I've said this before, but: many political discussions are heavily informed by who the person speaking identifies as the relevant agent in some situation. For instance, consider this dialogue:
Person 1: The US needs to institute policy X.
Person 2: That's impossible, all the Republicans in congress would just vote against it. What we really need is for the Democrats to push for policy Y, which they might be able to get bipartisan agreement on.
Person 3: That's not possible either, the Democrats are too beholden to corporate interests to support policy Y. What we really need is for the president to institute executive order Z.
Person 2: Oh come on, the president would never issue execute order Z, policy Y is much more feasible than that.
And so on and so forth ad nauseam.
I don't think these people are really arguing about politics, and I don't know if they even have any substantive disagreement with each other.
In general, when people talk about "what should be done", they are always implicitly thinking of some agent, they are speaking of "what should be done by someone". And of course when we speak about different agents, we will come to different conclusions about what they should do. This is not least because different agents have different options in front of them. For instance, if you were to give me suggestions about what I could do to make the world better, they would probably not be the same as the suggestions you would give to Bill Gates about what he could do to make the world better, or the suggestions you would give to Vladimir Putin.
Normative claims presuppose an agent, and the content of normative claims will vary by the agent that is supposed. It would be useless to suggest to me "end the war in Ukraine", or to suggest to Putin "be more selective about the discourse posts you reblog", or whatever.
The problem is that when we are discussing politics, there are many different agents that we can identify with and whose behavior we can present normative claims about, and we often do not specify which one we are referring to. Furthermore, political agents can be institutions instead of just individuals, making possible the existence of sub-agents with varying agendas, and so on. Individuals might conceivably be modeled as having these too, but that's a philosophical can of worms I won't open.
Anyway, this imprecision about what agents we are prescribing actions to leads to scenarios like the discourse above, where people who substantively disagree about very little might argue vociferously against each other because in truth they are prescribing behavior for different agents altogether. Person 1 is prescribing behavior for the US government as a whole, Person 2 for the Democratic party, and Person 3 for the president. They only disagree in that each imagines the other's agent as an object of nature governed by mechanistic processes and their own agent as possessing (practically speaking) free will. None of them are really per se correct or incorrect, I don't think.
My suggested solution to this is: specify clearly the agent you are referring to, and admit that for normative discussion to make sense at all you must model that agent as "being able to choose its action" even if deeper analysis of its internal processes reveals it to in fact be deterministic. When in doubt, recall that the only agent whose actions you can really chose (if you can choose any actions at all) are your own, and thus in a certain sense any discussion of what an agent other than you yourself should do is idle philosophizing.
Ethics (I claim) are in and of themselves only a system for selecting your own actions; their use in evaluating the actions of others is secondary at best.