I don't ask that an attribute system perfectly simulate all things. However, in a role-playing game, its goal should probably be to provide a full-spectrum of significant character distinguishments. That's the core of my issue with the PoE system. The only problem with Might that I have is that the system essentially says "Let's just assume anyone with X amount of Might can do all the exact same things that anyone else with X amount of Might can." It would be like combining armor and dodging. If you just say "Meh, at the end of the day, all that matters is whether or not you took damage," then you've diluted the whole role-playing aspect of the game world. Do you dodge things well, or actually nullify attacks that strike you? Ahhh, who cares. You avoid damage. Everyone is just a damage-avoider now.
The only reason Might "isn't that big of a deal" in PoE is because it's not really checked as significantly as it should be. This is the problem with Strength in 99% of RPGs. It's why there are dump stats. "Why should I take any Strength, as an (insert non-Warrior class her)? What... to pass those 2 Strength checks in the whole game that let me get past some minor obstacle, or achieve an outcome in a scenario that already had an alternative route to it anyway?" Same with Intelligence for Fighters. You maybe don't want them to be so dumb they can't function, but what's the use in their being smarter than a 10-year-old? You're spending an attribute point and getting FAR less for your "dollar." The solution to that isn't over-simplifying the attributes. Why take away the robustness of the system?
This is why I LOVE some of the things mentioned/suggested in this thread. Give Fighter-types a sub-grouping of abilities and talents with INT pre-requisites, for example. Make more checks to attributes, directly, and/or skills, directly.
And don't get me wrong... I know that with the first game, they had a pretty huge crunch, and were building all their assets and figuring Unity out from scratch, basically. But now that they've got a little more wiggle-room, I don't see it as unreasonable to try and get closer to some sort of ideal with attributes. The ideal isn't "make it just like DnD." But DnD had a pretty good foundation, coincidentally because of what all the stats represented. There are a ton of different attribute systems in different table-top games, but all the best ones are the ones that actually measure interesting, differentiating factors about your character, then are backed up by a system that actually uses those measurements to produce significant and dynamic results. When we went from table-top to CRPG, the tech at the time couldn't really handle the vast majority of the checks and attribute usages that you'd see in a tabletop session. Thus, the attributes were reduced to "how does this affect me, numbers-wise, in combat?", with dialogue checks trailing behind in second place and a sprinkling of other checks throughout.
Now that video games have come so far, the solution is not to reduce the attribute measurements down to the level of existing gameplay systems, but instead to bring the existing gameplay systems up to par with the attribute measurements.
Obviously Deadfire's budget only allows for so much, so if they can't do something, they can't do it. No one can feasibly be mad about that or demand that it be done anyway. However, to just say "Meh, the attribute system's okay, I guess. Better not waste any time even considering improvements to it" would be ludicrous. Proposing and discussing changes and improvements to the system and brainstorming reasons for them doesn't hurt anyone, and can only potentially contribute to a better understanding of attribute design on the part of anyone reading or taking part in such a discussion. So, I say collaborate away.