It's a perfectly valid point of view that effort might be better allocated to "finishing" major related systems as opposed to completing unrelated systems in a (seemingly) random order. I don't know why such an opinion is so controversial.
I, too, would prefer devs focus on either A2A or A2G related systems. The order isn't important to me, but it would be my preference to have either a fully working A2A or a fully working A2G solution as opposed to having both in an incomplete state.
ETA: I actually like getting individual systems added every few weeks. It makes familiarization easier, allows consolidation of learning, and is not entirely unlike real acquisition programs where planned functionality is added over time. I'm looking at you, F-35.