When designing an SWB (Short Wheelbase) recumbent I believe you should always try to get the weight as far back as possible. However, that doesn't mean that getting the rider back is going to improve the handling of an existing SWB, especially one with USS (underseat steering).

Tim Fairbanks' Haluzak is set up with the rider fairly far forward, giving it less than optimum (but not “unicycle with a training wheel”) weight distribution. When I pointed this out, he said he tried it with the seat and cranks further back but handling was worse.

Two possible explainations.
A. With a fixed steering pivot location, moving the rider back might have put him too far from the steering mechanism. Either giving him too much tiller, or putting the grips too far from his shoulders. This is primarily a problem with USS as most ASS SWB recumbents (except LowRacers) would be more stable with added tiller.
B. More likely, the problem stems from the suspension fork. My guess is that the spring rate is high enough that, without a fairly good static load, the fork is only active on good sized bumps, and may be topping out. With Tim over the front wheel, the fork may always be active so the front wheel stays on the ground better.

This suggests two things
1. With their short, steep forks, SWBs have a lot to gain from suspension.
2. Spring rates and/or preload should be carefully chosen to allow static sag for active suspension.

He is using a cast aluminum wheel with Suntour brake internals (the best he has found so far in other respects). There was little consensus among the gang about how much input force a coaster brake really needs, so Denis will just have to keep experimenting and settle the question.