m willing to pay.

FYI Sonic is now offering 10GB/s for $40/month in parts of the Bay Area.

10GB/s or 10gb/s?

I stand corrected, it's Gb/s

In Australia, it's $140 / month, plus the capex of $700 for the dish, which obviously won't work on any other provider. I suspect resale on that device, if the carrier allows it, would be reasonable, but I don't expect I'd be moving away from this service.

People are comparing (sub)urban US multi-carrier options, or heavily populated countries (like Italy), that have much cheaper pricing. It's not surprising that terrestrial services there are better spec'd and cheaper.

For me (in AU) it's compelling because I otherwise have an option of a 3G yagi / booster (nearest tower is 15km away, service is awful normally, but unusable in storms, frequently highly contended, etc), dial-up (there's very few places that still offer such a service, and just the landline connection fee is $40 / month to start with), or geo-stationary satellite (what I've just moved away from after 2+ years, which has > 600ms RTT to the nearest capital city, and a max downstream of 25Mbps, which is also frequently highly contended during peak times.

The 100W / hour usage is frustrating, and claims of electricity being 'just 10c / kWh' are quaint, I guess coming from well serviced parts of EU / US? Here most people are up for 30c (AUD) / kWh, and if on ToU that can go up to 60c during peak (2pm through 8pm). I've got solar, so on average I'm powering this 'for free' 25% of the time.

rough calculation in US dollars, if a starlink terminal is 100W constant load and electricity costs $0.30 per kWh, it would be about a $22.32 extra on the monthly electrical bill.

in your case about $16.74 if you subtract 25% for the estimate of benefit from your solar (grid tied?) kWh going back to the grid

That sounds about right.

For the average Australian user, it means you add ~ $200-350 / year to your effective ISP cost. Not end of world for most, but certainly not trivial for all, either.

The regrettable history of Australian internet provisioning (specifically our national broadband network) and electricity pricing (specifically the rorts and corruption, and higher-than-rest-of-world pricing) are described much more eloquently by others.

>but unusable in storms

Dumb Q - is satellite any better in a storm?

I would guess there is less interference between a ground station and a satellite than between two ground stations. Two main reasons:

1) The storm would have to be right on top of the ground station to be in the line of sight when talking to a satellite. For the 3g case, the storm could be anywhere along the 15km between the two ground stations.

2) Storms typically cover a larger area horizontally than they do vertically, so for most angles, communicating with a satellite should go through less of the cloud than horizontal 3g communication. I'm sure there could be storms where this is not true (eg a tall thunderstorm directly above the ground station), but my guess is that such storms are more rare.

Storms, and just heavy rain, consistently send our satellite internet offline. This is Australian “sky muster plus” pointing at a geosynchronous satellite at a higher orbit than starlink.

To augment the other two responses here with some anecdata -- yes, storms would knock off my geostationary satellite connection in ways I didn't fully understand (it might just have been a huge bump in attempted utilisation for all I could tell at the consumer level).

Starlink during storms, and we've have a half dozen big ones since I got this service 4 months ago, seems to do more re-acquires during those events (several seconds of drop-outs) but I'm not seeing significantly lengthy outages.

The dish is still sitting on some concrete blocks on the ground, so it may simply be that it's wobbling more during the wilder winds.

I pay like $94/month for DSL in a rural area. That includes a (mandatory) land line, and I'm "guaranteed" 10/1 Mbps--which I don't get.

The only other options, until Starlink launches in my area, are traditional satellites. My neighbor previously had one of those, and he had an 8GB cap per MONTH. Plus the ping time averaged like 600ms.

I'm not even mad about the price of $94. Internet is way, way worth it. The quality of the connection sucks, though. I would pay $200 for Starlink once it's reliable. I'm not actually sure what my limit is, but it's pretty high, because rural internet is painful

the economics of consumer grade oversubscribed geostationary are terrible - it costs something like $200 million to build and launch a 5500 kg satellite into geo orbit, it'll almost certainly die in 15 years (Standard lifespan), and its aggregate capacity is TINY

actual dedicated 1:1 geostationary network capacity costs at a minimum $1600 per Mbps per month each direction (eg: 2 Mbps down x 0.5 Mbps up circuit would easily be $3500/mo)

Question: How is starlink going to be a game changer in ways publicly funded infrastructure (either traditional cables or cables + 5G towers) aren’t?

Your anecdote doesn’t really cover “game changer” since land outside of suburbia is pretty either scarce, big and expensive, underdeveloped, etc. and not many people want to live there (and I doubt fast broadband is the only thing holding people back).

I honestly don’t think starlink is useful for that many people. Personally I would categorize it as a quality of life improvement for a small group of wealthy individuals (and maybe even smaller number of industrial workers which work onsite in remote locations). I have a hard time seeing it doing anything more.

there are a number of rural areas in the USA that politically show very low likelihood of getting something like a proper last mile broadband service built by wireline infrastructure any time soon (fiber and GPON or active ethernet). some rural counties in eastern WA are the exception.

Under-served rural areas in the USA tend to be pretty poor, and as is starlink (or starlink-like) internet is pretty expensive both for the service fee and hardware requirements. Perhaps the price will come down as operators proliferate and this will be affordable for these communities. But I still fail to see how this is a game changer.

Serving these communities with traditional infrastructure and maybe a 5G tower is very possible today, and frankly not that expensive for a country like USA. I think you undercount the political feasibility a bit. Politicians in the USA make a huge buzz about serving rural communities, infrastructure is a political buzzword at the moment. A good politician could easily capitalize on rural broadband. They don’t even have to be that well intentioned since they could take some lobby money from the telecom companies and bid out the infrastructure build out with conveniently specific requirements for contractors their family just so happens to own stock in.

that's true in terms of average income, but $110/mo USD for starlink is about the same as a 20GB quota per month consumer geostationary service that is absolutely a terrible and miserable experience (viasat/wildblue/hughesnet etc). Or at most, $10-20 more per month.

You seem to have very strict definition of rural that doesn't map to how the FCC or Census Bureau define the term. I live in an area that is very clearly 'rural', most of it has been built on in the past 30-40 years and the average home price within a few miles of me is ~$400k. I live by engineers, doctors, architects, technologists and none of us have had anything but 10M DSL and Hughes satellite options for Internet in the past 15 years.

Yes everything you've said about the possibilities is technically true, but it hasn't happened and there is no evidence it will happen. The RDOF has awarded another provider funding to bring fiber to within a few miles of my place but not directly to me.

Meanwhile I've been on Starlink since last February. I just checked and I'm getting 110M/12M with a 28ms ping time, which is 10x what I was getting for effectively the same price. Not only that, if I want to go on the road I can pack that little dish and take that bandwidth with me just about anywhere in the US. If I was broke I could pool funds with a few neighbors and we could split the cost.

With that available now, why would anyone spend millions of dollars to drop in a 5G tower to pick up another 50-100 customers?

Now my notion of under-served rural America might be biased. I have traveled quite a bit around rural America and by far the most people in under-served communities were on the poorer side. Now I did definitely see affluent communities with minimal infrastructure (particularly in California), but that was very much the exception (even in California).

> With that available now, why would anyone spend millions of dollars to drop in a 5G tower to pick up another 50-100 customers?

Your county would—and should—spend a few million dollars with some extra funding from your state and local telecom companies to drop a few 5G towers to pick up a few thousand more customers and voter happiness for sure. Your county counsel can call it something like Rural Broadband initiative and have a plan to bring a broadband connection to 30-70% of under-served rural households in the county before 2030. A good politician should be able to sell it even to city folks as serving the community.

> Under-served rural areas in the USA tend to be pretty poor, and as is starlink (or starlink-like) internet is pretty expensive both for the service fee and hardware requirements.

It's a utility. If it costs twice as much as elsewhere you still need to pay it, until it starts getting really difficult.

Our experience differs from your view. We recently moved from the Bay Area to the Sierra Foothills, near Sonora. There is plenty of land with homes already in place, as well as land suitable for new construction. Land is available in sizes from 10,000 sq. ft. lots to 1,200 acres and everywhere in between. Electricity, propane, and water are commonly available. As a rule of thumb, prices are ~0.25xthe Bay Area.

Although Comcast is a thing in town, most of the otherwise eminently suitable locations have no cable access. The choices are old-style satellite or Starlink.

We have met an increasing number of people who have left the Bay Area and moved to our area. Nearly all of these are remote tech workers. They have generally held their noses and signed up for geosynchronous satellite while waiting for their Starlink service. Those who have transitioned to Starlink are very happy they made the move when they did.

I think Starlink will prove useful to so many people that it will increase pricing for rural real estate.

Starlink is incredibly useful for areas that didn't have traditional Internet. It's not perfect but it makes working in rural areas possible where it wasn't before.

It's not even rural. It's less than 3 hours from 3 major cities (and within an hour of a minor one), in a popular weekend vacation spot, and yeah, it's pretty mountainous but it's also got plenty of infrastructure already.

It is amazing how our traditional telcos have failed us. Buildout of broadband was fully-funded by the government in the 90s, and it flopped; They took the money and never built anything. These communities still do not have access to reliable broadband as it was defined then.

I so agree with you.

Comcast could have built out service in so many of our current residential areas, but chose not to. For no apparent reason but that it cost capex.

There was so much funding to extend internet access to rural areas. None of the major vendors engaged these funds and used them to provide capacity.

If Starlink succeeds, I'll raise a glass to the memory of Comcast and others.

> ...increasing number of people who have left the Bay Area and moved to our area. Nearly all of these are remote tech workers.

spot on:

>> a quality of life improvement for a small group of wealthy individuals

Only question is, will the money from this small group of wealthy individuals be sufficient to pay for the full system? We'll see in about 4 years the latest, when satellites will start falling down, and will have to be replaced by new ones...

Money from a small group of wealthy individuals might be sufficient to pay for this system and perhaps few competitors. However it is quite a long shot at calling that a game changer. Quite the contrary this might be a game stopper for a larger number of poorer folks in rural communities that have been asking for traditional broadband for decades and now have to battle the “Technology will solve it” issue their local politicians will inevitably bring up now.

Hey neighbor, we moved into town so Comcast is available. Other options will be great, though!

> How is starlink going to be a game changer in ways publicly funded infrastructure (either traditional cables or cables + 5G towers) aren’t?

Starlink exists.

People spend money on things that aren’t “game changers” all the time. A business can certainly exist while only offering a quality of life improvements to a small minority.

What are the chances of Starlink getting worse as adoption grows and available bandwidth gets saturated?

You're only competing with your neighbors, since different regions connect to different ground stations. If you live in a rural location it's unlikely the user density will ever be high enough to cause a problem.

If you live in a dense metro area it will probably fill up very quickly - but then again Starlink is not really designed for those locations, which probably already have good connectivity.

100%. they're nowhere near the satellite/user ratio they're planning.

This shift is already happening, look at property values over the last 1-2 years.

My guess is that enough rich people moving into the middle of nowhere will cause the government to pay for terrestrial solutions ("incentivize") within the next 10 or so years.

If enough rich people move to rural counties and get involved in politics, they can create public utility districts to build open access dark fiber which 3rd party ISPs can ride on.

Wouldn't you want to wait for at least two providers before rearchitecting your life around a product like that?

There's not enough satellites to be useful yet, terrestrial astronomy is still possible. They'd need to have ten times the current number to stop it, and that's not due to happen for a few years.

I reckon if they put their mind to it they could cause enough pollution to make cellphones unusable, maybe even open-air wifi. They're smart, I'm sure they have the technology, they could make the whole EM spectrum so dirty the whole sky would glow.

My brother just rebuilt a house up in Maine. Previously the house had 1Mbps (with a tailwind) DSL. They wouldn't even reinstall it and houses further down the road couldn't get it at all.

You could get a Verizon hotspot (and ration data). But Starlink is pretty good. Video streaming does stutter a bit--you're better off downloading. I could actually work up there which I couldn't really with the old DSL.

That's awesome! On Starlink in Mendocino county (CA) (in the woods), I'm still dropping out for a few seconds every couple of minutes. Fine for calls, Netflix, and browsing, but most video games will boot me whenever they happen. Some streaming services seem to buffer far less than others. Disney+ seems to stop playing the moment it can't phone home.

Admittedly, most of my trouble is probably caused by the 70 meter coastal redwood on the western horizon. The network isn't quite dense enough yet that I can get a satellite visible at all times amongst the trees. It's just a bit sad, as the Starlink app reports like 95% clear sky. You really do need 100%.

I agree on the trees. Mine is also at a considerably further northern latitude so the simultaneous density of overhead satellites at any point in time is quite a lot more.

for people with trees they cannot cut down I see a lot more 40' aluminum light duty towers with foundation kits getting sold in the future. Much like people used to put up in very rural areas with yagi antennas on them to receive OTA TV a logn time ago.

http://www.texastowers.com/towers/ss.htm

Hello from Philo!

We were seeing drops of a few minutes each day, but as of December we've gone days without even the slightest interruption. We had to really work to get Dishy to not have any obstructions, but since then it's been great:

https://pics.rickosborne.org/blog/wp-content/uploads/2022/01...

We also have Ukiah Wireless, point to point microwave, as a backup. I use it for my work uplink as it has more stable upload bandwidth than Starlink. And when we had those really bad storms over the last few weeks, Starlink definitely struggled more than microwave.

Do you know if this based on some technical aspect, or simply that has Space X not yet to saturated their network capacity?

there was a 1-hour full outage at 0200 immediately preceding it. without inside info into what they're doing, I believe it's a combination of new cpe terminal firmware, satellite firmware/software loads, and more of the recently launched satellites being put into use at their intended orbits, so there's greater simultaneous satellite coverage over my area now.

What's your next hop time? Mine is about 20ms at home (FTTC - Fibre to the Cabinet) Satellite links are quite phenomenal to be honest (when they work).

The metrics to worry about are latency and loss and throughput. Jitter is really a SIPnRTP thing. Don't use ICMP to test for "quality" - it isn't even the same protocol that you worry about (TCP/UDP), it is what it is - ICMP (use it wisely).

the absolute minimum rtt ping i've seen to downtown seattle is 15.8ms but it's more typically like 25ms. That's keeping in mind the rtt for a ping is four one-way trips through space plus latency from the earth station in a far eastern suburb of seattle to downtown and back, by fiber.

jitter is really quite consistent at 23-28ms

I agree on the icmp, for me, icmp loss is really more like a cheap/easy early warning sign that something is wrong. if a properly implemented last mile connection has no icmp loss it's also unlikely anything else is wrong. if there is a few % icmp loss, very likely normal tcp and udp things will also behave badly.

Last time I checked, there were dozens of commercial satellite ISPs ramping up their launches. Do these companies have a plan to deal with potential collisions and the potential space debris that would cause?

personal opinion, space is huge, and the gap between companies issuing press release and the number of companies actually implementing hardware is huge.

oneweb is just now becoming barely usable above 55 degrees north latitude. they have a lot of launches to go before it's anything like a real network (and will not be for individual end user CPEs, a oneweb terminal is much bigger and more costly). other things like telesat LEO network remain in the land of vaporware, same as with amazon's.

"Space is big. Really big. You just won't believe how vastly hugely mind-bogglingly big it is. I mean, you may think it's a long way down the road to the chemist, but that's just peanuts to space."

(RIP Douglas Adams)

Space is hugely, vastly, mind-bogglingly big.

Useful low orbital shells around the Earth are an awful lot smaller. They're still big, like the sky is big, but airplanes do manage to, at least somewhat regularly, collide with each other "in the middle of nowhere" (most midairs are around airports, since there are a lot of airplanes heading in and out of them).

Aircraft debris, however, falls out of the sky in short order. Orbital debris doesn't.

The probability of an initial collision is fairly low, but once you have a few, or someone decides to show off their new ASAT weaponry (which is functionally a satellite collision, just one of them only having been in space for a short while), you now have flak that won't go away, moving in all directions at orbital velocity. That, then, has the potential to start causing all sorts of problems, and once that's taken out a few more satellites with orbital-energy impacts, you start having a real problem.

Will it happen? Those invested in littering certain useful orbital shells with satellites claim not. We'll see.

Kessler syndrome is definitely a problem.

I've heard SpaceX claims to deal with it by putting their satellites in an orbit that requires periodic boosting to stay put -- if they don't, satellites decay and de-orbit after a few years.

The plan is, I think, to regularly retire (de-orbit) and replace these satellites so they can continuously improve the design, and as a nice side-effect any collisions or otherwise misbehaving objects will at least stop being a problem after a known maximum span of time.

Another benefit is latency

And the birthday problem applies; the probability of a collision goes up quadratically. If you have N objects, you've got N^2/2 pairs that could collide.

Low earth orbit (where Starlink and the ISS are) actually has a fair amount of atmospheric drag. Random google searching suggests an inert Starlink satellite (or fragments of one) would come down in 5-10 years.

That's the claim made, yes.

Now, how many other satellites are likely to get hit in "5-10 years" of flak flying around at orbital velocities? It's nice if your goal is to say "Well, it probably won't render space useless forever..." - but that assumes one collision, and no ongoing collisions. If it's "5-10 years per collision," and each new one resets the 5-10 years timeframe, well, there's a lot of stuff up there that can make a mess.

Unfortunately, the low orbital shells are also "everywhere." It's not like geosync orbits where if you Kessler them, well... not going to have geosync stuff anymore, but you can still use other orbits. If you Kessler LEO shells, you pretty much have a high risk of anything you try to send to space getting blasted on the way up. It doesn't take much of a bit of debris, at orbital speeds, to trash your booster and make you have a bad day. I guarantee a bolt at 17,000 mph will do some damage (typical LEO velocities) - that's about ~22x the speed of sound in the atmosphere, depending on altitude.

If there are tens of thousands of things up there and you start things colliding and blowing up, you've basically put a defensive shell around the earth for a century, if not longer, by the time all the satellites get fragged and the debris eventually comes down.

> If it's "5-10 years per collision," and each new one resets the 5-10 years timeframe

It's not.

The concern is that unwanted collisions will become common enough that satellites break down or are completely destroyed within months. There will never be enough debris that we can't fly through LEO.

Therein lies the birthday paradox: the frequency of collisions among n bodies grows quadratically; but the probability of a one-time passerby being involved in those collisions only grows linearly.

> There will never be enough debris that we can't fly through LEO.

Flying through and flying through safely are very different things. With tens of thousands of objects in and below LEO and (presumably) increasing flights to orbit, it's only a matter of time until a rocket gets taken down. It doesn't require a particularly large piece for an involuntary update to your deorbiting plans.

Yes, space is big. Lower earth orbit is still very big. However if you have dozens of operators, each with thousands of satiates making multiple round the world trips. The chances of a specific satellite colliding perhaps is tiny (given how big space is). However the chances of one of the thousands colliding with any of the other thousands over any of their orbital periods increases exponentially as the number of satellites increases.

Tens of thousands.

> There are currently over 1,600 Starlink satellites in orbit, and that number will continue to grow; SpaceX has filed paperwork for up to 42,000 satellites for the constellation.

https://www.space.com/spacex-starlink-internet-satellites-la...

They're aware of the risk of collisions, but toss in a few more vendors trying for the same size, and you can have potentially hundreds of thousands of satellites in the shells. They're big, but they're not that big.

What worries me personally is that they are aware and simply don’t care. Our current climate catastrophe is a testament that big business can be completely aware of an impending disaster, but still ignore it while profits are showing.

Since space debris in the satellite zone is already a problem, that's not a solution.

There isn't just one satellite zone. Starlink satellites are about 4 times closer to earth than than they are to the geosynchronous Hughesnet satellites. The starlink satellites are close enough that even after full Gravity-style Kessler syndrome, you could just wait several years and start over from scratch.

Obviously that is something to avoid, but it's not like we're filling up a non-renewable resource in low earth orbit. It cleans itself naturally.

Note: Starlink orbits at an altitude of 340-350 miles . Hughesnet’s geosynchronous orbit has an altitude of 22,000 miles. That is over 62 times higher, not 4.

FWIW.

I apologize, you are quite correct! Not quite sure what my brain was doing there, I was looking at those same numbers and somehow typed 4.

Sorry if this a stupid question, but in the event of mass collisions a la Kessler, wouldn't some percentage of the debris get stuck in higher orbits? The kinetic force of collisions at those speeds seems pretty significant.

The phrase you are looking for is "Kessler cascade"

So is the earth, really. Reminds me of how all humans would fit in a ball of meat in Central Park [0]. The satellites aren't much bigger than humans, so even if there was a starlink satellite for every single human, it would still barely make a dent. Let alone the fact that the surface area would be larger 2000KM up, and without oceans.

[0] https://www.reddit.com/r/theydidthemath/comments/p0qws3/self...

Nit, but 550km up. You can visualize the Earth as a basketball with the Starlink satellites crawling around an inch above the surface. (Now, with my local folk out of the way, the rest of you can visualize a football with the satellites 2cm over the surface.)

This reminds me of my favorite "true but hard to believe" fact.

Imagine you have a string wrapped tightly around the earth, how much more string to you need to raise the string 10cm above the ground everywhere? The answer just feels so hard to believe, but it indeed only depends on pi and 10cm.

For someone like me, who's ability to do math stops when they run out of fingers and toes, what's the answer?

2pi multiplied by how much you want to raise the string, I believe: https://en.wikipedia.org/wiki/String_girdling_Earth

Do you have a list of these? I'm aware of Amazon's Kuiper and also ViaSat / OneWeb. Neither are available yet.

Yes, they have a plan to deal with collisions.

There's not dozens for broadband internet. A lot more if you count companies intending to build low throughput non realtime m2m data networks with cubesat size things.

Announced but not implemented Leo other than SpaceX include amazon, Russia, china, telesat.

Note that this would also be the observable effect of simply prioritizing ICMP while dropping other protocols.

Probably a better thing to observe is loss and congestion on normal TCP sockets, using retransmit counters, congestion window collapse, etc.

Was going to mention that a lot of ISPs de-prioritize ICMP but I get a sense that OP might be aware of that.

It'd be cool to see what a long running mtr in TCP mode says about Starlink.

I'm aware, this is through the isp to endpoint destinations, not pinging intermediate hops in the starlink cgnat path to internet.

> For anyone who has seen packet loss, jitter and transfer quotas on deeply oversubscribed consumer geostationary services, this is amazing.

starlink is not even close to oversubscribed, so I'm not sure why you're making this comparison.

Reply on news.ycombinator.com