My source is bitter experience designing and testing counter-rotating electric ducted fans for distributed propulsion drone applications. You have to consider

My source is bitter experience designing and testing counter-rotating electric ducted fans for distributed propulsion drone applications. You have to consider specific failure modes. You might have an object strike, like a bird. You might lose sealing and get water in somewhere, frying anything electronic inside. You might have a loose bolt, a fatigued bracket, or a frayed wire. When you start considering scenarios rather than parts lists, it becomes clear that consolidating things together spatially costs you most of the advantage of redundancy. This is especially true in the case of props. In point of fact, if a prop breaks the net force is upwards, because that's the direction it was pushing in when it failed. However it's basically irrelevant - unless you perfectly and flawlessly detach the blades symmetrically with explosive bolts, with impeccable timing such that no detached blade hits the airframe, so much energy is released with a moving-part failure that anything in the vicinity is pretty well destroyed. Experience you might have with normal-sized drones is not applicable if you scale it up even a little - big props make a big mess, and you're lucky if the airframe survives at all. You certainly won't be able to count on that shared pod. We can only guess at why they chose to design it this way - personally I think they're a bunch of cowboys who will kill someone the way they're going. I suspect it's because they want to keep the footprint small, and because it's kind of uncool to have to pick your way through a dense ring of props to get to the cockpit (if you look at some of their videos you'll see that they've experimented with non-coaxial x8 configurations in the past). But stacking the props two-by-two like that only costs you reliability.