Z--the AK variant was a Valmet M76. I believe it is a specimen defect, perhaps with excessive headspace combined with excessively vigorous and perhaps early case extraction. On the feeding problems, I believe all the mags are incorrectly desiged--the front is cut out a bit from the sides=only about 1/16 inch of metal up front should be enough to tip the bullet noses up into the chamber before the feed lips release the case. There is about 1/16-inch of copper smear at the 5:30 and 6:30 positions at the back of the barrel below the chamber, showing where the bullet meplats hit on feeding.
With USGI Ball ammo, it works flawlessly. Reloads with bulk Winchester 55 FMJBTs get pulled apart mid-case, and jam the bullets into the barrel below the chamber at least once per magazine.
BB: looseness in the gas system and other places you mention does nothing to improve M14 reliability AFIK. Loose tolerances in the op rod raceway, the bolt raceway, the op rod/bolt lug interface, and the trigger group/bolt interface and related areas, will affect reliability. However, these areas are not a demonstrated problem in the M14 series, because they are already loose enough.
There are some theories that loose gas cylinders can reduce reliablity, by allowing gas bypass to reduce the operating power available--compare to the NM specs on gas pistons outside diameter and gas cylinder inside diameter on the M1 rifle. Too loose is rejected because of its effect on reliability, not for its effect on accuracy.
Gas cylinders to barrel fit on the M1 affects *user* accuracy because that's where the front sight is mounted.
Gas cylinder to barrel fit on the M14 is reported to affect accuracy, but AFIK, it's not guaranteed to improve accuracy. IOW, it's not a reliable variable in an individual rifle's accuracy.
BTW, how did the M16s perform in that North Pole ice cap exercise? Were they even deployed for that play session?