It isn't debunked, but not well modelled and thus very large error bars (from a friend that works in atmospheric chemistry)

The problem with that criticism is that the payoff function is absurdly nonlinear. Think of it in terms of expected values: because the outcome at one end of those error bars is so severe, it's rational to treat the "gun" part as true until it's shown not to be.

I'm not criticizing the hypothesis, simply pointing out it's not well understood.

But it is not true that the precautionary principle applies here regardless of the risk.

There is a cost to avoiding the behaviors that raise the possibility of run away methane release; carbon based energy was absolutely necessary to get out science/industry to the point we can look for alternatives.

If we muddled around in the pre-industrial age in perpetuity, eventually we'd get wiped out by some large event (astroid, super volcano, ice age etc). So the risk is worth it to a point since the cost of the alternative is equally high.

We're not in that situation, because we've already made the choice not to muddle about pre-industrially.

But even so, that argument is flawed: we're still at risk of getting wiped out by all the same extreme natural phenomena as we were 500 years ago, only we've added some new ones of our own: nuclear holocaust, clathrate guns, runaway warming of some other cause immediately spring to mind. Probabilistically we're worse off now than we were then, just because of the additional ways things can go wrong.

We don't have the choice to stay pre-industrial. That's off the table. We do hopefully still have the choice to avoid a scenario that drops us irrecoverably back to pre-agricultural.