I understand what you are saying, but I think we are both looking at this from slightly different angles in terms of the stresses involved. I was speaking in terms of a 2d vertical movement of the shock towers.
I suppose it would also depend on the particular design of the rear bar you are using also. I can't see how the rear shock mounts would see a lateral shear load, like a strut would. The load that we were talking about was a vertical load on the mounting location. If a bar was designed that had two different bases that had a bar bolted into them I can't see a single bar being effective. The bolted joint would allow some pivoting, significantly decreasing the effectiveness of the bar on reducing vertical movement. In this case a triangulated bar going down to the trunk that bolts directly to the mounting bases would be effective. *However* if the bar were a solid one piece welded unit, as many on the market are, you loose this weak joint and have a much stronger design, in this case the need for triangulation is nill.

If we are just looking at stiffening up the rear frame of the trunk the RS bar handles the job just fine.
I still can not see what loads the fore and aft bars in this "cage" are designed to work against. I fail to see the shock mounts and rear trunk being stretched/compressed. If there is a lateral force trying to twist the trunk on an axis closely parallel to the ground between these four mounting points a rectangle is not going to provide much support without a diagonal.

 

Exactly! Consider a structure like |==================| with vertical loads applied at the ends. To first order a small vertical displacement has no effect on the length, which is the only constraint something like a strut tower bar can apply.

I couldn't have said it better (I know, 'cause I tried). My question is still, what problem does this trunk bracing address? The rear end of the evo (at least mine) is already stuck to the track. It's the front end that needs more grip.

Running more roll stiffness in the rear is a perfectly reasonable thing to try. On turn-in running more roll resistance in the rear makes that end do more of the work in total roll resistance, unbalancing the load on the rear tires (reducing grip) and maintaining the load balance on the front tires (improving grip). The only way that increasing chassis rigidity in the rear can have the same effect is if the chassis were so flimsy that its deformations were of the same magnitude as the suspension movement! Joe Random Evo driver's comments about how it feels carry 0 weight without real measurements of chassis deformation or lap times. For 99.9999% of evos products like this have more effect on the driver's head than on the car.

Dave

 

I'm glad what I said made sense to someone, getting these thoughts into words without a diagram is hard, and it's only harder to understand.
I think the best thing to do here, short of a scientific analysis, is to see if someone can dig up a pick of the trunk of one of mitsu's WRC cars to see what happens to the cage in the rear. If anyone knows how the unibody is going to deform and where it needs bracing it would be the works team that made the cages.