Your problem is a structural crisis. You would need a reform of your voting system to effect a stabilization of the political system, o.e. abolish fptp and gerrymandering. I don't see that happen.
So you are fucked.
We need a system that allows a freer market of political parties. With two entrenched private organizations that are impossible to unseat, an extremist takeover of one is like taking over the country... Which is what has happened. Imagine a system where rational Republicans could have formed a breakaway conservative party and had a chance at success without being a spoiler.
And I think the new voting system has to have preference built in. I know approval voting is simple to explain and calculate, but I want to maximize the "expression of voter preference" as much as possible while keeping the process simple. Ranked choice has some issues but it is conceptually simple and eliminates the spoiler effect. And it's already in use in parts of the US.
It feels like pushing for this (and for effective government structures in general) at the state level is the main backstop we have.
Unless there happens to be some helicopter pilot that ends up taking his oath to defend the United States seriously, or some other exceptional event or sea change that stops the neofascists in their tracks, it seems a foregone conclusion that the federal government will have been thoroughly destroyed in two years.
Ranked pairs seem like it would be cumbersome for "the average voter" to fill out on voting day if the number of candidates exceeds about four. And other methods like Schulze which do simple ranking but complicated math on the calculation are too math-y for "the average voter".
I'm open to ideas but I think in this country we need simple simple simple, while getting rid of the spoiler effect.
---
EDIT: Since I have been throttled by the "You're posting too fast, thanks" passive aggressive message, I'll respond here.
I thought ranked pair method involved actually going through each pair and doing a 1v1 vote for every combination. If it is the same "input" as RCV, meaning a single list of preference, then I'm really interested.
Systems that require centralized counting to calculate a winner are much easier to rig and much harder to trust.