A lot of the things here I'm seeing (like the don't do X to my favorite character list) are more things you don't like rather than lessons. Which I can fully understand disliking, but aren't exactly the things of which lessons are made.
Instead, why don't we look at the specific design decisions/experiments executed in ME3 and judge whether or not they worked? We all know that Bioware tests designs and concepts in one game that it's thinking of using in another. Here's my contribution:
Did the Kirkwall/Citadel experiment work?
In the last two Bioware games, the designers have made the very obvious decision to move us into one location, one hub. This is a pretty extreme departure from the Bioware norm, which averages 4 hubs per game. KOTOR, DAO, DAO:A, ME1, and ME2 all had four pretty sizable hubs each, if I remember right. And it's been a long time since I played it, but I remember NWN having 4 hubs. In fact, that became known as a pretty constant part of the Bioware formula. The last two games--huge, polished, high-budget affairs (DA2 and ME3) featured one central hub from which you ventured out to accept missions. Do you think this change had a positive effect on the game?
I'm strongly opposed to the shift, by the way. STRONGLY. Considering a large part of Bioware's strength is their ability to weave an interesting universe that your PC gets to interact with in a variety of places, confining us to one place makes for a much less immersive game. Exploration of the Asari corporate world (Ilum), the Krogan homeworld, the Citadel, and Omega brought us four completely different, fully fleshed out and engaging parts of the ME universe, each strengthened by the contrasts available. Having just the citadel in 3 meant that we had one place which seemed far blander than any of the four I just listed, as it had nothing to distinguish itself. On the other hand, I'd imagine it's much quicker to program and saves a lot of money. Plus in theory we could establish a solid connection to that one place instead of being travelers.
I think discussions on what lessons we can take from a game would be MUCH more productive if we focused on specific design aspects/decisions such as this. Other examples include Bioware's transition from plot-driven narratives to much more character-driven narratives over the last 4ish games. The quality of Bioware's story writing has gone down dramatically over the last 5-10 years. Dragon Age and most of the ME series didn't really have a plot at all. On the other hand, they had EXTREMELY strong character development. You could say that's what both game series were completely about. Do you agree with that decision?
What about the idea of time progression over a story? KOTOR, DA, NWN, and ME1 had very non-linear progressions in which you could visit whichever major plot points you wanted in whatever order you wanted. ME2 and DA2 had an interesting system in which the story was broken into acts but we could do plot-quests from each act in whatever order we wanted. JE and ME3 were completely on rails. What do you feel the pros and cons of the ME3 model were? The obvious pro is that it's much easier to weave a full, compelling storyline if you have fixed plot points at fixed times. On the other hand, ME3 didn't really have a storyline--really, we could have gone to each species in a completely different order and it would have changed nothing. We could have gone to the Asari, then the Salarians/Krogans, then Turians, then the Quarians/Geth. Given that they didn't make use of the more fixed structure we were forced into, I feel a much looser structure would probably have been better for 3 (and allowed for much more interesting DLC), but I could see how one might argue differently.
So instead of focusing on matters of opinion and whether we liked specific parts of the writing--which is an area that we pretty much will have no control over ever--is there any chance we could look at decisions that reflect design philosophy?