In article , Eric Flesch
writes:

Let's pretend for a moment that there is no dark matter, but that the
gravitational lensing that we see happening out there is affected as
follows:

(1) The lens / target are at different distances than we suppose, and

(2) There is a migrating universal "constant" such that in earlier
epochs matter bent light more per kg than it does today. In other
words, lensing effectiveness is proportional to 1+z, or maybe the
square root of 1+z.

Either you are assuming a change in the gravitational constant (which is
ruled out---at interesting levels, at least) or proposing some method
other than that described by GR to bend light. However, if GR is not
valid, then it probably makes little sense to leave the cosmological
model otherwise alone and just change the lensing effectiveness.

For (1), my question is, if we alter the distances to lens and target,
even if very unreasonably so, can we recover the lensing that we see?
Or is the only working solution to make the lens much larger & further
away? A smaller closer lens can't bend the light that much, is that
right?

I don't follow you here. If the lensing strength is higher at high
redshift, why do you want to make the lens larger and further away?

For (2), my question is, is there broadly a redshift dependency in
lens power, that is, lens mass? Are high-z lenses seen to be more
powerful than low-z lenses, or is that susceptible to a Malmquist
bias?

Check out the classic Turner, Ostriker & Gott ApJ paper for a plot of
lensing effectiveness as a function of redshift or, for a non-zero
cosmological constant, Fukugita, Futamase, Kasai and Turner, also in
ApJ. Lensing effectiveness is roughly Gaussian when plotted against
redshift, but is down to almost zero well before the redshift of the
source is reached. It depends on the redshift of the source and on the
cosmological model.

A while ago I speculated that time dilation might go as the square
root of 1+z instead of the standard 1+z.

This seems rather ad-hoc.

This is because if there is
a migrating universal constant which operated on the space-time
manifold, then redshift would be half time dilation and half spatial
lengthening.

Why? Unless you have an underlying theory, I don't see how you arrive
at this.

In other words, the past would look bigger but this
self-corrects via Riemannian geometry. I'm wondering how this would
affect the lensing that we see, thus these questions. Appreciate any
help.

You need to have a more concrete model in order to make concrete
predictions. People have looked for signals of unorthodox models in
lensing and found none.