I've had Starlink deployed for a client since mid-January or so up around the 45th parallel (near the US/Canadian border). Installed it in a brisk -20 windchill

I've had Starlink deployed for a client since mid-January or so up around the 45th parallel (near the US/Canadian border). Installed it in a brisk -20 windchill too, good times. That was kind of a best place replacement scenario even early beta, as their previous solution was a 10/2 connection that cost >$200/month, so Starlink didn't have to be stellar to beat the competition there. I was able to place it with zero obstructions. This was never used with the native router which just stayed unopened in the box, it was right into a OPNsense gateway. We left the old landline connection up for a few months with Starlink as primary and the old one as a failover. For the first month and a half or so, there were regular dropouts every day, though only for a minute or two. That noticeably dropped over the course of March, and we dumped the previous service by mid-April. Overall experience has been excellent for wireless, as good as a high quality WISP despite the much greater technical challenges. And it's continued to improve: as well as latency and uptime, bandwidth u/d has increased from around 100-130/10-30 to nearly always 220-250/30-45. At no extra price. Tell me the last time you got that from an ISP ;). Overall experience is similar to this review, a few observations following along the review: >Hard cable This is annoying, but my perspective is coming from someone who does a lot of network termination and has thousands of dollars in fluke testing gear, rolls of cat8 cable and certified jacks and so on which is obviously absolutely not the norm. I needed it to go farther, but I ended up just plugging the 802.3bt injector it comes with into fiber optic anyway which thus also gave me guaranteed isolation from the rest of the network without having to worry about grounding. I suspect once it's in full service and they start to branch out they'll do a "Business Class" version or something aimed at more advanced deployments. >Your own router This does work fine. In order to get statistics though you'll need to static route 192.168.100.1. It's not needed for it to operate, but probably still a good idea. Lots of instructions around on the net for that. When I first set it up, IPv6 was still wonky whether you had your own router or not. Easily worked around with a permanent WireGuard tunnel for a VLAN to a cheap VPS with a static IP. Anything that needs a static IP can just get put there. >Power consumption Real, but as the article said in the context of rural internet pricing it's frankly fairly meaningless. Effectively means it's $105-115/month which is still so much frigging better than most options available at the speed that it's plenty competitive. If we had fiber we'd use it, but we don't. >Latency Where that will REALLY get interesting is once intersat optical links go live constellation-wide (can't remember the timeline on that, v0.9 for polar went up last year, I think they're now on new ones going forward). Most latency tests are local hops only, which is relevant of course and Starlink will always be at a disadvantage for. But once you start cross continents or oceans, Starlink definitely has the potential to easily beat many people's fiber connections (particularly rural ones its aimed at). CDNs are a lot, but not everything. It'll be cool to see how that affects the market. >it's potential to increase the risk of the Kessler syndrome This is a meme, not reality. I wish it would stop coming up. Starlink sats are in quite Low Earth Orbit (LEO), and the majority of the constellation yet to be launched will be V-band ones in V(ery)LEO. At those altitudes, natural orbital decay will eliminate debris or sats even if they fail, and if they don't they deorbit themselves. SpaceX has thought this through, part of the delay on optics was specifically figuring out how to make sure they'd all burn up. Other planned megaconstellations higher up do indeed represent more concern, but Starlink does not. SpaceX is leveraging its overwhelming and soon to grow more-so advantage in launch capability and economics in a host of self-reinforcing ways, and this is one of them. They get to lean on higher numbers and faster replacement to get more performance and stop Kessler risk. >Astronomy They've been working on shading them to reduce apparent magnitude, but that only fully works when sats are fully deployed and the trains going up are there. Even reduced magnitude will still affect some instruments and observations. I'm afraid though astronomy is just going to have to deal with that for a decade, looking forward to major development of cheap space based industry allowing lunar/space telescopes like never before eventually supplanting ground-based entirely. Cold comfort to a few who really will be messed up at a point in their careers that will be awkward for that, but this is frankly more important. >Other This will be MIND BLOWING FOR MARINE/AEROSPACE. If you think your rural connection is bad go look at merchant marine pricing. Aircraft too obviously, but one area I think many people haven't thought about yet is the potential for other satellites themselves. SpaceX is really setting themselves up as a major space infrastructure company.