Utilitarians and lying

Utilitarian types tend to worship science, when probed as to why they worship science they will give an answer resembling ‘because it is the most reliable way to approach truth’. But of course, these are not truth maximizers, they are utilitarians… so even if it is the most reliable way to approach truth, that truth is only valued for the utility it brings.

Utilitarians are aware that their position implies that is just a means to an end. They react to this implication in one of two ways. The Sam Harri’s path, which is that truth is in alignment with welfare. Or you can take the position that it is sometimes okay to lie like Noah Smith claims. The second position is just biting the bullet, so let us focus on the Sam Harris path.

Utility and truth are always in alignment so you should always be truthful. This is a particularly heroic assumption, indeed, probably the clearest example of this is to imagine somebody on their deathbed asking for glimmers of hope. I doubt even utilitarians would deny these cases exist, but maybe they can just make a weird sub clause for people on their deathbeds. To avoid this clause, let us take a more relevant case: There is an unprecedented pandemic, if the doctors don’t get the masks while they deal with patients, two thousand more people will die, so you want to make sure that normal people don’t rush to buy masks to ensure the doctors can be supplied. It should be obvious that a utilitarian would lie! Heck, the nice thing about democratic systems is that ‘trust’ is usually person specific, If Fauci lies about it, he can be replaced with the next useful idiot who doesn’t have a record of lying and bam, credibility is reset. Utilitarians can try to argue that it isn’t about trust in a person but about trust in the ‘institution’ but I think that this critique is overblown, most people usually forgive the lies and hope the next person of influence isn’t going to also lie.

Utility is furthered more on average by truth-telling. This point of view relies on the agent not being able to distinguish between situations where truth helps utility and situations where it doesn’t. Indeed, this seems to be one of the advantages of science, science helps us find state variables where it is easier to predict the consequences of our actions. So if it is true that on average, you should tell the truth, then a corollary of this is that as our understanding of the situations improves, our lying should increase.

It is also interesting to understand what this does to the probability of being truthful as a function of influence. Take Harris’s main reason for telling the truth, that it allows agents to form a relationship (which increases welfare). Now it seems quite clear that when somebody of influence speaks, there are not proportionately as many relationships being formed. It would be rather absurd to claim that Mehmet the Conqueror formed a relationship with everyone who listened to him. Instead it seems clear that when Mehmet speaks, very few relationships are being formed but there is a unilateral effect on a lot of people. So the tradeoff for a person of influence, seems skewed towards lying.

So to sum it all up, Utilitarians are telling us they will lie more, the more their understanding increases and they will also lie more the more influential they become. As soon as they have a position of influence, they will in fact lie. This is but an implication of what it means to value truth for its effects on pain/pleasure.

Utilitarianism entails no duties

As mentioned last time, if we have a given situation, utilitarianism entails that there will be one right answer about the right thing to do, in other words, everything will be morally obligatory, which leaves no room for character, today I want to highlight an entailment of this problem.

The utilitarian often asks you to take decisions as if you are in a spaceship disconnected to anything. If for instance we are asked whether one should save one baby or ten babies, the utilitarian would say, save the ten. But there is no obvious way in which they revise their views should relationships emerge, for instance, what if you are the mother of the one baby?

Classical tradition would say that a mother has a duty to her child, and though this duty may not lead to a moral obligation to save her child, it at the very least makes saving her child, morally permissible. But a utilitarian does not care, the prescription of a utilitarian is about total utility, and hence no relationships can be used as a criterion.

A utilitarian may attempt to revive duties by using something called ‘rule utilitarianism’, which means follow the rules that maximize utility, but really this now shifts the morality to plauisibly rule utilitarianism out. The rule ‘we ought to accept the rule that maximizes utility’ may actually imply that we ought NOT to be rule utilitarians. But this actually gives the whole game away, when we are actually arguing about ethics, we are arguing about what rule set or source to use when making moral decisions. Saying that we should adopt the rules that give the best consequences is only SLIGHTLY refining the set of rules. Instead of being ‘let us pick the best ethical system’, it is now ‘let us pick the ethical system with the best consequences’ but this may get you to a more or less ANY ethical system. A deontologist may be a rule utilitarian!

Action only needs correlation

This is an example that came up in conversation which I think is quite clear. Suppose that you are a teacher at a school, and you can recognize children as either gypsies or non-gypsies.

Children periodically stop showing up at school. When they stopped showing up at school, you went to their houses to give them homework. Over time, you noticed that whenever you went to a gypsy house to give them homework, only 10% of those kids completed the homework, whilst whenever you went to a non-gypsy household the kids did complete it.

You know that the ultimate cause of the gypsy kids not completing their homework was NOT their ‘gypsyness’ but some other attribute, perhaps those with noisy neighbors did not complete their homework because it is too loud, and those without, could complete it. It so happens that gypsies have noisier neighbors than non-gypsies. However this information about neighbor noisiness is not available to you.

You do have a cost to bringing the homework to the students, there are other more valuable activities that could be done. You believe(perhaps based on some calculation) that it is only worth making the trip to give the kids homework, if there is at least a 50% chance they will complete it, which excludes the gypsies.

So as it turns out, you do NOT visit the gypsy kids to give them homework even though you KNOW that the gypsiness of the kids is NOT the reason they do not complete the homework. Nevertheless, the decision based only on correlation is a reasonable allocation of resources.

Utilitarianism: everything is obligatory

Other ethical systems may 1) give you rules on what not to do, or 2) may outline criteria after which an action becomes morally permissible, in contrast, utilitarians claim that everything they say is a moral obligation. If something maximizes utility, it is your obligation to do that, there is no room for disagreement or tolerance. If you are utilitarian, and somebody is not maximizing utility, they are being unethical. This means that it has no room for pluralism in ethical values, there is no room for character at all. In fact, the best moral agent to a utilitarian is a robot, any kind of subjectivity in calculating the right action can only lead us astray. Whilst classical ethics, in the form of virtue ethics, actually allow one to look within for the answer, utilitarianism focuses on a vague cardinality of future pleasures and pains, which are best quantified, and hence take the human out of the equation.

Since utilitarians are likely to acknowledge that you can do more to maximize total utility by focusing more on the utility of others than your own, they also in fact prescribe that everybody do the same thing. That is, if volunteering at charity X does more to help total utility, then everyone should volunteer to work at that one charity. Their only mechanism for generating diversity in actions is diversity in competencies, they may say, if you are a good investment banker, focus on doing that and then give your money to the most effective charity. The ideal utilitarian simply earns a lot of money, even by wronging people, and then simply gives that money to the most effective charity(since that charity specializes in maximizing utility).

So by definition, if Bob meets Anna, Anna is a utilitarian, and Bob agrees with Anna that his actions are not maximizing utility, this implies Anna thinks Bob is unethical. Of course in practice, utilitarians will try to hide this aspect of their view by simply saying that you are too ignorant so you are only acting under incomplete information. In sum, a utilitarian thinks non-utilitarians are either unethical, or idiots.

I do not think this requires any more elaboration but maybe for constrast let us simply compare this to a rule like the doctrine of double effect. The modern elaboration of the doctrine was made by Phillipa Foot, when she introduced the Trolley problem. Her formulation of the doctrine is still being debated but what I want to focus on here is that she has her four criteria, which I include below, and IF those criteria are met, the act becomes morally permissible. This means Foot would never say you have a moral obligation to kill somebody, simply that killing somebody can become permissible. This means that her view can be scaled in a society such that people who hold various answers can co-exist(in the sense of find each other ethical).

Notes on the doctrine of double effect

Foot in her original trolley problem article outlined the docrtine of double effect. The doctrine comes from Aquinas Summa Theologia and it goes something like this. For an action to be morally permissible, four things must be satisfied.

  1. The action must be good or permissible (nature of act)
  2. The bad effect must not be the means by which the good effect is achieved (means/end)
  3. The intention must be only for the good effect, the bad effect is simply a side effect.
  4. The good effect must be at least as important as the bad effect

Point 3 says that the END that is intended is the good and not the bad. An important discussion she has betweem action(direct intention) and allowance(obliquely intended). Allowance can be further broken down into letting things as they are, versus removing an obstacle or commiting an ommission. If we are deciding who to save, we can choose the higher number and if we are deciding what not to allow, we may also choose the higher number but we cannot compare the duty to save with the duty to help.

Equality of opportunity vs ergodicity

Different starting conditions lead to different probabilities to transition to a variety of professions. Some people say that the probabilities depend on innate talent, there is probably some truth to that but here I want to focus on the environmental aspect. Specifically, the idea of equality of opportunity.

First, let us begin by describing how things are. For the sake of simplicity assume there are only two possible professions, and that they mary each other such that there are only carpenter parents and mathematician parents. It should become clear soon enough that this generalizes for any number of starting positions and professions. In the image below, one should see that somebody born to carpenter(mathematician) parents has a 0.6%(0.3%) of becoming a carpenter and a 0.4%(0.7%) of becoming a mathematician.

Transition graph with two professions/starting positions

Now under my interpretation of equality of opportunity, this picture shows a lack of equality of opportunity. For equality of opportunity to be achieved here, it must be that the transition probabilities from both states be equal. This seems like a fair representation of the concept.

It should also be clear why this is a terrible idea, it entails the abolition of the family. How can the parents fail to pass on even the slightest edge to their children about their own profession? Either the parents are neglecting their children to an absurd degree or the children have been confiscated from their parents.


It should be clear that this is true even if children have different natural affinities, like below where I label children with math(carpenter) families that have math affinities as MMA(CMA), math(carpenter) families with carpenter affinities as MCA(CCA). Even if a child is naturally gifted in mathematics, it will never have the same probability to transition to mathematics as a gifted child from a math family. As a rather technical note, it should be clear that if there is no equal total probability of transitioning to each profession, then some professions will gradually go extinct.

Transition graph with 4 starting positions, Mathematician Parents with Carpenter Afiinty(MCA), Mathematician parents with Mathematical Affinity(MMA), Carpenter parents with carpenter affinity(CCA), and Carpenter parents with Mathematical Affinity(CMA)

This is part of the sickness of trying to construct a rational order, most human constructs will have the property of stationarity. The rational construct has a tendency to destroy diversity under the guise of empathy. These constructions are an attack on the inheritance we have as humans, the family being the most vital. This is a problem with our intellect, it so happens that stationary constructs are simply easier to make. However, as our ideas become developed, we can better articulate principles that are in line with our inheritance. We just need to make sure that we supress the tendency to rationally construct until it is in line with our intuitions.


In short, equality of opportunity ensures one of two things. Either a low diversity of opportunities, OR a lot of mediocrity. For everyone to have an equal probability of becoming everything, that must mean that either the professions in question are rather easy to learn OR it means that nobody has learned enough to give them an advantage. A better idea that equality of opportunity is ergodicity of wealth. The idea is rather simple, we want the wealth distribution to simply converge.

Wealth transition graphwealth distribution is ergodic if p is decreasing in w

For simplicity we assume you can only jump up one bracket but the idea does not depend on a single jump. So we can now present what the criteria for ergodicity are: The probability of going higher has to be decreasing in wealth, so for example p2 > p1 and p3 > p2. Technically, ergodicity in this context just means that the distribution of wealth can be modelled. That is, if the rich get richer faster than the poor, then the distribution kind of explodes(loses continuity). It is not clear WHY we should care about whether the academic should be able to model the distribution as a normative criteria, but at the very least this criterion allows for more diversity than a concept like equality of opportunity and perhaps more importantly, it does not call for abolition of the family.

The most obvious problem here is the measure, wealth, it is not clear what it refers to. If wealth measures physical stuff, then clearly this either entails a zero sum world OR that humans can infinitely expand their physical consumption. Yet it should be clear that if Anna has halloumi but prefers blue cheese, and Bob has blue cheese but prefers halloumi. Clearly if they exchanged, the total amount of physical wealth would stay constant but there would be more value.

So we could interpret wealth as a non-physical category. But then, we might have that somebody is better off by having lower physical wealth! Which means the criteria loses it’s normative force. Anyway, though, I don’t think this ergodic criterion is all that great, it is still heuristically better than equality of opportunity.

note: I am using the word ergodicity rather loosely here but I think it conveys the point. The idea for this blogpost comes from this paper, which gives a much more rigorous definition.

Utilitarianism: It entails Eugenics

Most people who know me, know that I am quite staunchly anti-utilitarian. I have a few reasons for this and I plan to give them all at some point but today let me simply give one. It seems to me that the most consistent utilitarians have no choice but to be Eugenicists. I salute Singer for being true to his position. I only wish more people would advertise this implication of utilitarianism. He of course tries to stick to disabilities when he discusses this, but in his book, ‘Should the baby live’ in the first 20 pages he makes clear that even moderately disabled children should be aborted, even if there is a family ready to adopt them. Let us give a general argument:

  1. All finite resources can be allocated to X or Y
  2. We ought to choose between X or Y by the measure M
  3. ==>X has a higher M, therefore we ought to allocate all finite resources to X.

Now we need only substitute: X = genes that have the best pleasure-pain tradeoff, Y = genes that don’t have the best pleasure-pain tradeoff, M = of higher pleasure-pain tradeoff. It seems pretty straightfoward if you ask me. But let us articulate a few things anyway just for completion. I am aware this is strictly not a valid proposition so I will include one in the bottom for the nitpicky.

So the argument does not say that all resources are finite. Indeed some could be in abundance, it simply says that IF there exist resources that are finite, those resources should be used to support the best by the measure M. This works pretty straightforwardly with utilitarianism but it could work with other things, says the person who wishes to maximize output. In general, anybody who views ethics as an optimization problem is automatically a eugenicist.

It should also be clear that the disagreement between utilitarians and a sort of Darwinism is only marginal. They both structure their argument in the same way, they might simply selecting different subsets from the population. This parallel is sometimes not obvious because utilitarianism is presented statically, but a coherent utilitarian understand that it is the long-term total utility that must be maximized.

Notice also that this is about the tradeoff. If some genes can have a great pleasure for little resources, then a utilitarian might favor numerous smaller organisms over larger ones. I suspect this is not the case since ever since Mill they might distinguish between higher pleasures and lower pleasures. So a higher consciousness can achieve greater efficiency, but if there are resources which the higher consciousness does not use, then maybe he will advocate for the smaller organisms.

In Joh Grays’s book, ‘7 types of atheism’, he puts light on how odd a phenomenon it is that two people as different as Michel Onfray and Sam Harris can converge on a single ideology. I would say that almost all non-religious scientists are of this persuasion. There very few ways to justify temporary pain and destruction to achieve a greater goal other than utilitarianism.

I don’t want to pull a Foucault here but the glove fits nicely: scientific types are merely trying to maximize their own power. First, a scientist is someone who spends their life studying IS and not the OUGHT. If scientists want to maximize their influence, they must inflate the importance of IS statements and deflate the importance of OUGHT statements. Otherwise, the scientist king would not be a good ruler. So the simplest way to deflate the OUGHT is to claim there is but ONE ought and to claim that it is so simple that they know it(utilitarianism fits perfectly).

So if one claims that only the material exists, and there is a certain material composition that is more desirable than another. They are unhindered in their attempts to reshape the world. There is no room in this mindset for admitting numerous desirable states and making morality an open-ended process, it instantly deflates the authority of the scientist, the scientists cannot make recommendations if there are numerous competing ought values.

So if somebody asks you, why you are not a utilitarian, I think it is straightfoward (but rather tongue in cheek) to answer with: ‘Because I am not a eugenicist’.

edit, for the nitpickers:

  1. We ought to choose how to allocate finite resources by the higher M
  2. X has a higher M
  3. ==>All finite resources should be allocated to X.

Review: How Innovation Works: Serendipity, Energy and the Saving of Time

Patents and copyright are justified with the question, “Why would people innovate if they didn’t expect to gain?» At first glance this argument sounds like common sense; however, if we dig a little we quickly understand that this is “lawyer logic”. That is, it is a story that conveniently has lawyers legitimacy reinforced.
Economics teaches us to think in a rather different manner. The narrow interest versus broad interests, specifically for two sides that have equivalent interests, the side that has the narrowest base has an advantage. It is simply easier to coordinate their interests. The lawyers are an example of such a narrow coalition. So if you think the question is sufficient to justify patents, the narrow interests are winning.
A more neutral framing can be presented, one such formulation could be “do most innovations occur faster with or without patents?”. Note that this question doesn’t have the conclusion baked into it. An infamous paper shows that more innovation is sequential, the more likely it is that patents will slow it down. The logic here is really quite simple, it may be true that the next innovation will be created faster with patents, but the innovation that improves on this innovation may be created slower because there is a decreased incentive to work on a patented invention.
This is the reason the sequentiality of innovation is such a contested territory. If one can show that innovation is a highly sequential phenomenon, then the lawyer logic is under attack. Matt Ridley’s book is an important contribution to this debate. In “How Innovation works”, Ridley isn’t interested in the theory of the debate, he wants to stick to what is important, the innovations.
His book reads like a series of short stories, indeed hundreds, about many different innovations. He attempts the impossible task of trying to classify them, in the beginning, he classifies them by industry (transport, health, food, energy) but by chapter 5 this approach is already limiting its limit, he then has a chapter on pre-historic innovations and low technology innovations.
Ridley doesn’t just focus on debunking the lawyer argument through the sequentiality argument. A running theme throughout the book is that people seem to be motivated by creating in itself, indeed the academics are usually the exception with their litigating behavior. The problem itself is academic; the working motto when it comes to evaluating innovation is “if I can’t measure it, it isn’t there”.
He makes a rather sharp distinction early on, the invention is not innovation. The person who brings the product to market and changes the norm is the innovator. According to Ridley, if patents were always given to innovations then they would be less harmful, the problem is that innovation is defined as ex-post; it isn’t possible to know which invention will turn out to be an innovation. As such most patents are given on things that never had any potential to generate revenue. An invention may or not turn out to be ready for the market, if it isn’t ready, then protecting it can only slow down the invention that will be able to penetrate the market.
Except for the sequentiality and the motivation of entrepreneurs, an important reason to undermine the lawyers’ argument for patents is the pushing of inferior products. In a world without patents, firms have an incentive to search the best possible product for any given end and then attempt to sell that product. In a world with patents, firms have an incentive to sell products that are protected. A safer more effective drug will not be pushed over a less effective and more dangerous one because they have property rights over the latter and not the former.

It is not until page 240 that Ridley takes a break from his short stories to try and synthesizes what has been learned so far. He also briefly stops by the economic literature to take a few digs at Mariana Mazzucato (who thinks the government is an important source of innovative activity). Perhaps the most fun chapter in the whole book is kept for the ending,”fake” innovations, with some absurd stories, such as Elizabeth Holmes faking her way into becoming a billionaire as well as becoming Obama’s ambassador of global entrepreneurship.
The elephant in the room is how come patents are so pervasive? While part of the story is the narrow versus wide interests. An even bigger part of the story is the United States forcing it upon others. Ridley’s book is rife with examples of both historical and contemporary interest with examples ranging from why coffee is subversive to the public good, European regulations being dictated by German manufacturers of vacuum cleaners, and how much Ridley’s ancestors were paying in patent fees.
The book isn’t perfect there are some important areas that are not expounded upon. Though household innovation as a concept is sprinkled throughout the book it is never quite given due credit in the conclusion. In some places Ridley is rather too cautious, he correctly claims there is no evidence that patents increase the rate of innovative drugs but does not advocate for the abolition of the patent system. The inverse problem occurs when he is discussing GMO’s(Genetically Modified Organisms), he does not sufficiently explore the problems raised which are both about the combinatorial nature of the disease and about the industry structure which has the potential to increase the cost of errors.
Nevertheless, Matt Ridley makes a powerful contribution to the debate by sticking to the stories. Innovation is a bottom-up process; no single mind is guiding it. Innovation can only emerge when the cost of errors is low and so Matt Ridley is on the correct side of the issue when he says “evolution” not “revolution”.

Why do firms exist? Three answers.

Legal: A compelling one, though not necessarily an interesting one from an economic point of view is that the legal system favors them. That is, firms can write contracts that regular people cannot. Firms are subject to limited liability and a different set of bankruptcy laws, can play accounting tricks to avoid taxation especially through debt instruments etc. This is a common-sense view but it seems like it could be historically falsifiable: By the weight of evidence that there was time rate compensation before the 18th century.

Uncertainty: A more classic answer is Frank Knights’s view, who claims that it is a way of dealing with uncertainty. Specifically, he says, under perfect knowledge, there would be no firms. With uncertainty, the focus of economic activity shifts from “doing” to determining what must be done and how it should be done. This task involves forecasting what consumers want and this is the role of the entrepreneur(the firm owner). In effect, the entrepreneur will guarantee the wage of his employees and reap the potential losses and rewards from his ability to forecast. This view does have limitations: just because entrepreneurs have better judgment and knowledge doesn’t explain why they start firms, indeed they could just outsource their forecasting services to other people. However, this explanation is usually paired up with a behavioral explanation. That is, employees don’t want to face the uncertainty and so prefer the certain payment to the uncertainty inherent in forecasting.

If we imagine this from the point of view of forecaster/entrepreneur, a natural question arises: When would one prefer to hire rather than simply compensate employees as a function of profit? It seems like the more competent the forecaster was, the more they would prefer to give employees time compensation. At the same time, as employees confidence in their boss increases, they would prefer to shift to proportional compensation. However the forecaster would then have to decide between two competing values, stability, and higher profits. If he splits the profits, the downside and upside will be bounded, if he chooses the profits…

Transaction costs: Coase has the modern answer to the question. Coase says that it is fairly clear that if an entrepreneur could simply buy all the goods at a competitive price, there would be no reason to produce them internally. So why might the entrepreneur not have access to a competitive price? There could be many reasons, Coase calls a “transaction cost” whatever it is that creates a barrier to the agent from accessing the price system. However, even if the price mechanism could be accessed, imperfections in the price system would also create an incentive to create your own product, for instance, if the good is being provided by a higher price than what it would cost if it was made within the firm.

Ends through truth

We have some end we want to attain. Regardless of what that end is(even if it is truth), we must take into account hazards in the way to getting there.

Suppose we suspect that getting closer to the truth, will in fact lead us to pursue our ends better. However the truth level required to pursue our ends better is level 3 and we are currently at level 1. To get to level 3, we must first pass through level 2 truth.

The state of level 2 truth however is dangerous. We suspect that going to level 3 by passing through level 2, will in fact lead us through a period where we can potentially be stuck. As a stylized illustration, imagine the following:

We are on a cliff if we fall off the cliff we die. However, it so happens that there is a spot on the cliff where a small jump can take us to another area where we suspect life will be better. So far, we are operating on the belief that “we instantly die if we go to the edge of the cliff”. The implication of this belief are twofold: 1) We don’t fall off the cliff because we are scared of going to the edge of the cliff. 2) We can’t move on to the other area where we want to go. Let’s try and apply our level’s of truths to this situation:

Level 1: “we instantly die if we go to the edge of the cliff”
Level 2: “We don’t instantly die if we go to the edge of the cliff”
Level 3: “We can jump to the better area by jumping off the edge”

So the problem is to be able to go to “level 2” as a transitional stage. The problem is that going to level 2, may lead us to be less careful around the cliff and cause us to fall down the cliff. Clearly, falling down the cliff will kill us and we will never be able to get to level 3.

In other words, to get to the truth we must be able to make our actions invariant to the truth. If we can somehow change our beliefs to level 2, but NOT increase the risk of dying then this is what is needed to get to level 3. It is vitally important that we don’t base our behavior on our beliefs until the beliefs reach a sufficient level.

So EVEN if you value truth, one should worry more about behavior. To reach truth you must have the right behavior, and sometimes, marginal progress towards truth may cause an absorbing barrier which stops the way towards the truth.

This is also why cultural attitudes are very important. If a culture makes their behavior contingent on beliefs, it will be very difficult to convince them to change their beliefs. If on the other hand, behavior is independent of beliefs, say how many atheist Jews treat it, then a culture can make progress towards truth without feeling the need to resist it.

In sum: The problem of belief that sometimes payoffs are not monotonic in truth value. To determine this non-monotonicity, we need the cost of error. If the cost of error is high, it is non-monotonic, if the cost of error is low, then it is monotonic.

Value hierarchy

If someone has a lack of strict hierarchy in values they may exhibit seemingly inconsistent behavior. Take for example, the case where somebody has to choose between three objects, {a,b,c}. It so happens that the best one in value 1 is “a”, on the other hand the best one in value 2 is “b”.

So to break the tie, the decision maker chooses a third value, value 3 and applies it to the winners of the first round, {a,b}. It turns out that “a” is the best in value 3 so the decision maker chooses “a”.

Now suppose that the choice was between {a,c}. Once again, “a” comes out on top for value 1 but this time, “c” comes out on top for value 2. So once again the decision maker breaks the tie by applying value 3, this time “c” comes out on top.

Conclusion: If “b” is available, the choice is “a”, if “b” is not available the choice is “c”.