The fetishization of numbers in policy

Note: this article is not about how politicians find numbers sexy. The word ‘fetish’ in this context  designates something that is used or done ritualistically rather than pragmatically.

In the world of UK public policy, everyone loves quantitative skills. Making policy ‘evidence-based’ is considered a matter of significant importance, and the National Health Service is held up as the shining example – primarily because of the prominence of randomized control trials and value-for-money. In the past decade this has led to the development of ‘What works’ centres, such as the College of Policing’s ‘What Works Centre for Crime Reduction”, and LSE’s “What Works Centre for Local Economic Growth”. Based on my research on these centres (from articles such as this), they tend to subscribe to the ‘evidence-based’ movements hierarchy of evidence, with Randomized Control trials at the top and anecdotal experience at the bottom (for those unfamiliar with this hierarchy, my main point is that data is on the top).

This increased emphasis on evidence is not inherently a bad thing. Indeed, in many ways it is a positive development. However, I have a number of concerns on the ways that these work centres interact with policy.

Concern #1: Begging the question – What works… for what?
The title ‘what works’ begs the question of what the interventions are
working for.  On the College of Policing’s What Works Crime Reduction page (link), we are presented with a list of interventions, with data on cost, effectiveness, where it works, and the like. But this cannot answer the question of what the Police should do in the first place.

Concern #2: The ‘what works’ frame
‘What works’ neglects the reality that problems can be described in multiple ways. For example, are we concerned with young hoodlums who have not been taught proper values, or are we concerned with oppressed minorities who are lashing out due to opportunity deprivation. There is often not a natural way to interpret data – our personal values play a significant role. The ‘what works’ language seems to cover up these ambiguities by assuming a common frame.

Concern #3: Politics and evidence
‘What works’ does not seem to engage with the reality that policy overlaps with politics. Evidence is regularly used as ammunition to support pre-existing positions, rather than forming a basis for re-evaluating positions. Not that I mean to be entirely down on politics, I just mean that evidence will not transform politics.

Concern #4: Stifling innovation
When practitioners focus on ‘what works’, their attention is necessarily backwards looking, because we necessarily cannot have evidence on new ideas and approaches. ‘What works’ therefore cannot help us prepare for future problems, and may potentially hinder policy that looks to deal with future problems because ‘its not evidence based’.

Concern #5: The gap between theory and practice
Models and quantification represent reality, but there is always a gap. Further, small errors compound quickly. This is why our ability to forecast the weather drops off very quickly. This does not mean that we should not use models (we must!), but we should do so while being mindful that they are not infallible.

I intend to write a more in-depth article on this topic, but here are my immediate thoughts. Let me know what you think.



via Blogger

Elected Responsibility

Now that Canada has a new government, I’m going to dedicate this post to exploring the moral responsibilities of elected officials. Elected officials are in a bit of an odd spot, morally speaking, because they must factor into account the moral weight of their responsibilities to the citizens that elected them even when the expectations of those citizens run counter to their own moral precepts. That is to say, I believe that being elected carries a notable moral burden to represent those that elected them, and it seems quite easy for a politicians moral responsibility towards those that elected them to come into conflict with their own moral beliefs. What this entails, however, can be difficult to parse, and this post will examine what exactly this responsibility entails, as well as how far the responsibility extends.

What is the nature of a politician’s moral responsibility towards those that elect them? I think that the key word or concept is that a politician is a representative: they are to represent the interests and beliefs of their constituents. There are two immediate problems with this, those being paternalism and minorities. First, should a politician represent their constituents expressed interests, or can she act as she sees best for the people. If we believe in the acceptability of paternalism then a great deal of the complexity is eliminated from this topic: the responsibility of a politician seems analogous to the responsibilities of a caretaker or parent – by taking up political responsibility the politician takes on duties to look after the well-being of their constituency. However, this paternalistic interpretation should be considered cautiously, since it is rather contrary to the spirit of democracy. For paternalism necessitates the assumption that certain people know what is best for others better, but democracy seems to assume that the electorate should have a say in their governance (at least on its face). There is a great deal more to say on paternalism, but I will not examine it any further here. The issue of minorities is more challenging, because it gets at a problem that democracies regularly struggle with: does a government represent the people, or just the majority? It is often the case that an elected offical has but one vote to represent the nuanced variety of opinions that they represent. On the other hand, it seems that they still bear the responsibility of advocacy for and attentiveness towards their constituents. The resolution of majorities and minorities seems like it must ultimately come down to judgement calls of the elected offical: for example, they may decide that it is important to speak for a cultural minority, but disregard, say, flat-earthers.

This leads into another issue – are the electorate electing someone to enforce certain beliefs, or are they electing a person that they trust to make good decisions? The answer will likely not be uniform. It is tempting to say that the answer depends on what a politician promises, if a constituency elects a clearly proclaimed feminist then they cannot reasonably expect that she will ban birth control. However, the electorate is limited in the choices they can make: they can only vote from a small selection of candidates, which inevitably means that no candidate will fully embody the beliefs of any constituents. This certainly makes it strategic to compromise one’s own beliefs to get votes, but it also raises the question of whether such compromising also has moral elements: should candidates seek to offer platforms in order to provide greater choice to constituents?

One particular feature of the Canadian system that I also wanted to mention is the fact that we do not elect our prime ministers to be prime minister – instead our prime minister is determined by the party with the greatest number of seats. This renders the representative responsibilities of the prime minister and MP’s even more difficult to discern because it makes individual votes play double duty – voters are simultaneously selecting who they want to represent their local area and who they want to represent their country. These desires do not necessarily align.

Those are my thoughts for the moment, let me know what you think,

Thank you,


via Blogger

What is an informed voter?

Given the amount of time that that I have spent volunteering for the NDP (New Democrat Party) for the upcoming Canadian election, I feel that I am more informed about the issues in this election than for any other election I will have voted in. This brings up a question that I have previously reflected upon. What is a responsible voter? Crucial to democratic rhetoric is the notion that voting is both a right and a responsibility. That even if you do not know who you intend to vote for, you should, at the very least, show up and spoil your ballet. This is a very minimalist notion of voting responsibility, even if it is one that many people fail to achieve. The question remains, however, how much people should aspire to know or understand in an election.

There is an economics argument that I have read a notable amount on which argues that informed voting is a kind of tragedy of the commons – we would all benefit substantially if all voters spent an hour researching who to vote for (if nothing else, this would lead to higher levels of scrutiny for political parties), but we would all prefer it if others spent that hour researching their vote and we spent the time relaxing or researching where we would like to go on vacation next year. Similarly, party loyalty is arguably an effective means of saving time and energy – after an initial investment deciding on which party to support, the voter can assume that the party will continue to represent their beliefs in the future. In asking the question of how much voters should know, I do not think that this kind of economic logic can be completely disregarded. Our time is precious, and I do not want to take it as a given that pursuing political knowledge is a responsibility of citizens. There is something to the rebuttal that is often uttered that people have fought and died to have the right to vote. This, however, does not quite draw out why it is a responsibility. Perhaps our responsibility is to our fellow citizens, though this is possibly contradictory given that each vote in the voting pool dilutes the votes of your fellow citizens. There is some evidence that group decisions can actually be better than individual decisions in some cases (see more on this here), though the extent to which this is the case is difficult to discern.

So how much should people aspire to know? It is obviously not a formal or lawful responsibility, at least not in Canada (there are 13 countries where voting is compulsory, according to wikipedia, including North Korea, where ‘Dissenting votes are possible but lead to repercussions for voters’). The question is instead how much a ‘good citizen’ should aim to understand in making their vote. Of course, notions of ‘good citizenship’ only have power in a very existential way, quite distinct from what is advantageous as I’ve explored in the previous paragraph. We all ‘should’ be good people, that’s pretty much the definition of good, but the main force behind such a concept is just our desire to be such. This, then, is my question: for someone seeking to be a ‘good citizen’, how much should they set out to know about politics. Even then, though, I do not think there is a uniform answer. When it comes to doing good work, we split responsibility. We can not be masters in all areas. Having a strong grasp of politics, like being a good teacher or an active member of a volunteer fire brigade* requires a significant investment of effort. It seems quite reasonable to say that some members of a community should be knowledgeable about politics, but it also seems reasonable to say that someone who’s interests lay elsewhere may lean on political savvy people they trust when it comes to political decisions.

As is often the case, after exploring what I know of the topic, I must conclude that I have less of an answer than when I started. I’ll probably write on this again next week.

Thank you for reading, let me know what you think,


*There are probably better examples in existence.

via Blogger

Are Deserts Earned or Revealed?

Are Deserts Earned or Revealed?
When meting out justice, should we proportion our punishment to criminals’ intentions or the consequences of their actions? In his essay, Criminal Attempts (2003, 77-102), Feinberg proposes that the consequences of a criminal’s actions should be an irrelevant factor when determining their punishment. Whether someone succeeds at realizing their intentions is a matter of chance. Unless we can provide a reason that success is somehow relevant to blameworthiness, Feinberg argues, we are being arbitrary and therefore unjust when we give different sentences to criminals based on chance. I am sympathetic to Feinberg’s goal of reducing arbitrariness, but I am unconvinced that his reformist proposal persuasively succeeds in making justice non-arbitrary. In at least some scenarios the fact that a particular individual attempts a crime seems to be a matter of chance in the same way that succeeding in our intentions is a matter of chance. To the extent that possessing a particular intention is a matter of chance, to escape arbitrariness I propose that Feinberg must rely on the counter-intuitive idea that deserts are character based instead of action based.
Section 1: Feinberg’s argument
Feinberg’s central argument is that a justice system should proportion its punishments to an individual’s intention instead of the consequences of those intentions because the consequence of actions is not relevant to an individual’s blameworthiness. Fienberg takes as given that ‘blameworthiness is normally compounded out of such factors as beliefs, emotions, motives, objectives, and intentions’ (79). It then follows that what a person deserves depends upon these mental features, and we should proportion our punishment to an individual’s intention. Since succeeding in a crime is irrelevant to an individual’s blameworthiness, it is arbitrary to base our punishments on success or failure.
Section 2: Blameworthiness and Character
Before getting to my main argument, I need to explicate further the concept of blameworthiness. I believe that Feinberg speaks somewhat loosely when he states that blameworthiness is compounded out of mental features. Blameworthiness and action are inexorably linked – I think most would agree that the mere contemplation of a crime is not a sufficient condition for blameworthiness. Someone is not guilty of murder the moment they begin plotting a murder. I do not mean this as merely an epistemic point, though it is true that intentions are epistemically unavailable until they are put into action. What I propose is that if we want to attribute responsibility for an action then we need to assume that humans have some kind of enduring element for responsibility to attach to. This feature could be dispositions, or it could be some kind of enduring character. Describing the exact nature of this enduring element is not crucial for the purposes of this essay. For simplicity’s sake I will refer to it as character. Since intentions and thoughts are not obviously selfsame with the actions someone undertakes, it is at least theoretically possible that a person could be unaware of their own hidden character until they are in the midst of an action. This account suggests that when we punish crime that we are really punishing this character that an action reveals, an idea that I will explore further in the next section.
Section 3: Luck of Temptation
It follows from the preceding paragraph that there may be crimes that people with particular characters commit when exposed to temptation: for these kinds of crimes it is a matter of chance that an individual commits them. It seems intuitive to me that these types of crimes exist – it seems to be the definition of a crime of passion. The perpetrator of the crime can even haveconceivably been unaware of their own temptable nature. Imagine two different men come home and discover their spouse with a lover. One man (angry1) becomes enraged and murders both of them, while the other man (calm1) politely asks the lover to leave and proceeds to have a rational conversation with his spouse. It seems quite conceivable to me that both could have been the same epistemic circumstance of not knowing their own character in relation to such circumstance. We can also imagine another man (angry2) who would act as angry1 did if exposed to the same circumstance, but who has had the good fortune of marrying a loyal partner. It then seems to be a matter of chance that angry1’s negative character was revealed, while angry2’s negative character remains hidden.
That certain crimes are committed as a matter of chance reveals an ambiguity in the concept of just deserts: are people blameworthy because of the actions they commit, or are we punishing them for the objectionable character that their act revealed? I will not endeavour to do full justice to this question here, but I do propose that Feinberg’s concern about arbitrariness in justice should drive him towards the latter option. If we fail to acknowledge that some crimes are an interaction between character and circumstance, then we seem to do a conceptual disservice to those we are punishing. The Milgram experiments1come to mind as good evidence that the fact that one individual commits an act rather then another can be a matter of chance. If someone actually committed a murder in a Milgram style scenario, it seems we would be derelict if we did not acknowledge that over half of the population would have done the same
The purpose of this section is to pose the following challenge to Feinberg: for crimes of passion or of extreme circumstance, is it unjust that we only punish those who have the misfortune of being exposed to the temptation to behave badly? It seems that fairness would dictate we should also punish those who are lucky, since they are of equivalent moral character.
A response to my challenge seems readily at hand: though it is unfair and unjust for individuals of the same quality of character to receive disparate deserts, this disparity is not one of our making. One of the ends of our justice system is to make sure that people (specifically criminals) receive their just deserts, but the justice system can not be faulted for not possessing infinite resources. Our epistemic circumstances are such that we can not give people deserts based on their hidden character; we must instead rely on the actual manifestation of character we can observe in their actions. We have not erred if we fail to do that which we are incapable.
Section 4: A Life’s Deserts
Though my previous challenge was resolved without much difficulty, the answer has led to a more significant problem. The issue is this: are deserts deserved at a particular moment in time, or are deserts accrued over the course of a lifetime? In asking this question I am assuming that our character or dispositions change over time. This is an assumption that would bear further investigation if I were trying to propose a theory of character, but it seems to be both intuitive and probable by my anecdotal experience.
First I will illustrate the problem. Let us imagine that we have two people who, in their youth, were of such a character that in the right circumstances they would shoplift. One of these persons (Unlucky Lily) was exposed to temptation and subsequently punished, while the other (Fortunate Fabian) was not exposed to temptation. Both of these people then grew out of such character. Let us imagine, however, that we somehow became aware that Fortunate Fabian had previously been temptable to shoplifting. Does justice dictate that since both were of equivalent character we should therefore punish Fabian? It seems strongly counter-intuitive to say that we should punish Fabian for a disposition that he no longer possesses, but if we do not punish him then we seem to declare that punishments are only deserved for character traits that are actually realized. In other words, if we do not punish Fabian for his previously negative character we endorse luck as a factor in the dispensation of punishment because we are making punishmentcontingent on being exposed to temptation at the right cross-section of life. I will call this the cross-sectional problem of justice.
One may criticize the problem I just illustrated by saying that the reason it seems counter-intuitive is because it requires that we imagine ourselves in an impossible epistemic circumstance. The only evidence that we can have of an individual’s character is their actions – it is nonsensical to talk of punishing someone for a particular non-realized character in the past, because we will never be in such a dilemma. The closest analogy would probably be punishing people based on the statistical probability that they have or had a particular moral character, and this kind of action is clearly egregious. The cross-sectional problem of justice may not seem like much of a bullet to bite because we will never need to actually pass judgement in the counter-intuitive way that it describes. However, I have a proposal of an interesting consequence that follows from endorsing non-arbitrariness in the cross-sectional problem.
If we endorse the idea that temporary character traits or dispositions leave enduring stains of blameworthiness on an individual, then it seems to follows that the duration or persistence of temptation-realized character traits is a relevant factor in the meting out of punishment. The primary point that the cross-sectional problem identifies is the question of whether we care about whether an individual’s blameworthiness is based on them possessing a trait now, or that they have ever had the trait. A desire for consistency drives us towards the second because it can be chance that causes one individual to reveal themselves as possessing a particular temptable character. From this it seems to follow that the persistence or duration of such a trait is a relevant factor in its punishment: a person who possesses a particular negative character trait for years seems to be of more blamable character then someone who has the same trait for only a month. Again, this observation is difficult to use in practice, unless we are comparing between crimes. Though it would be near impossible to determine the persistence of a trait withinan individual’s character, it seems much more feasible to me to say that generally or on average traits have different degrees of persistence. A perchance for vandalism could be a generally short-lived trait, while petty theft might be an enduring one. If this were the case, then vandalism would generally express a shorter duration blemish on character than petty theft, which seems to be a relevant consideration for just deserts.
What I’ve aimed to demonstrate with the preceding arguments is that arbitrariness is not avoided by simply punishing people based on their intentions rather than their actions. For the concept of blameworthiness to avoid the accusation of arbitrariness due to circumstantial luck, some conception of deserts being character dependent seems to be required. Once we divorce deserts from actual behaviour, however, we encounter some odd consequences. The first odd consequence is that character based deserts seem to demand that if we learn someone previously had a negative character trait then they justly deserve punishment. The second odd consequence is that the persistence or duration of a trait becomes a relevant factor when determining an individual’s just deserts.
Though I am sympathetic to Feinberg’s goal of reducing arbitrariness in the justice system, I am not convinced that we succeed in doing so by attributing substantive responsibility based on intentions. In this essay I have provided an account of some unusual and counter-intuitive results that seem to follow from adopting Feinberg’s position. My aim has been to demonstrate that his reformist position has significant conceptual obstacles to overcome before I would accept that he has successfully reduced arbitrariness.
Feinberg, Joel. Problems at the Roots of Law: Essays in Legal and Political Theory. Oxford: Oxford Scholarship Online, 2003.
1The Milgram Experiments were a series of famous experiments on obedience. Milgram found that under relatively gentle pressure over 60% of subjects would obey instructions to the point of committing murder.

via Blogger