it seemingly values targeting minion line with a near-queen higher than targeting the champion line...?

I think the most relevant part for me is 17-18 being a minor difference, because what that changes is "I have to move to row 3 vs. row 4 to get this effect". I believe the minion line is more important because the near-queen would be in support of a potential weak square, not the one that's attacking.

There is also the thing where the more expensive a unit is, the less likely it can be traded for anything but a King, and this implies additional melee moves have diminishing returns from trading perspective [...]

It's true that it implies it has diminishing returns but I am not sure it's necessarily only something that happens for expensive pieces. I feel like undercutting is usually annoying for the player with the slightly-more expensive pieces (ex: Legionary+ player against some Legionary++s, the Legionary++ player has the worse deal). I almost suspect extra granularity makes this worse because an incredibly discrete system (like if you only had 6-cost, 10-cost, 14-cost, etc. champions) gives you more control over undercut situations in hopes to make it more balanced.

The history with the 4 minions was: F3 put in a piece in our game that was C in the picture, with 16 cost (so cheap-to-mid-cost minion) and two other upsides. Nobody put it in their army until one time I was just looking for a final minion to put in, and put in that one. Ryan instantly complained about whoever put that in and thought it was obviously overpowered, then made that balance test to try to prove it:

Me: A>D>C>B
Ryan: C>A>B>D
F3: A>C>B>D

Naturally agreement was nowhere to be found.

Here's my case: A is Militia, obviously really strong. D is more like a defensive Militia. The moves are much more awkward to use but I believe its downside can be mitigated by having it on defense. The range 2 minions can be used on defense as well and potentially threaten a forwards-fork, but with some risk that the blockable attack could just let one of the forked pieces move away while blocking the other threat. I don't believe it is a remarkably good attacker, but it's a decent defender (while its blockable spaces are in ally territory). The worst piece is B as it can only hit 1/4 of the board's spaces and most of its moves create overlapping threat ranges (while even though D itself is ¼bound, its attacks can hit a bit below 1/2 the spaces even though it's rowbound).

There is an alternative strat with C/B though, and it is trying to use them almost strictly like a control piece, by 'blocking off' an entire row (imagine them on the d/e columns, which targets b/c/d/e/f/g). This does have value cause this row is one higher than you'd get by trying to do the same with A/D (putting them on c/f).

-main_gi