Pre-MechClix BattleTech's lore didn't make sense in terms of how real-world conflicts work, but it was at least mostly internally consistent. The people in the setting were all humans, they had human motivations, and while the way the setting works requires a lot of very far-reaching assumptions (starting with why would anyone use big stompy robots instead of tanks, and ending at why fight battles on the ground at all when space is the ultimate high ground), everybody was playing by the same rules. Literally, since all units are built using the same construction rules and all fights are fought using the same combat rules.
That's what drew me (and I suspect a lot more people) to the setting. The lack of technological ass-pulls, the consistency in rules, the fact that even if the Davions were clearly the authors' pets everybody delivered and received regular ass-kickings, and that the setting was actually well designed for eternal conflict. Compare it to Warhammer 40,000, where the lore bends over backwards for the sake of Games Workshop's overpriced plastic sales. Sure, they have some very good novels (and whole cycles) published by Black Library, but the good lore happens in spite of the game. Because if the market research team tells the design team that Orks need a new range of vehicles that go completely against the Orks' design concepts then by the Emperor the design team is going to make those new vehicle models, throw them to the intern writing the lore, and it's all canon now. BattleTech's lore might have been rocky in places but until MechClix at least it was more or less coherent.
Post-Clix... yeah, I'm sticking with the FedCom Civil War. I love how the pie chart map got broken up and there are more avenues for brushfire wars and mercenary shenaniganry, but Jesus Christ on a pogo stick these writers don't know how to write compelling characters. Or conflict, for that matter. It's not unsalvageable (they managed to mostly salvage the Jihad period) but it would require some pretty intense staff turnaround first.