Theoretically, my criteria for an ideal world can be characterized by two pairs of numbers. Each pair includes a measure of decision-related increase in longevity and a measure of decision-related increase in happiness (satisfaction); with one pair dealing with the individual most impacted by the decision, and the other pair dealing with the most people affected by the decision.
Practically, deriving such numbers (especially the pair dealing with the species) to any reasonable amount of accuracy for even the simplest kind of decision is all but impossible. Such a feat requires a comprehensive model of the world with current and comprehensive information about every part of it, not to mention computing power that far exceeds anything we are likely to ever acquire. Even approximations demand an unwieldy set of assumptions that may not even be testable.
Fortunately, since my criteria involve the maximizing of these four numbers, the minimum information we need to know when making a decision is whether or not all four numbers are negative. That is, we should reject any choice where one or more of the numbers is negative (indicating a reduction in longevity or satisfaction with life). For comparison, in our present system, buying decisions are typically made based on whether the individual buyer will be satisfied or not, representing the sign of only one of these numbers.