Yes that's what I'm seeing. Thanks everyone for verifying it isn't possible. The reason there were no de-couplers was that it's easier to illustrate the problem -- only one motor will be sheathed, showing that there's only one connection point. If I put four de-couplers on, four motors will be sheathed, but it still doesn't snap in more than one location, and still collapses on the launchpad, except the cause is much less obvious to see. As for answering the higher-level problem: I actually want to do this with the nukes, but as they were blowing up on launchpad, I figured I'd get a less volatile engine working first, before going back to that. I had already tried strutting -- eight "200 I-beams" in turn clad with M-2x2 structural panels, but that didn't brace it enough. I admire the cleverness of the docking port design, but I've found docking ports to be even wobblier than decouplers. In general, the physics engine is more wobbly than "real life" as far as I can tell. Mild Steel has a shear strength of at least 50,000 pounds per square inch; alloy steels much higher. The connecting area of even a small decoupler or other ring connection would be in the hundreds of square inches, but they behave like bubblegum and duct tape. I think this is because KSP uses the default 32-bit game physics engine of Unity?