[lxgr]

> The "Oops, the Sat-to-Sat links are not working, so we now have to build base stations everywhere and can not do load distribution" might have cost Starlink something like $10 BILLION? I guess I would have tested my stuff first before launching it. With now two generations of Starlink sats already being outdated and/or falling from the sky

You don't seem to understand their strategy: Constant replacement is a feature, not a bug, to them.

And in that paradigm, why wait any longer than absolutely necessary with any given launch? The problem is already fixed – at least inter-satellite links seem to be working well enough now (as evidenced by global coverage on the oceans).

> Starlink will never ever be profitable, just as no Sat ISP in history ever has been.

How do you explain the non-zero stock price of e.g. Iridium and Viasat?

> You have to be EXTREMELY remote for Sat internet to make sense. No, not rural USA. Fiber will be cheaper.

Are you sure laying fiber to every last home is really more capital efficient in the long term? Have you done the math on that side too?

And what about mobile coverage? Even solar-powered low maintenance cell stations need to be installed, repaired after storms, have their solar cells dusted off etc.

> No, not Africa. Fiber through the desert will be cheaper. Sat Internet may make sense if you live in the artic or on mount Everest or something like that.

Mount Everest has pretty good cell signal, as far as I know. It's a tiny area, compared to actually remote but still (sparsely) populated regions.

[Fischgericht]

Due to the nature of the business I am in I very well know Viasats customer base. They are too important to fail for multiple european military organizations.

As discussed elsewhere in this thread, the intra-links still do not seem to be enabled. Can not verify myself due not having a yacht and/or time, but I am constantly flying between Asia and Europe with various airlines, and so far none of them have switched to Starlink but keep paying the outrageous pricing from ViaSat & co.

[lxgr]

> Due to the nature of the business I am in I very well know Viasats customer base. They are too important to fail for multiple european military organizations.

So there is demand :)

> As discussed elsewhere in this thread, the intra-links still do not seem to be enabled. Can not verify myself due not having a yacht and/or time

Are you arguing that everybody reporting successfully using it far away from land is part of some conspiracy? How else would SpaceX get away with claiming that they have global coverage?

> I am constantly flying between Asia and Europe with various airlines, and so far none of them have switched to Starlink but keep paying the outrageous pricing from ViaSat & co.

Installing a new satellite terminal on the outer hull of a commercial aircraft costs millions, including the lost time spent in the hangar, and that's to say nothing about all the required certifications.

That said, Hawaiian Airlines have been using it for a few months now. Seems to be working great, and their routes are also definitely not possible to cover from LEO without inter-satellite links.

[Fischgericht]

No conspiracy, but let's say that it is rather hard to get proper benchmarks done by actual users, and one has to rely on a lot of anecdotical data. Have you seen any real-life benchmark reports with traceroutes, measure downtime, handover time etc that impressed you in a positive way? If so, please share.

Hawaiian Airlines - very interesting. Sadly wrong side of the planet for me to test it myself :)

It very well might be possible that the intra-links are only used for special customers like airlines for now, and not for consumers, and that this is the reason that all people I know who use Starlink still handover downtime...

[lxgr]

"Handover downtimes" for stationary or mobile users? If they're stationary, that's not something inter-satellite links are needed for or would help with.

[Fischgericht]

You are very wrong here:

Right now Starlink claims to be operating a mesh, but they are not. If they would want to build a mesh, Inter-sat links for NOT be used used to pipe through bandwidth to the "best" base station. It would be used for shared state to be able to prepare a handover. Synching state obviously is much easier and more stable if the neighboring sats can talk directly, instead of sharing it over their slow, high latency and lossy base stations.

See IEEE 802.11r for the equivalent for WiFi.

[lxgr]

…what? Where do you see the claim that they are running a mesh? Why would they do that?

The main point of inter-satellite links is to provide coverage to areas beyond single-hop (subscriber to satellite to ground station) coverage. (Theoretically they can also be used to provide extremely low latency intercontinental routing, but for most traffic, the goal would be to minimize routing in space.)

Since the entire constellation is known a priori, all paths can be precomputed centrally, just like in a non-moving network, and that routing information can then be propagated to terminals and satellites. There’s no need to dynamically make complex “mesh” routing decisions at the edge.

802.11r controls faster key exchanges in 802.11 roaming scenarios – what’s the relation to satellite ISPs?

It seems like you have some axe to grind with Starlink and are collecting evidence through that lens.

[Fischgericht]

Can not reply to your anymore, guess we are nested too deep now?

I think we are simply talking about two different things here.

I mentioned 802.11r not due to the key exchange implementation details, but to point to the general point: Seamless handover requires shared state between cells.

This is not about static vs dynamic routing, you are thinking on the wrong layer here. We are in L1+L2 land.

On Starlink, the last time I tested a handover between two Sats in 2025 still involves a downtime of at least 5 seconds, and both L2 info and NAT state being lost.

In regards of axes: I am not much into emotions. Of course the data says that Elon Musk is the cancel cell that will play a huge part in destroying the western civilization. But as I do not like the western civilization and humans in general much, this does not trigger much emotions.

And even if I hated Elon Musk: We are talking about technology, R&D and implementation details here (which I enjoy!). I do not have emotions on IP protocols and such :)

No, in reality it's really very simple: My data says that Starlink just is not worth it. It is not commercially feasible. It pollutes the space with tons of trash that will harm productive future space missions and projects. It's highly overrated and overhyped. It's very hard to find positive reviews that haven't been paid for.

Or, executive summary: Starlink is a dead end, and without the Elon cult nobody after looking at a hypothetical business plan would invest.

And finally: Anecdotical evidence collected from my own tests and those of friends all says: It's just shitty. However: That of course depends on your use case. For some an 8 seconds drop-out might mean "patient dead". For others it might be "I will retry loading this after grabbing a cup of coffee". My peer group might have higher standards than others.

Of course Sat internet has its place as a niche business. But as you surely are aware in the US it was and is tried to steal tax money meant to build fiber by claiming Starlink would be equivalent. And you might also remember that if someone would not have pulled the emergency break, you know would have air traffic controllers seeing planes with 100ms+ of latency AND every now and then losing contact to all airplanes for 8 seconds.

And all of this has been tried before. Over in Europe, we 10 years ago had those fights where Viasat & co claimed to be an alternative when we got the "basic human right to broadband".

[inemesitaffia]

please talk to "panuvic".

pan &a@t& uvic.ca

I believe he's a University professor in Canada. Working on this data wrt OneWeb and Starlink.

[inemesitaffia]

United, Zipair, Hawaiian, Qatar.

The lasers work and I really don't know where you got the idea they don't.

They've worked since at least late 2022.

We're in 2025