An other for blowback pieces of evidence. The Thompson
Blish lock rifle which used the Blish lock phenomenon with a rotating non-locked blowback bolt design. Lately it has been noted that dissimilar metals moving against each other under high pressure demonstrate fluid-like behavior, so that accounts for the increased friction vs predicted.
(Link to TFB article about this and the Thompson auto-rifle)
The rifle worked as basically as a blowback rifle, where as soon as the cartridge fired, it would push the bolt back and the bolt would begin to rotate it. Some of the rotation of the bolt and increased friction from the blish phenomena helped to slow the bolt down and give a greater "holding" power then just a strait blow-back or mass. This was by the way
with a .30-06 and 7.62x54r and not a .223, so significantly MORE bolt-thrust then what a .223 would deliver. Also, this rifle did need lubricated cases, but that is not anything that can't be solved with chamber fluting, as they serve the same purpose.
Again, I offer historical evidence for consideration that blowback rifle have been made, and they have worked safely.
And to re-iterate, for this discussion, when we are talking about blowback, we mean a system where the bolt is not locked to the barrel, and requires a separate outside mechanical force to unlock it, be it direct gas impingement, short or long recoil operating rod, or in the case of the G3 a second recoiling mass.