Deontology as Pareto Optimization

Morality is divided broadly between deontologists and utilitarians. Some characterization of the two and their differences are along the following lines. Deontology broadly considers some things just right and others just wrong, the founding characterization is the Categorical Imperative which Kant formulated a few way for our purposed here is his second:

"Act in such a way that you treat humanity, whether in your own person or in the person of any other, never merely as a means to an end, but always at the same time as an end."

Utilitarianism considers things good or bad in different degrees and that what matters is optimizing the total "utility" -- specified differently depending on who you ask.

In a highly, highly simplified views there are many moral philosophy like rule and preference utilitarianism as well as various versions of deontology and other ethical systems that don't fit neatly into either box. Even the domain over the Categorical imperative is not without controversy. And now that we are done with the disclaimers section.

You might also characterize deontology/deontological moral systems as optimizing in a similiar manner (a rather convoluted) other objective function where if you are doing something that is in violation of the moral law it is worth negative some very large amount of utility and if not it is worth say 0 utility. i.e. specifying it in a way to fit the moral dictates of an arbitrary value system usually one that is considered non-utilitarian.

Part of the issue is it does not really add anything or parsimoniously describe the system it is not then "natively" utilitarian. It is a hack. But what if we approach it from a different tack; instead of changing the definition of the objective we change the optimizer?

Maximization is only one of large varieties of ways of specifying an optimization problem. We needn't have one or even a fixed set of objectives, they need not be the same toe they need not be prioritized in exactly the same way relative to each other (indeed if they were it would generally reduce to simple maximization). We can have hard constraints we can have things be conditionally weighted. Optimization is a big field for a reason!

Here specifically of concern though is Pareto Optimization, where you have a set of objective functions which you wish to maximize however changes that do not decrease any of the functions can be considered. If we take the individual functions to be utility functions for individuals, this interprets to:

"Take the actions which maximize the utility of individuals subject to those actions not impairing the utility of others."

Which to me sounds sort of like a synthesis, let's take another stab at interpretation and make a substitution of utility for end and associated grammatical adjustments and supposing the ends you personally try to pursue are your own (perhaps rather tautologically):

"Act towards your ends subject to those actions not impairing the ends of others."

Which at that point to me bares quite the resemblance to the second formulation of the categorical imperative as mentioned above:

"Act in such a way that you treat humanity, whether in your own person or in the person of any other, never merely as a means to an end, but always at the same time as an end."

To me switching the optimization framework creates a clearer path to formulating deontology in a broadly utilitarian framework than shoehorning a utility function to do the same job. Turning it back on itself you can also interpret usage of Pareto Optimization as encoding a deontological value system on the optimization process!