I have never been able to consider the main triangle, structurally speaking, as much of a triangle at all.
Considering that the head tube is cantilevered to the top and down tubes, this part of the "triangle" would seem to offer little in the way of triangulation with regards to the loads imposed on it, and thus little of the stiffening effect of triangulation.
Unless the top and downtubes are relatively over-engineered, it would seem that the guage of these tubes might play a significant role in the frameset's flex (out at the fork tips where this might best be measured and compared to the entire fork's own flex). My experiences repairing crashed frames tells me that the head joint areas are not so over-engineered.
As well, any flex in the top/down tubes will be amplified by the considerable distance from the fork tips, though these joints are relatively large in diameter and, of course, stressed as a pair (as are, to a degree, the fork blades).
In regards to the seat stays, it's hard to imagine them compressing significantly. To the extent that any added curvature enhanced such compression, this would have a very undesireable effect on keeping the rim centered between the brake pads, as small side loads (where road meets the tire) would be highly leveraged by the relatively narrow spacing of the dropouts, thus compressing the seat stays un-evenly. A flexible mono-stay above the brake bridge would seem the only practical place to introduce rear triangle flex in the interest of comfort (an old and successful idea to say the least!).
I'm almost ready to conclude that the fork (together with the front half of the main triangle) and the above-bridge part of the seatstay treatment are the only practical places to add flex to a minimalist (weight-wise) frameset structure. I don't much buy into Colnago's or Ritchey's or Specialized's (relatively recent) paired, and supposedly flex-inducing, manipulations of stay cross-sections.
Auburn, CA usa
\r?\n> From my experience, the fork blades make the biggest difference in comfort
\r?\n> (after tires). (We measured flex of different fork blades, and it was
\r?\n> quite a bit different.) I am willing to believe that seatstays also can
\r?\n> make a difference, but haven't investigated this in a repeatable test. The
\r?\n> rest of the tubes is so well triangulated that I have a hard time seeing
\r?\n> how they could affect comfort.