Place your bets -- Epsilon redux

Tamino asks:

What’s the chance that if we continue with business-as-usual, man-made global warming will lead to disastrous climate change? It isn’t zero. It isn’t one. What is epsilon?

My estimate (with links added):

Even if benefits were somehow remotely comparable to costs (2% chance of that, maybe?), benefits will be concentrated among the rich, while harms will be concentrated among the poorest and most vulnerable. So disaster is highly likely.

Enough disruption and direct harm to reverse the overall trend to greater wealth and prosperity — 10% — 70% (highly uncertain) — say 40% over the next few centuries.

Collapse of civilization, barbarism, going a long way backwards in terms of knowledge and human adaptive capacity even as impacts continue for centuries — 5%.

About the first estimate I (and most economists who have studied the issue) are very confident, and it’s easy to make the case for mitigation based on that alone. The other estimates are highly uncertain, and many analyses ignore them, foolishly in my view.

Add your own estimates at Tamino's and/or over here. He forbids much supporting argumentation, to avoid distractions, but not being cursed with his degree of popularity, I think we can differentiate ourselves by welcoming rationales and arguments for the numbers.