"we don't need empirical evidence to know that at some point interest payments will require highter tax revenues which will reduce disposable income and hence consumption."
We always need empirical evidence. For your example, no, we don't know that, e.g. when the debt is monetized by the central bank. A bias towards sovereign debt elimination may be more dangerous than a bias towards sovereign debt accumulation - empirical evidence, not intuition, is how we validate the models that guide our trade-offs.
We will never have empirical evidence (with any predictive power) for what really happens if the biggest economy in the world, the one that provides the reserve currency, the preeminent military power, goes bust, because the sample size is 1.
[Edit] And about your counter example: We do know that sovereign debt accumulation will eventually become impossible, so it's not relevant whether or not it would be beneficial. Once domestic lenders (who can be forced to lend) are exhausted, the government will have to borrow from foreigners in foreign currency under foreign law. So debt accumulation is only possible as long as foreigners are willing to lend.
A nit-pick, but also maybe a useful FYI, "sovereign debt" is different than public/national debt.
The US doesn't issue sovereign debt AFAIK, and you cannot just print more money to repay sovereign debt, that's the whole point of it. (Sovereign debt by definition is issued in a foreign currency.)
Correction: 74% of the American population carries net debt, even if it's "good debt". Inflation would act as a mini-jubilee to rebalance the economy between producers and financiers.
We always need empirical evidence. For your example, no, we don't know that, e.g. when the debt is monetized by the central bank. A bias towards sovereign debt elimination may be more dangerous than a bias towards sovereign debt accumulation - empirical evidence, not intuition, is how we validate the models that guide our trade-offs.