An intriguing definition of Morality

Question

The definition of morality is a topic that is heavily debated in philosophy. In this post, I will provide a possible definition of morality and will try to argue why the definition is generalized and consistent. It would be appreciated if someone can provide some insights into the intended argument.

Rather than defining morality, I would define a "moral set" which is a little more mathematically rigorous than the abstract concept of morality. The definition goes as follows:

Definition

For a defined Goal, the set of instructions/laws that maximizes the probability of attaining that Goal is called the moral set.

Application

Let us say that the Goal is the happiness of humankind. We can see that much of the actions that are considered "moral" in the normal sense can be categorized in the moral set under this definition. For example, we can consider telling truth moral. We see that truth-telling (with some exceptions) guarantees a greater probability of success than utilizing falsehood since falsehood can be extrapolated to many more evils. Hence, it suffices to state that truth-telling can be categorized under the moral set. (There is a minor problem of behavior after a long time in this argument. For example, it is surely possible that even though falsehood may result in a decreased happiness for the time being, but after a finite time, the happiness due to a moral set containing truth-telling laws may be approximately equal to a moral set containing falsehood laws. Hence we would have no preference for truth-telling in this situation. However, this is accounted in the modified definition given below.)

A subset of examples warrants a little modification to this definition. If our Goal has a measure, then there can exist a Law A and Law B where both maximizes the probability of attaining the Goal but Law B also guarantees that Goal, it attains, has a higher measure than Law A.

Assume that the Goal is maximizing mathematical knowledge. Law A advocates for symbology that is very hard to learn and use. Law B, however, advocates using symbology which is very easy to use and master. We see that both Law A and Law B, in terms of maximizing the probability of attaining the goal, are both virtually indistinguishable. But it is undeniable that we should go with Law B since it makes the whole process easier.

Hence a modified definition is proposed

Definition

For a defined Goal, the set of instructions/laws that maximizes the probability of attaining that Goal and optimizes the Goal itself is called the moral set.

First off, this definition makes morality subjective (over the domain of possible Goals). Moreover, an advantage of using this definition of morality is it reduces moral dilemmas to problems of optimization and statistics.

The above arguments that were given in support were in no way a rigorous proof of the definition. All of them were intended for shedding some light on the motivation behind the definition.

It would be appreciated if someone were to scrutinize the above definition and provide some reference (if the definition is already adopted in the philosophical literature) or provide some insight (that can be used to polish the argument or understand why the definition is purely wrong).

Edit:

On the subject of choice of Goal:

The problem mainly comes with choosing the Goal. I have left it for further analysis at this point, and deciding the optimal choice of Goal (if necessary) is the reason I had to post this question.

It is entirely possible that the Goal can be chosen something very ridiculous like building Lego structures or, plainly to contradict the main intention, immorality. However, originally it was hypothesized, mainly due to the nature of humanity, that the Goal would be chosen rationally. If the rational consensus of humanity agrees that we should be striving for immorality (and consequently define the Goal to be some immoral aspect), then I think then that will define the construction of the moral set. However, I would disagree that a rational consensus would result in such a poor choice of the Goal. (This is entirely subjective at this moment).

Answer 1

In a pragmatic sense, a definition similar to the one given may be relevant to artificial intelligence. In fact, your definition closely corresponds to a class of machine learning algorithms. One could define an objective function corresponding to probability (or quality) of the goal, and use search, optimization or learning techniques to maximize it. Such are the subject of active research.

Further, if we assume that the moral set is the set of permissible moral actions, your first definition would be a sort of probabilistic deontic logic. Most studied in philosophy are standard deontic logics, which adapt Kripkean semantics for alethic modal logic. If you're interested in additional background, details and history on standard deontic logics, you may want to see the SEP article. However, there have been potential issues with such axiomatizations.

Independent of the precise logic though, such a definition implies a type of consequentialism with respect to a good corresponding to the probability of attaining (and quality of, for the second definition) the given goal (although usually consequentialism is presumed to refer to hedonistic consequentialism or utilitarianism - perhaps corresponding to your sample goal of the happiness of humans).