Consider two courses of action:
- One has a low probability of success but promises to mildly increase welfare (however defined). Call this “meliorism.” Rawlsian liberals, Burkean and Oakshottean conservatives, and Hayekian libertarians frequently identify with this view.
- Another has an unknown probability of success, but promises to massively increase welfare (however defined). Call this “perfectionism.” Marxists, anarchists, neo-conservatives, and cosmopolitans frequently identify with this view.
It seems to me that the perfectionist critique of meliorists frequently concerns the size of the benefit, while the meliorist’s response focuses on the probability of success.
But common-sense meliorists also sometimes offer the cliche that “the perfect is the enemy of the good.” By this, I think they intend an addendum to the perfectionist position:
- That perfectionism entails a greater probability of creating a massive decrease in welfare (however defined)
In support of this addendum, meliorists might offer various historical examples of populism-cum-tyranny and good intentions leading to tragedy.
But perfectionists can play the addendum game too. For the perfectionist, there is an unnoticed tendency in meliorists to accept moderate decreases in welfare over a disruption in the status quo. The addendum might be:
- That meliorism entails a greater probability of mild and continual decreases in welfare (however defined)
In support of this addendum, perfectionists can offer both historical and contemporary examples of institutional degradation, continued barbarism, and a general blindness to suffering by meliorists, as well as, in some cases, a sophisticated theory of causality and history to account for both the decline of meliorist institutions and the epistemic ignorance of the meliorists to the weaknesses in their position.
It seems, in this case, that both groups are engaged in basic probability and effect-size assessment. Thus, it would seem incumbent on both groups to begin by working on a definition of welfare and then on a consensus for the probabilities assigned to various consequences. And yet here perfectionists tend to reject both the project of defining success and the project of assigning likelihoods, retreating to a skepticism about metrics and about the evaluation of future probabilities. Frequently, the perfectionist bolsters these skepticisms with the sophisticated teleological or messianic view of history, or the error theory I mentioned earlier, now applied to available metrics which allegedly blind us to real depredations.
Does this sound right?