Fast reactor inherently have more fissionable material in a smaller volume, and don't have a moderator that can either keep things separated or remove itself in an accident to kill the chain reaction. If a supercriticality on prompt neutrons alone occurs in a fast reactor the neutron population will increase orders of magnitude faster than in a thermal reactor, because the neutrons are moving so much faster.

Edward Teller in 1967 famously expressed his skepticism about fast reactors because of this safety issue.

That's not the argument I was making. I said anything of sufficient scale carries significant risk, and we can in time reduce that risk to acceptable levels via engineering practices and principles. As with airplanes, which are metal tubes full of humans we load up with kerosene and send into the atmosphere. A bad plane will crash, a bad reactor will melt down. We can overcome the risks to reap the rewards. That's like our whole thing, as humans.

Chernobyl happened for a very specific reason - you know, the late-night tests and the positive void coefficient. Do you think we haven't learned? Please cite sources. The thing is, we have learned. That's why reactors are so expensive. You're double-dipping by citing the risks and the costs. It's really one or the other if you're not being disingenuous.

It's not the argument you were making, but it's a point your argument is compelled to address. Yes, any system of sufficient scale carries risks. In the case of breeder reactors, one of those risks is the possibility of something approaching an actual nuclear explosion, with fast neutron physics similar to what happens in actual fission bombs, not pitiful low yield moderated explosions as happened in Chernobyl.