Anyway, to supplement that discussion, the roof may have a UL rating at which point the keyway may need something in there, possibly concrete. Alternately insulation might be stuffed in that area. I mention this because I started poking around on this topic and ended up here. As far as the tested strength of an untopped plank as a diaphragm, I'm not convinced there's anything formal involving testing. Some engineers may have the planks broken out to interlock them with rebar and site-cast concrete, but I'd question if that actually takes place versus it being on the drawings and nobody inspects for it, particularly in low-seismic areas.
Precast planks, if you ask me, don't get quite as much design attention as a diaphragm outside of the higher seismic areas. It's possible there are other things glued to the roof that would function (without formal documentation or testing) as a diaphragm, i.e. cover boards, insulation boards). You aren't supposed to rely on that for life-safety, but that might be part of the picture of why there isn't a lot of awareness of this loophole/flaw in the designs, particularly in older buildings (say 1990 and prior?). And there's always the "never saw design event" cop out.
For a fair discussion of the background, which wasn't brought up in that thread since it's from 2018 and the NEHRP/NIST guide was fairly new at the time:
NEHRP Seismic Design Technical Brief No. 13, Seismic Design of Precast Concrete Diaphragms: A Guide for Practicing Engineers, Ghosh, Cleland, Naito, Sept 2017.
My sentiment is that NEHRP is more interested in double T (i.e. parking garages) than it is "normal" buildings with precast planks.
