Could you explain what I'm missing? Have we reached a point where the returns of increasing size are very low or is it more that there are enough big mysteries to solve at the smaller size so it isn't much worth increasing until it can be done cheaper?
Say for JWST, there are 18 segments. Could you have 1800 mini Webbs and offset the inaccuracies that come from that with the vastly higher total area? Does that eventually reach an asymptote where a adding more units isn't useful? Does it reach a point where the data re-integration becomes the intractable issue? Can it ever be as good as a gigantic monolith even with an arbitrary number of units? Would the cheapness of individual units be outweighed by the number of them needed?
Combine a wealth of data sources in many different directions and many different observations can also lead to new insides.
When majority of the community, a whole generation of scientists, all have to wait for one massive telescope that can then only do 1/1000 of what people like to do with it, are you really gaining more insight?
Sometimes yes, sometimes no.
These mega project that cost a gigantic amount and are hilariously oversubscribed are not actually the best return on investment in my option.