# Market Utilitarianism

(~1900 words)

Abstract

Two classic versions of utilitarianism are average and total utilitarianism. They are classic, but they have well-known problems. Both have rather simple formulations. They begin with a reckoning of utilities across the population of individuals, and contend with a simple, linear aggregation of those utilities: average utilitarianism takes the mean of utilities and total utilitarianism takes the sum. I propose an intermediate, nonlinear version of utilitarianism, predicated on the local population of utility-experiencers in their abstract statistical space(s). When optimized, this formulation of utilitarianism recapitulates certain properties of a marketplace. I will show how “market utilitarianism” resolves issues with both average and total utilitarianism, though it introduces issues of its own, and I will consider some theoretical ramifications it leads to.

1. Definition

Given a number of individuals experiencing utility, average and total utilitarianism can be given simple mathematical expressions:

Total utilitarianism:

(1) $U_T = \sum_i u_i$

Average utilitarianism:

(2) $U_A = \frac{\sum_i u_i}{N} = \sum_i \frac{u_i}{N}$

Where $U_T$ is aggregate total utility, $u_i$ is the intrinsic utility experienced by individual $i$, $U_A$ is aggregate average utility, and $N$ is the total number of individuals, and $\sum_i$ is of course the summation for all individuals of the subsequent expression.

In market utilitarianism, the contribution of an individual’s utility to aggregate utility is attenuated by the existence of similar individuals, in proportion to the quantity of similar individuals. Market utilitarianism creates a rift between intrinsic utility as experienced by the individual and extrinsic utility as recognized by the aggregator, a quantity which relates inversely to the abundance of similar individuals for any individual. Market utilitarianism can thus be expressed:

(3) $U_M = \sum_i E(u_i)$

Where $U_M$ is aggregate market utility and $E(u_i)$ is the utility of individual $i$ as possibly attenuated by others extrinsically. $E(u_i) \simeq u_i$ when individual $i$ has no similar matches and $E(u_i) \simeq 0$ when $i$ has indefinitely many. More precisely:

(4) $E(u_i) = \frac{u_i}{L_{S, P}(u_i)}$

Where $L_{S, P}(u_i)$ is the local population of individuals in the statistical space $S$ around individual $i$, as defined by some nearest neighbor or local population parameterization $P$. $L_{S, P}(u_i)$ can never be less than $1$ because an individual is always near itself, and it can never be more than the total sum of individuals, $N$, identified previously. Putting it all together:

(5) $U_M = \sum_i E(u_i) = \sum_i \frac{u_i}{L_{S, P}(u_i)} \quad 1 \leq {L_{S, P}(u_i)} \leq N$

Thus market utilitarianism is intermediate between total and average:

(6) $U_T \geq U_M \geq U_A$ because $\sum_i \frac{u_i}{1} \geq \sum_i{\frac{u_i}{L_{S, P}(u_i)}} \geq \sum_i{\frac{u_i}{N}}$

The denominator $L_{S, P}(u_i)$  of the extrinsic function depends on both the choice of statistical space (one example for $S$ could be Blau space) and the choice of local population parameterization (one example for $P$ could be a count of individuals falling within a similarity hypersphere centered at $i$ in $S$ with some radius $r$, with $r = 0$ equivalent to total utilitarianism [given no two things being exactly identical] and $r = \infty$ equivalent to average utilitarianism). $S$ is somewhat arbitrary and often a subspace of a richer space, but $P$ should attribute monotonically decreasing importance to less similar individuals (and which individuals are similar or dissimilar obviously depends on the choice of $S$).

2. Behavior of the optimum

What is important about aggregate utility is the optimum that it achieves under the range of possible conditions. These conditions relate both to the set of individuals in a population of utility-experiencers and to the intrinsic utilities that each individual in the population experiences. This optimum prescribes the appropriate behavior—a choice or policy—assuming a prescriptive understanding of utilitarianism.

Let’s assume until further notice that the population is fixed at some size with some specific list of individuals. Then total and average utilitarianism will always experience an optimum of aggregate utility together. The reason is that total and average utilitarianism both aggregate all individual intrinsic utilities linearly (and without additive inversing, i.e. multiplication by $-1$, which is a linear operation but one that flips the maxima and minima). Thus, total and average utilitarianism always prescribe the same behavior (again: given a fixed population).

Market utilitarianism aggregates non-linearly, however, so the optimum of its aggregate utility will not necessarily co-occur with that of total and average utilitarianism. In particular, the same amount of intrinsic utility will contribute more to aggregate utility if spread among individuals in a sparsely populated region of the statistical space, such that relative to the optimum for total and average utilitarianism, intrinsic utility can be sacrificed among common individuals to provide it to more unique individuals. The conclusion here is that even though aggregate market utility is intermediate between aggregate total or aggregate average utility, its optimum is less similar to either of those than they are to each other (again: given a fixed population).

Let’s remove the assumption of a fixed population. This is where total and average utilitarianism both break, leading to absurd prescriptions for behavior. If the population can be adjusted, then the optimum for total utility occurs when every last individual exists who experiences net positive utility, even if barely non-miserable, and the optimum for average utility occurs when no individual exists except the one experiencing the most utility. This result is discussed in the literature on the mere addition paradox.

The optimum for market utilitarianism on the other hand is influenced by an important statistical fact: the more individuals that exist, the more likely any individual will have similar matches (and this is true for any $S$ and $P$). Thus, aggregate market utility does not increase past a certain point of diversity saturation (dependent on the footprint of $P$) because any new individual added is statistically likely to be similar to any other. Conversely, the optimum for aggregate market utility occurs at more individuals than the optimum for aggregate average utility, because any additional individual is statistically likely to be unique in that regime.

3. The absurdity of market utilitarianism, with a caveat

In market utilitarianism, aggregate utility can be changed without changing the number of utility-experiencers nor any of their experienced utilities. It can be increased merely by making the experiencers different from one another. This seems intuitively like an absurd result. It seems like individuals should be treated directly equally and not indirectly through terms regarding who else they’re similar to.

Note, however, an extremely keen analogy with the behavior of the job market. The job market apportions utility to individuals in the form of monetary compensation. This compensation depends on their ability to perform a set of duties, but also on the abundance of other individuals that can perform the same duties, i.e. similar individuals, analogously to market utilitarianism. Job market compensation can be fit to the model of market utilitarianism by finding the appropriately parameterized space of skillsets and other job-related characteristics $S$ (with an unfitted, out-of-the-box $P$), that is shaped such that an equal distribution of intrinsic utilities across individuals yields larger extrinsic utilities in sparsely populated regions as densely populated regions.

With market utilitarianism as a model, the job market can thus be understood to “see” human variation in a very specific way. Humans can in turn see with similar eyes in their choices to increase their own utility/compensation, and they will behave in such ways as to migrate the population from dense regions to sparse regions in the relevant space. If market utilitarianism is absurd, then the job market deserves intense scrutiny, along with the economic system built upon it. In the other direction, if the extant economic system is not held to be absurd, then market utilitarianism shouldn’t be considered particularly absurd either.

4. Utility-experiencer space vs experience space

Individuals experience utility, and individuals vary in their experiences of utility, with varying degrees of similarity between pairs of individuals. This fact presents a conundrum to market utilitarianism; if the space for individual variation chosen is the one-dimensional space of their experienced utility, absurd consequences quickly follow. Utility is but one measure of an experience, closely related to if not synonymous with valence. That and the other measures of experiences, all of them ostensibly qualia, together construct a space of experiences that utility-experiencers populate that is separate from the space that captures their “external” characteristics, even if they are highly correlated in some or perhaps even most domains (such as with the frequencies of incident light on the retina and perceived color or biological sex and felt gender).

Market utility can be computed in both utility-experiencer space or experience space. The latter might make more sense with the utility/valence dimension removed, but that still leaves behind rich structure. The correlations between the two spaces entail that a lot of the activity in one will be reflected in the other. Again, these are but two of endless possibilities for the choice of space, but they highlight some odd properties that market utilitarianism can exhibit.

5. Towards a reverse-engineering of the Universe’s actual objective function

Given the history of the Universe as data, there are many quantities predicated on this data or specific subsets of it that have increased or decreased mostly monotonically over time. Some are well-established constructs like entropy or Gibbs free energy or various other thermodynamic permutations. I’m interested in the realm of high-level compositions where utility-experiencers live, and so I ask what constructs predicated on that subset of data—the subset referencing high-level compositions—are actually being maximized in the Universe? Note that this question is emphatically not about quantities that subsystems may be maximizing each on their own, such as biological species maximizing their fitness, but about the aggregate. Are these constructs generally aligned with each other or totally scattered (especially modulo Occam’s Razor)? Is there any sense in which a moral compass of the Universe’s own can be detected on the basis of the constructs and quantities it maximizes?

Finally, I’d like to reflect on the role of utility/valence as a significant player in the development of compositions. As I have noted elsewhere, there are more levels of compositionality in the biological and cultural ecologies on Earth than have been observed in the rest of the Universe as a whole. Utility/valence has been organized into the psychologies of some life forms in a specific way, most importantly in the process of individuation and the separations between utility-experiencers. The properties of that space of experiences (and the distributions within it which can only be defined after a process of individuation)—along with its associations and correlations with the space of externals—drive much of the evolution of compositions, and essentially all of it within the human economy. Humans have the additional ability to share their experiences either by communication or faithful and intentional re-creations. If such experiences would be traded on a market, then the full exploration of the space of experiences would be incentivized. How would such an economic arrangement, especially if widespread, align with the Universe’s high-level compositional “moral compass”?

## 3 thoughts on “Market Utilitarianism”

1. Thanks for this, really found this interesting. On point 3, I would make an important distinction between the job market aggregations and market utilitarianism. Its an interesting parallel to draw but whereas market utilitarianism imposes values based on statistical sparsing which is uni-dimensional; job markets are two-dimensional with the ‘value’ being determined by both demand and supply. The supply mimics the market utilitarianism in a sense but demand does not – you may have a very unique ability to build cd-roms but that is no longer in demand. You would have strong relevance in the market utilitarianism if such an ability also meant you had a unique utility with cd-roms, BUT not in the job market. I may have misunderstood this though. Thanks!

Like

• Thank you kapuras! You’re right that the economic calculus has a two-pronged approach that market utilitarianism doesn’t have, and it might be interesting to try to recapitulate that in a further iteration of MU. Still, I think your point can be subsumed under the ability to choose a space. You can simply choose a space that does not have a cd-rom building ability as a dimension. Thus anyone who is “far” from others along that dimension would have their distance from them projected down to naught. That’s what I was intending with “the job market can thus be understood to ‘see’ human variation in a very specific way,” and it wouldn’t see the variation of cd-rom builders.

Like

2. […] According to most utilitarian calculi these two scenarios ought to be perfectly equivalent. In both cases the total amount of positive and negative qualia is the same (the full duration of the meal) and the only difference is that the latter also contains a large amount of neutral experience too. Whether classical or negative, utilitarians should consider these experiences equivalent since they contain the same amount of pleasure and pain (note: some other ethical frameworks do distinguish between these cases, such as average and market utilitarianism). […]

Like