I just found this little footnote at the bottom of the “dynamic modelling” page.
*Designing it was not that hard - optimizing a mechanism that accelerates backward at 2000 g’s was the hard part.
An external hammer makes sense. There’s not much room for a sear to release a striker like in a Glock. The feeding mechanism is in the way. You could build a sear that reached over the feed area and make it in a U shape, like the 1911 trigger bar. This would make it rigid on both sides, but it would be an expensive part. The hammer is a proven mechanism, and this thing has some radical ideas. If I thought of it, I would want to use established methods wherever I could so that any problems would be easy to isolate.