Stating "you don't have as much agency as you might think you have" isn't diminishing your agency; it simply makes you feel as though your agency is being diminished because your internal framework as to how play works is being challenged.
That kind of framing shows a lack of consideration for the creative goals of the person you're responding to. What counts as meaningful agency depends on those goals. If your model of agency is tied to one set of assumptions, and theirs is tied to another, telling them they have “less agency” isn’t neutral—it’s misaligned and can come across as dismissive, even if unintentionally. That’s how people end up talking past each other.
The same problem arises if someone assumes there’s an absolute scale of agency without first unpacking how the other person defines it or what kind of game they're trying to run. When that context is ignored, the conversation stops being productive.
"Sandboxes have near-maximal agency within the framework of trad GM-world creation play" is something I would generally agree with. The social contract of sandbox play allows for me in-fiction agency than one where the table agrees to run through module X.
"Sandbox play has near maximal agency for TTRPGs" is something I'm going to disagree with, because it seems to have blinders on to a lot of TTRPG play.
That’s part of why I started digging into the assumptions people bring to these conversations. I noticed a recurring pattern: we were talking past each other because different people were using “player agency” to mean very different things.
Eventually, I proposed that player agency in TTRPGs can be understood as a combination of two broad types:
Character agency—what players can do as their characters within the world.
Meta-agency—what players can do outside the fiction to shape the narrative or structure.
Framing agency this way gave us a more useful vocabulary for understanding how different systems prioritize or combine these elements.
Also, though it hasn’t been discussed much yet, each of those types has subtypes. So we can’t reduce a system to something like “80% meta, 20% character.” You have to look at the specific mix of subtypes to understand how that system expresses agency in play.
I know this adds layers of complexity, but that’s because the subject is complex. We’re talking about different play cultures, rule structures, procedures, and creative goals. If the goal is to understand what’s really happening at the table across styles, then we need to let the analysis be as complex as the situation demands. Oversimplifying just creates new confusion.