"If someone eats unrefrigerated pizza ten times without any Adverse Events, then in fact that's an experiment providing (admittedly crude) scientific data that, in fact, the chance of an Adverse Event is no worse than ~10%."
Is that a statistically valid sample?
Let's say the "chance of an Adverse Event" is 0.000001%; literally 1 in a million. There are about 330,000,000 people in the US, all eating leftover, unrefrigerated pizza every morning. That means roughly 330 people get sick every day. If I've got the numbers right, ~0.3% of those who get sick are hospitalized. That is about 0.88/day, or ~6/week. 2.3% of those die, or about 7/year. Sounds perfectly acceptable to me (although those numbers are based on the current USDA regime, not the "try it a few times and if you don't die, you're good to go" approach).
"quick, dirty, wrong calculation because the exact value doesn't matter here"
Is that a statistically valid sample?
Let's say the "chance of an Adverse Event" is 0.000001%; literally 1 in a million. There are about 330,000,000 people in the US, all eating leftover, unrefrigerated pizza every morning. That means roughly 330 people get sick every day. If I've got the numbers right, ~0.3% of those who get sick are hospitalized. That is about 0.88/day, or ~6/week. 2.3% of those die, or about 7/year. Sounds perfectly acceptable to me (although those numbers are based on the current USDA regime, not the "try it a few times and if you don't die, you're good to go" approach).
"quick, dirty, wrong calculation because the exact value doesn't matter here"
It kinda does.