Notionally, opening new banks and/or buying an existing one that lacks current customers is feasible. Presumably once you do that, you could easily plug into FedWire/FedACH/FedNow, which would allow you to process electronic payments on behalf of whomever you choose to bank, as long as you comply with all the KYC/AML and other finance regs. You might not be able to process visa and mastercard, but virtually any of your customers' customers presumably still have financial accounts that are enabled on Fed rails, which will not be as likely to discriminate. Requiring end-customers to initiate payment (push) as opposed to doing pull, would also reduce transaction reversal risk.

I don't really the ultimate purpose of the position taken by either side here, but it does seem to me that there's a path by which both can coexist, even if difficult.

I don't think blockchains are the way forward, but the crypto people had one thing right: The incentive towards providing storage/compute/connectivity for transaction settlement should be built into the protocol, not provided by some corruptible institution. The idea that coins would be mined by the people providing the compute has unacceptable consequences, but it did solve the problem that it set out to solve. So we need something like that, but with fewer problematic side effects.

If I were to take a stab such a protocol, it would live on a web of trust that kept track of interpersonal debts. So if Alice owes Bob, and Bob owes Charlie, then Charlie can "pay" Alice by instructing Bob to cancel both of those debts. In this case Bob is the clearinghouse. Some network connectivity is needed (Alice, Bob, and Charlie all on the same network) but it still works if the internet partitions. We incentivize Bob to keep his device available for this by creating a new debt when the transaction settles: this time it's one where Alice and Charlie owe Bob for having been the transaction processor.

If Bob didn't want to act in this capacity, he wouldn't have trusted Alice and Charlie to begin with. Charlie would then need to find a different path to Alice if he wanted to use the network to pay her. (Presumably they don't trust each other directly, otherwise they'd be creating new debts instead of cancelling transitive ones). In this scenario I suppose Bob would be acting as "censor" but his capacity to do so is not greater than any other user of the network.

For Charlie to be fully denied the ability to pay Alice, one of them would have to be behaving so badly that everyone refuses to trust them. That's a desirable outcome, if we must have something like censorship, it should not depend on the feeling of some guy who owns a bank but is otherwise unconnected with the parties of the transaction. It should be decided by the people who are near to Alice and Bob on the trust graph and therefore have to deal with the real world consequences of Alice and Bob's behavior. If they want to collectively prevent those two from doing business, that's their right since it's their community that that business is happening in.

One imagines this featuring in cases where some rich foreigner wants to set up mining operations that would poison the drinking water. The locals can collectively prevent that foreign money from buying local groceries, thereby limiting the ability of the foreign actor to harm them.

So that's my harebrained idea. But I'm not making up fictions and asking people to treat them like money. The people involved with USD are. If they want us to continue participating in their harebrained idea, they should recognize that giving us a system with problematic properties represents a risk that they might lose their privileged position. My point is just that it's on them, the designers of the system, to figure it out how to make it all work--we need only accept or reject it.