Think net expected benefit, not just probability (Score 1)

So these may be crackpot theories. Let's put emotions aside and try to describe these risks in scientific terms. What is the chance of a supernova-sized Pac Man materializing in space tomorrow and swallowing the planets of the solar system in quick linear succession? Small, too small for the word "small" to be appropriate. But what if there's a God with an overly nerdy sense of humor? For all we know, the chance isn't quite 0. It's perhaps 1e-10000000, or 1-e1-e30. Hard to quantify for sure, but I'm pretty sure that as far as risk evaluation goes, I can put a lower bound on it that's greater than 0.

Clearly the same can be said about these theories. While the risks involved are greater than that of the Pac Man theory, they are probably smaller than, say, the risk that any of the scientists involved will die in a car accident tomorrow, and possibly smaller than the risk that they will ALL die in simultaneous, unrelated car accidents on the same minute of the same hour tomorrow. But even if they all did, in what is undoubtedly another statistical near-impossibility, in the end we would have lost a bunch of scientists, not the entire human race and the planet to boot.

This is why the expected value of the loss, not the probability alone, should be the relevant measure here. The expected loss is the probability of the loss times the value of the loss. There are very few ways we can endanger the entire planet in a catastrophically irreversible way. If this astoundingly improbable event should come to be, we'd lose the human race, the animal kingdom, and the Earth. While the value of all that is almost too large to ponder, it exists and it is finite.

The point here is that it's too easy to throw one's arms in the air and dismiss this risk assessment because the probability of the loss is inconceivably small and the value of the loss is inconceivably high. Very very small numbers multiplied by very very large numbers can yield very accessible and understandable numbers. As far as I know, the expected loss could be 1 cent, or $1 million trillion dollars. After reading TFA and a couple of related sites, I can plug some wild numerical guesses for these two factors and get either result.

Which is why we also need to look at the benefit half of the equation:

(net expected value of the LHC) = (probability of scientific discoveries) x (value of discoveries) - (probability of black hole swallowing earth) x (value of the earth)

The left side is easier to quantify, but not by much. I'm pretty sure that even the most optimistic scientist, the one most willing to bet the farm on this will come up with a number smaller than $1 million trillion dollars, so a risk evaluation is order. IMHO, it would be reasonable to spend as much as 10% or more of the total budget of the LHC in coming up with the most accurate possible risk assessment. Now intuitively, I'm sure the actual risk assessment budget was much smaller than that - orders of magnitude smaller.

Which brings me to the truly scary side of all this: the inevitable difference between the social expected value of the scientific discoveries ("social" meaning to the entire human race, which is betting the farm on all this no matter what the probabilities are) and their private value to the scientists involved, the scientific community at large, and the particular politicians and bureaucrats associated with the project. These are classic externalities in economics: the managers at the upstream factory that's polluting the river have to drink the polluted water as well, but the private benefit they derive from polluting outweighs their share of the social cost, so the decision to pollute brings a net private benefit to them. Now I don't mean to offend the scientific community with that example: scientists aren't evil and certainly aren't on the same moral ground as polluters. But they're certainly more gung-ho about scientific experiments and their benefits than laymans are, or in other words are generally predisposed to overvalue the left side of the equation.

I don't know how many times we've bet the "big farm" in the past. We certainly did in 1944-45 during the development of nuclear bombs. Perhaps we did when one of our caveman ancestors went deer hunting on a full moon against the warnings of his religion. But in the 40's the prizes were victory in the war and the development of nuclear energy (while in the caveman's case, prehistoric superstition and statistical risk assessments don't go hand in hand.) I don't think it would be unreasonable to force never-before-tried high-energy physics projects to invest 10% of their budget in developing the most accurate (and unbiased) possible risk assessment.