In act utilitarianism you choose the action that most increases utility, there is no ambiguity (assuming utility can be calculated, etc.). Conversely, in rule utilitarianism rules are chosen that generally increase utility, even if they sometimes don't.

So a rule utilitarian might have the rule

never kill

But then they might think, well what about if that person is going to kill you

never kill, except in self defense

What about someone else

never kill, except to prevent killing

What if the person they are killing is planning to murder hundreds of people?

never kill, except to prevent the killing of someone who is not going to kill anyone else

And you can keep going like this ad infinitum until you get back to where you started — act utilitarianism. Obviously that defeats the point, so a rule utilitarian would not do that. But it is also clear that they would not pick the most general rule, such as

never cause harm

since this is both impossible and fails in a large number of cases to increase utility. Where can a rule utilitarian draw the line without it being an arbitrary choice, while also not being at either extreme?

  • 2
    Act utilitarianism has nothing but ambiguity: how does one calculate what is "best" without a crystal ball, even assuming that they can come up with "utility" which is less vague than "well-being"? At least, a wide range of possible utilities and circumstances can presumably endorse the same rule. And the rule does not need to anticipate all the caveats, exigent circumstances are assumed in any practical rule by default. They neither can nor need to be anticipated, the decisions are always subject to judgment calls in specific contexts, but rules give somewhat more guidance than "utility".
    – Conifold
    Feb 20, 2019 at 19:30
  • @Conifold A few things to unpack here 1) I attempted to addressed the ambiguity of act with "assuming we can calculate utility" 2) The problem of ambiguity is more an issue of difficulty to calculate than actual ambiguity, there is presuppose to be a "right" answer. 3) "rule does not need to anticipate all caveats" My question is essentially about this, how many caveats accounted for and why 4) "decisions are always subject to judgement calls" that sounds in direct opposition to rule utilitarianism, since it is putting on the spot calculation above the rule
    – rtpax
    Feb 20, 2019 at 19:58
  • "Assuming we can calculate utility" is rather strange since the switch from act to rule utilitarianism is largely motivated by the obvious fact that we can't. There is no point to "accounting for caveats", one can insert explicit ones for more guidance if they are likely to be beneficial on average, and leave it at that. Again, "on the spot calculations" are in most cases a fantasy, hence the rules, and judgment calls are needed even to begin any "calculations" in the first place.
    – Conifold
    Feb 20, 2019 at 20:15
  • @Conifold, I created a chat at chat.stackexchange.com/rooms/info/89996 since this is getting kind of hard to do in comments, I would appreciate being able to continue discussion there
    – rtpax
    Feb 20, 2019 at 20:31
  • 1
    Do rule utilitarians all follow the same set of rules? Or is it determined on an individual basis? Feb 21, 2019 at 4:33

2 Answers 2


Yes, there is, but to see how and why we need a bit of exposition. Bear with me...


RU, in its more subtle versions designed to avoid "collapse" into AU, construes right acts as acts permitted by a set of "ideal" rules the general acceptance of which by most, but not all, people in the relevant community would produce at least as much utility as comparable acceptance of other moral codes. The rules are taken to be public, teachable, socially reinforced rules that should be internalized by all. Such rules may include "built-in" standard exceptions, but they would not be useful if they were too complex. The rules in the "ideal" set or code may sometimes conflict, but allowing too many conflicts among the rules would diminish their usefulness as rules. RU can offer various procedures for resolving these conflicts, but to stipulate that we should resolve conflicts simply by attempting to maximize utility in each particular instance of conflict would threaten to undermine what is distinctive about RU ... (Thomas E. Hill, Jr., 'Assessing Moral Rules: Utilitarian and Kantian Perspectives', Philosophical Issues, Vol. 15, Normativity (2005), pp. 158-178: 162.)

The crucial point distinctive of RU [rule utilitarianism] is that it implies that particular acts in accord with a useful rule can be right even though an alternative act [sanctioned by act utilitarianism] would actually produce more utility - and particular acts contrary to useful rules can be wrong even though they would actually produce more utility than any alternative would. This feature, and the structure of RU, enables rule-utilitarians to argue that the ideal moral code includes many intuitively important moral rules regarding honesty, justice, gratitude, loyalty, etc, that demand or forbid acts independently of the utility they produce on the particular occasion. RU says it is right, not just a generally useful rule of thumb, to conform to rules provided general acceptance of those rules by most people in the relevant community produces at least as much utility as such general acceptance of any alternative moral code.

Sophisticated versions of RU take into account that blindly following rules ideal for a non-actual world might be disastrous (and morally wrong) to follow in our imperfect, real world. Various strategies help to address the problem: for example, in assessing the utility of general conformity to a given set of rules we must take into account the feasibility and costs of maintaining it as a moral code as well as legitimate expectations generated under the actual moral code to be reformed ... (Hill: 163.)

First charge of arbitrariness

This is the charge that RU conflicts indefensibly with ordinary moral thinking. It:

entails judgments of right and wrong that conflict with judgments that are widely and deeply held by most competent, reflective people. Just as AU (act utilitarianism) is said to recommend acts that are unjust, ungrateful, breaches of promise, and insensitive to moral deserts, it is charged that RU recommends moral codes, or sets of rules, that are not adequately sensitive to claims of justice gratitude, fidelity, and desert. The reason is that, although rules requiring justice, fidelity, gratitude, and desert obviously promote well-being in many ways, the moral code that maximizes (expected) utility (e.g., aggregate well being) does not necessarily or obviously accord with our convictions about these matters. Given the complexities of the relevant empirical facts, the utility maximizing code would apparently allow many more exceptions than our reflective moral convictions would.

The problem with this line of objection for present purposes is that we are not in a position to know exactly what a utility maximizing code would prescribe regarding matters involving justice, fidelity, gratitude, and desert. We cannot argue that its recommendations are inferior to those of an alternative normative theory without knowing what the specific recommendations of each theory would turn out to be. Fictional counter-examples that simply stipulate what maximizes utility do not show that RU yields unacceptable results in the real world for which, presumably, normative ethical theories are intended. Actual or realistic counter-examples, however, are always controversial because it is uncertain what the long-term consequences of various complex sets of rules would be. (Hill: 164.)

I am not sure that mere departure from ordinary moral thinking makes RU arbitrary but the key point is that at the current level of discussion RU is schematic. We don't know what rules it would licence, a fortiori we do not know if and to what extent they would depart from ordinary moral thinking.

Second charge of arbitrariness

This is the objection which your question poses. It is that RU rules become so overloaded with qualifications that the situations to which its rules relate are so specific that RU becomes de facto indistinguishable from AU. But you omit a key point. There is every reason why RU should not incorporate densely complex qualifications. Remember a consideration from above :

The rules are taken to be public, teachable, socially reinforced rules that should be internalized by all. Such rules may include "built-in" standard exceptions, but they would not be useful if they were too complex.

The ever-more qualified rules you conjecture would be too complex to guide practical reasoning. AU aims at total finesse, at the exactly best outcome in every individual situation. RU lowers its sights and in my view becomes more reasonable and realistic in doing so. The aim is to formulate rules the general acceptance of which by most people in the relevant community produces at least as much utility as such general acceptance of any alternative moral code: and the calculation of utility can from the nature of the case be only approximate given our fallible judgement and imperfect information, the very limitations which make AU impracticable.

Endnote on AU and rules

Some act-utilitarians deny that the AU principle should be used as a working guide to moral decision-making. Instead, they treat it as stating what would make agents' acts objectively right even if the agents could not know what is objectively right, should not even try to discern it by estimating utility in the particular case, and are often utterly blameless in failing to do it. Such utilitarians can consistently recommend following rules the general acceptance of which would apparently maximize expected utility, but they can only recommend this as a useful heuristic guide likely to result in right action more often than trying to estimate utilities in particular cases. Sometimes, AU must concede, following such rules may lead to wrong acts because what maximizes utility (and so by AU is objectively right) in the particular case is not always what is prescribed by rules the general acceptance of which would maximize utility. (Hill: 162.)

So AU can find a place for rules and allow that in certain circumstances it is best to follow them. But it cannot argue that following rules all or most of the time is the best way to go about things - otherwise it loses its distinctiveness and collapses into RU.


Thomas E. Hill, Jr., 'Assessing Moral Rules: Utilitarian and Kantian Perspectives', Philosophical Issues, Vol. 15, Normativity (2005), pp. 158-178.

  • This is an interesting formulation of RU that solves the problem nicely. I am a little concerned about it's reliance on being taught universally, though I suppose that is a different discussion entirely. Do you happen to know if this is a commonly accepted formulation among contemporary rule utilitarians?
    – rtpax
    Mar 5, 2019 at 20:54
  • @rtpax. Thanks. I don't think that 'universally' can be taken absolutely strictly. My understanding is that it indicates that rules should be taught and inculcated within a moral community. Unless the generality of people know (and follow) the rules, the benefits of RU can't be gained. I doubt if one and the same set of rules would produce the best results across all times and communities. Sorry if I caused any unnecessary misunderstanding. Best - Geoffrey
    – Geoffrey Thomas
    Mar 6, 2019 at 9:34
  • @David Blomstrom. Sorry for the delay in replying. I'm not clear - straight admission, not a criticism - what you intend by 'that balances RU and AU'. I think rules for the taking of another human life would have to be numerous if they were to cover the different types of situation in which such killing would have the best consequences. In fact I don't think the task of enumerating and specifying the rules for best-consequence killing (if one allows such a thing) could possibly be codified. The variety of situations needing to be differentiated would defeat the most assiduous casuist. Best:GLT
    – Geoffrey Thomas
    Mar 7, 2019 at 11:54

A number of people have referred to my personal flavor of act utilitarianism as rule utilitarianism, so I will say how I choose rules under that system (though I disagree with the equivocation). Note that I these "rules" are only guidelines, and do not carry moral weight in and of themselves.

Rules here are essentially guidelines that a person creates for themselves (possibly based on rules of others) that they have determined to have a positive impact on utility. A person ought to constantly update these rules to have the best impact, bearing in mind that if a rule is too complicated they will be unable to follow it quickly enough, decreasing it's utility. Since you are comparing the expected utility, there is theoretically an unambiguous choice between two options -- though it may be difficult to calculate. A person need not follow these rules if they have the time to make the calculations more precisely. Because some rules are more intuitive to some than others, which rules to choose will vary from person to person, especially since any given person would be choosing from the subset of rules which they have thought of.

In short: when choosing between rules, pick the rule that has the greatest expected utility when followed, accounting for the fact that more complicated rules are harder to follow and thus decrease expected utility.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .