"Expected value" is standard probability theory. Check it on Wikipedia. Expected value equals the value of an event times the probability of occurrence of that event. For example, if a deal has a 0.5 chance of yielding $1, the expected value of the deal is 0.5 x $1 = $0.50. It is standard decision theory to evaluate deals in this way. In the case of collider risk, the theories that permit trouble appear to be a small subset of the set of all theories, so I will concede that the probability of trouble is "low." However, many proposed safety factors have eroded, so the probability is not zero. Let us arbitrarily and subjectively stipulate that the risk is in the range of 1/100 to 1/10,000. That is unlikely to happen, nevertheless a passenger airplane with a 1/10,000 chance of crashing would not be allowed to fly. If one multiplies a probability of 1/10,000 times the lives that would be lost immediately (to say nothing of future lives) the negative expected value in lives is 6,500,000,000 (the population of Earth) x 1/10,000 = 650,000 lives. That is not trivial.
If you care about credentials, I have an MS in biometry and statistics.
A good paper on this type of thing is Adrian Kent, "A critical look at risk assessments for global catastrophes," Risk Analysis, Vol. 24, No. 1, 2004. See our references section for a link. That paper was inspired by the collider issue, albeit written before safety factors eroded.
James Blodgett, MS, MBA, MA