At 27 pages now, it's a long read, but I've tried to explain as best I can in prose and some screenshots the limitations of the FIC, which is true for any piggyback. The stumbles have nothing to do with the MAF signal "getting out of whack". The AEM is definitely the best piggyback once you get the MAF signal at an appropriate voltage level and the O2 modification working, but it's still a piggyback and at the mercy of the PCM. Mine was perfectly stable, with no hiccups or stumbles, but it took many weeks of data logging and experimenting to get the fuel and O2 offset perfect in closed loop. What I couldn't control was when exactly the PCM would switch between closed and open loop, since there are scenarios that the AEM could not account for.
The EMS4 does not have those limitations, but is priced accordingly. I am leaning towards the Megasquirt now, because it is a lot cheaper, has limitless expandability, and open-source software. I haven't ever used one, and not really sure if I have the time needed to get it running right, but it's not a terribly expensive experiment, and I figure I'll learn even more than I did with the FIC. I just need to stop looking at these expenses as blowing money on an old car, and more just as the education and entertainment of a hobby.