Ann: Results of Annual General Meeting, page-194

This is a good point - technically the key difference is 4ds is interface switched over the entire surface area of the cell, vs wbt which is filamentary and creates a conductive channel from one electrode to another, before breaking this down again (it forms the filament using silicon nanowires, which is pretty cool!). Obviously both approaches are just changing the resistive state of the cell (hence ReRAM - 'Resistive' RAM for the newbies).

To your point - both approaches, interface switched (non-filamentary) or filamentary (in wbts case) - are doing the same thing with the the oxygen ions (so end result of both approaches is the same). The real proposed magic comes in with geometry scaling, and also the least number of 'moving parts' in the architecture of interface switched, which means cost of manufacture can be kept minimal. If 4ds can pull off the 20nm in PL6, we'll really be looking at something special (imo). As others have mentioned above, there is plenty of cake for all reram candidates; it's a big market. Exciting times ahead.