We have data, we know how efficient driving the wheels directly can be (and an electric motor still has to do that as well), we know how to keep the ICE happy and how not to use it when it isn't happy. We know that series starts off behind based on peak efficiency, and no reason to think that it will ever catch up to an equally optimized pusher. Everything else is speculation as to why we should think series is going to be more efficient under some loads/speeds. The default assumption here is that parallel starts out ahead and thus stays ahead. And if it isn't, well it is more than fair to say I will believe it when I see it.
I do not lord over empiracal evidence. I do not see things in absolute terms, but I don't see the rationale for the belief without the evidence, yet I see a lot of believers. I am comfortable with probability myself.
I don't know how series can close the efficiency gap with all the conversions, what do I have to even speculate about it? More speculation? No thanks, I've considered much speculation already. We have a few data points, and it doesn't look so good.
Even the fvt aero didn't do so great well to wheels, since mpge largely compares electricity (which is basically temporary storage for far less efficently burnt coal) vs gasoline based on energy equivelant. A big part of the reason to keep these variables separate instead of mushing into one meaningless number. There is literally no good comparison from an ICE car to an ev, until you attach an ICE to your ev, in which case there is a pretty direct comparison. Ask any hypermiler how to get good mpg with an ICE, plenty of data there.
Maybe something like fuel line magnets, only little pouches of fuel that you tie around the generator leads would help?
