Stop Adding Zeroes

A sharp defense of EA/x-risk expected-value reasoning against Dylan Matthews' 'you're just making up the probabilities' critique. Core move: 'nothing is ever 10^-66 and you should never use that number', justified by the famous tornado-meteorite-terrorist-double-lottery-Trump chain (~10^-65) as a calibration anchor for how absurdly tiny such probabilities have to be. Adds the 80,000-Hours doctor-vs-earning-to-give example ('even your ass has some standards'), the 'staring into the darkness' passage on trusting weird math (hedge-fund-vs-doctor, praziquantel, chickens), and the Muggeridge train vision. A widely-cited statement of EA epistemics; keeps Matthews inside the tent. Excellent floor.

Quality 76 · Excellent. Excellent floor (76): a memorable, genuinely clarifying defense of EV reasoning and calibration of tiny probabilities, frequently cited; not higher because it's a targeted rebuttal to one critique rather than a standalone framework.

Claude’s paradigm shift 56 · Moderate. Major-shift floor (56): the 'you can't just add zeroes / sanity-check your made-up numbers' argument and the tornado-meteorite calibration device are a fresh, memorable contribution, building on the existing Pascal's-mugging and EA discourse.

Real-world impact 4 · Moderate. Moderate (4): 'staring into the darkness' and the never-use-10^-66 calibration argument circulate as touchstones within the EA/rationalist movement; discourse-level influence in an educated subculture.