My objection to cost/benefit analysis arguments is that the people making them usually fail to understand that there are FOUR, not two, elements that are part of the calculation:
- The probability of a specific set of benefits
- The probability of a specific set of costs
- The group that will receive the benefits or suffer the costs
- The time frame during which the costs will be suffered or the benefits received
The addict who is deciding whether or not to rob the convenience store to get money to buy heroin is not considering the costs to third parties and he is not considering the costs to himself beyond the next 24 hours. His group is too small and his time frame is too short.
If he included the costs to the clerk/bystanders whom he may shoot or the costs to himself over the next 12 months the result of his cost/benefit analysis would be very different.
If the individual uses a longer time frame over which the costs may be incurred he must recognize a higher level of costs; if he considers a wider group of people who may suffer costs the equation changes; if he will be performing a series of similar events he must use a higher probability of suffering those costs.
And, as you say, if one of the costs is total ruin then even a small probability of suffering it is far too much, to put it a little differently, a small chance of a fatal cost is actually a very large cost, not a small one.