Starlink shines when it comes to rural/remote environments, not cities where towers and fiber can reach.
Side note, look at HughesNet and Viasat in their data, LOL!
edit: instead of downvoting if someone could find the sampling method?
When you make up an accusation from whole cloth, people are going to downvote it. They'll probably downvote even harder after you ask for other people to give you proof to refute the accusation that you made up from whole cloth. If you find the sampling method, and it turns out your suspicions were true, and then you write a comment explaining that, they'll give you hundreds of upvotes to offset these two or three well-deserved downvotes.
Just because there's a lot of people in highly populated cells that drag down the speed for people in those cells has nothing to do with the speeds of the lowly populated cells.
tldr; it makes no sense to average all cells together, as the goal is to improve the areas where existing infrastructure have failed in specific regions.
(and in those areas, where hughesnet, or viasat, or old DSL were the few options, Starlink does it's best)
(This is oversimplified but puts the capacity scoping in context)
So if you take a cell that has LA in it, it will be much more congested than if you took a cell in Wyoming, or Nebraska.
In theory most or all of the people in a major metro area would be opting for faster, cheaper ISPs. Starlink is for people way out in the boonies that don’t have any other good options.