It's made out of the same plastic as the other Ditra uncoupling mats, if that helps. Once installed, and you've filled it up with the thinset and set the tiles, the actual flexibility of the mat is what provides the stress-relief in the lateral plane. The air space in the towers is what allows it to move horizontally without cracking the thinset or breaking a tile(not vertically)- IOW, it MUST be flexible to obtain the uncoupling. Keep in mind, you still have the rigid tile bonded to the thinset that keep the towers in place, only deflected as a whole assembly (tile/thinset). The areas where the thinset goes to the base of the mat are what provide the vertical support, and their spacing is the reason why there is a minimum tile size requirement - the tile needs to be supported by a minimum number of thinset towers to provide adequate compressive strength, only limited by the compressive strength of the mortar used. If you don't wear kneepads when working on your knees, you might want to use a kneeling pad, but you won't hurt the mat in the process - it will be more comfortable for you. That's pretty much true for any uncoupling mat.
That is one reason why, as opposed to say a crack isolation membrane which needs to span a ways over a crack to have enough flexibility to actually isolate it, an uncoupling membrane does that from its inherent structure over the entire mat, regardless of its size - things can move any direction laterally, independent of the subfloor or slab and the width (contrary to what John thinks) has no impact on its decoupling capabilities.
Any test trying to exceed the industry standards with any mat are misleading and mostly useless...all membranes of this type must meet a minimum of 50psi to pull it off the substrate (the fleece attachment layer will fail first), and that will be the weakest link in the system. The only way to determine how well it works for uncoupling would be to install multiple strain gauges in the sandwich, and measure how much stress can be applied before either a tile or a grout joint fails or breaks. No backyard test is going to do that, and with crude measurement tools, is mostly a joke and pointless. These things are designed for use on a floor, and you're not going to have something yanking the tile up (gravity doesn't work that way) or pull it sideways. Yes, the tile and the substrate move separately, but that's the whole reason for an uncoupling membrane, to isolate them. For any practical installation, it far exceeds anything needed. It does NOT reduce the need for expansion joints IN the tile - it only works between the substrate and the tile - the tile's expansion can exceed the strength of thinset regardless of whether it is attached via bonding directly or whether it is placed on an uncoupling membrane. The beauty of the uncoupling membrane is that thinset does NOT stick to it (at least on Ditra - why one would want fibers as part of the tile bond is beyond me, and probably mostly to avoid patent issues - they state that they tear in the process, anyways, so why have them!).
The whole concept of you must use a modified thinset for attaching porcelain or glass IS true, IF you direct bond them in many cases. When using an uncoupling mat (and the installations from centuries ago used their version of an uncoupling mat - a packed sand bed - no cement, just sand and their glass and porcelain installs work just fine), you have probably at least a 300% stronger bond to the tile than any lateral stress that can be applied to it. IOW, it isn't going to come off unless you try to destroy it!