AFAIK the pressure vessel was never breached at Fukushima. They had plenty of problems whose primary cause was the absence of cooling water (including a hydrogen gas explosion), but there was never an uncontrolled criticality because of it. (Also see further comments below on criticality.)

I am talking about short-term "everyone evacuate the building" types of disaster scenarios. You can do precisely that with a LFTR because of the passive decay heat cooling combined with the loss of criticality that happens when the freeze plug melts.

Ok, this makes it clearer where you are coming from. I certainly agree that the LFTR is a big improvement over the standard PWR design.

We don't really even know how bad the damage can get.

It's true that Chernobyl could have been worse, so we can't really judge the worst case from what happened there. However, these worst-case scenarios have been simulated in great detail; the physics is actually pretty simple.

With any design that isn't Soviet-built, a loss of coolant will not cause an uncontrolled criticality; losing coolant decreases the reaction rate, causing loss of criticality. The problem with the Chernobyl design was that it had a "positive void coefficient of reactivity", meaning that the reaction rate increased on loss of coolant. That feature, as I said, is not present in any non-Soviet design, which means it's not present in any commercial design currently operating (since all the old Soviet reactors have been shut down).

I just REALLY don't like the idea of the balancing act that has to be performed in 1 and far prefer the kind of intrinsic safety that you get from 2.

So do I. Now that we have such designs, we should certainly be building them, and should not be building the old designs that lack those passive safety features.