The part of the NS article visible above the paywall says "the carbon footprint of each satellite constellation is potentially 14 to 21 times higher per internet subscriber than the emissions associated with land-based..."
But the arxiv paper says nothing like that! It focuses exclusively on carbon emissions from satellite internet.
Because, obviously, the baseline here is important. Starlink exclusively provides internet to rural users-- it has to, each footprint is limited by physics. If the limit is 10 subscribers per square kilometer, then obviously it's going to be a lot more useful in the Mojave than in downtown Vegas.
Running fiber 10m down the street to connect two houses in a suburb is much cheaper than running it 100km to someone living on the top of a mountain. Is the story comparing the carbon cost of connecting people in Manhattan, or the cost of running a cable to Srednekolymsk? No way to tell, since I'm not paying to read this. But it's definitely not addressed in the arxiv paper.
---
Minor aside:
>by the number of satellites in each constellation, along with a factor which represents the average amount of time each satellite will spend idle over ocean (rather than over land where subscribers are located).
Huh? Shipbourne users are a substantial market for Starlink, and obviously one that is completely impossible to address with terrestrial solutions. Unless you want to carpet every ocean with cell phone relays, which would have a... substantial carbon footprint.
But the arxiv paper says nothing like that! It focuses exclusively on carbon emissions from satellite internet.
Because, obviously, the baseline here is important. Starlink exclusively provides internet to rural users-- it has to, each footprint is limited by physics. If the limit is 10 subscribers per square kilometer, then obviously it's going to be a lot more useful in the Mojave than in downtown Vegas.
Running fiber 10m down the street to connect two houses in a suburb is much cheaper than running it 100km to someone living on the top of a mountain. Is the story comparing the carbon cost of connecting people in Manhattan, or the cost of running a cable to Srednekolymsk? No way to tell, since I'm not paying to read this. But it's definitely not addressed in the arxiv paper.
---
Minor aside:
>by the number of satellites in each constellation, along with a factor which represents the average amount of time each satellite will spend idle over ocean (rather than over land where subscribers are located).
Huh? Shipbourne users are a substantial market for Starlink, and obviously one that is completely impossible to address with terrestrial solutions. Unless you want to carpet every ocean with cell phone relays, which would have a... substantial carbon footprint.