The year is 2026. Surely the orbits of all the satellites are known well enough, and optics are modelled well enough for telescopes to know which few pixels to ignore at any given moment?
1. many astronomical observations are long integrations (many minutes of open shutter) and you often cannot read out the pixels during the exposure. So you can't selectively ignore pixels on a "by moment" basis. And with enough satellites, enough pixels could be affected to render an entire image effectively unusable
2. It seems you're thinking only about optical astronomy. There's also a ton of radio/millimeter wave astronomy that's done from the ground. Satellites have radio-wave downlinks that can be powerful enough to destroy the electronics used in radio astronomy receivers. The US's National Radio Astronomy Observatory has been working on data sharing with Starlink to mitigate this, but it's basically up to companies to agree to work together. Other satellites / companies don't engage in this coordination and so some radio telescopes need to go to a "safe"/stow position to protect the electronics. This costs valuable observing time.
I suppose the only downside is cost. So, the actual problem, is how to tax the shared resource that is our LEO. Aside from that, wed also want to incentivize not leaving space debris in orbit. Require insurance, and perhaps rent for orbital trajectories?
But, without working international order, theres no way thats gonna happen...
Right now for the price of that one project one could have a humongous radio telescope in orbit without any interference. If only people could really look forward.
Radioastron required cross-correlating the signal with ground-based radio teleacopes in order to do science. The collecting area, dish distributons, etc. would need to be fully replicated in space if one wanted to avoid using ground-based facilities entirely.
Now we can launch two. Or two dozen. This completely obviates any need for ground listeners as for resulting angular resolution.
Since ground-based radioastronomy has other uses, will stay there anyway and since crosscheck is always better, future long-base radiointerferometry will use legacy ground-based observatories.
One would need to put the telescopes into fairly high orbits in order to avoid the interference. I think that even images made with the Hubble space telescope have sometimes seen the impact of satellite constellations.