[mturmon]

The telescope in the OP whose observations were interrupted has a 4m primary mirror. Hubble has a 2.4m primary mirror.

The optical corrector lens -- just the corrector lens alone -- that is part of the camera instrument of OP has a diameter of 1m, with a mass of about 200kg. (https://www.darkenergysurvey.org/the-des-project/instrument/). The entire Starlink satellite mass is 227kg.

I don't think SpaceX is going to be putting instruments of this class on its satellites.

[kd3]

That is not what I am suggesting either. Smaller telescopes and cameras but on every satellite which can then be combined computationally as well. Tons of possibilities. Of course, like I said, they could also equip some of them with more powerful and larger hardware. But the smaller ones would already provide lots of capabilities for very little extra cost relatively.

[mturmon]

OK, but that's not addressing the problem raised in the OP.

Dedicated CubeSats have largely filled the smallsat niche. I would expect the Starlink mass has been carefully optimized already.

There are interesting mission concepts (telescope + star shade; NEO detection [1]) that are in the works with dedicated CubeSat constellations, but yoking them to another spacecraft bus with its own constraints doesn't sound likely to succeed.

[1] https://www.aanda.org/articles/aa/full_html/2017/07/aa29809-...