This seems to me to be talking about system - ie what system is being used to resolve encounters?
I think it can be about system. The system itself can provide detailed rules for resolving some challenges (most often combat) and leave others to GM fiat (perhaps social/interaction activities), exactly as you say. It can also be on the GM where the system provides mechanics for resolving some type of challenge, but the GM ignores these in favour of a GM fiat approach.
This might result in player skill, rather than character ability, being determinative of success or failure. For example, where the GM bases success in social encounters on the player's persuasiveness, speechmaking, glibness, etc. rather than the skills and abilities of the character representing persuasiveness, speechmaking, glibness. In this case, investing character resources like feats or skill points in such abilities is pointless, as they will not influence the character's success.
Or it may simply be that the GM wants combat to be the method of resolution, so all of the important NPC's/adversaries are immune to persuasion and the PC's can succeed only by battle, not by other means.
Either approach logically results in characters focused on combat strength, since investment in interaction skills carries no reward.
I am not that interested in systems where the resolutin method is "GM fiat". It puts too much work on my shoulders as GM, and also leads to conflicts of interest - I have to, at one and the same time, (i) push my NPCs hard to create pressure on the PCs (and thereby on the playes) while (ii) decide whether or not I want the PCs to win or lose the conflict. In my experience that does not produce very satisfying play.
Agreed. To me, a good game comes from the GM placing the PC's in challenging situations with, perhaps, some thoughts on how the players may resolve the challenge (ideally with thoughts on multiple approaches) without forming a vested interest in one method over others, or even in the success or failure of the characters (again, ideally with a plan for how the game will proceed either way - sometimes, failure may carry more interesting results than success). To me, the GM forcing a specific approach to problem resolution is the epitome of "railroading" in its most negative form. Hey, why not just tell me how you want us to resolve the situation, or better yet, save everyone's time and email us the preconceived story you've worked out - obviously, our characters' choices won't change the story anyway. I think you and I are on the same page in this regard.
Of course, approaching the game in this manner can be a difficult skill to cultivate - the GM builds the world, and it's tough to eliminate any bias creeping in - but the better GM's are certainly more skilled at minimizing that bias.