Morality is not math, but, as always, things might be more foggy: lack of free choice and a really skewed ratio (e.g. punish one to save 100 million) may test the limits of morality.
In that case it's clear that I won't pull the lever. It's not a question of life and death, and because I am not sure whether pulling the lever helps preventing lots of damage, I won't do anything, especially because perhaps it's just a vague and abstract story you made up.
Government builds a dam. It clearly recommends people to not build in front of the dam, it may crack and a flood would be a tragedy. 100K people ignore it and slowly houses appear: it's cheaper, better land, views, etc. 1K people choose to comply with Government and build in a safe but ugly and infertile area. One day, suddenly, a huge crack appears. At the last minute engineers discovered "a lever" to flood the safe area instead and save 99K lives. They don't have time to evacuate anybody. Should they pull the lever?
By limits of morality I mean:
- Free choice: if the people living in the unsafe area didn't understand the danger or didn't have another choice but to live there?
- Skewed ratio: if only 10 people lived in the safe area and 1 million had chosen to live in the unsafe area?
If you're enterpreneur, see whether 100K people is more beneficial than the other 1K. For example if you have a factory that needs many workforce and can utilize the 100K, save them.
If somehow you pull the lever, make a press conference showing sorrow and guilty face telling it's inevitable. If you don't pull the lever, don't expose the info until asked.
If you're not a decision maker, leave it to your supervisor. If somehow you have the loved one in 100K, pull the lever anyway.
Sadly this is the shape of our world right now.
"A new car built by my company leaves somewhere traveling at 60 mph. The rear differential locks up. The car crashes and burns with everyone trapped inside. Now, should we initiate a recall? Take the number of vehicles in the field, A, multiply by the probable rate of failure, B, multiply by the average out-of-court settlement, C. A times B times C equals X. If X is less than the cost of a recall, we don't do one."