A problem I’ve just got my head around with TypeScript is contravariance. What is contravariance? Well, it’s responsible for errors like this:

Type ‘(x: string) => string’ is not assignable to type ‘(x: string

number) => void’.

Types of parameters ‘x’ and ‘x’ are incompatible.

Type ‘string

number’ is not assignable to type ‘string’.

I found it baffling that TypeScript would complain about string|number not being assignable to string. In my head, I’m doing exactly the opposite: I’m assigning a more specific type to a more general type. The reason? Contravariance.

Technical explanation

Contravariance is a rule that enforces that a function’s input type must be a supertype of the input type of the function it’s being assigned to. It’s a result of the Liskov Substitution Principle, which is one of the SOLID principles. The Liskov Substitution Principle states that an object should be replaceable with any of its subtypes without breaking the program.

Here, TypeScript is saying that a function that takes a string|number as an argument can’t be replaced with a function that takes a string as an argument. That’s because if we replace the function that takes a string|number with a function that takes a string, we can’t guarantee that the function will work with a number argument. We often run into headaches because we’re narrowing a function for a specific use-case, just as we would a variable, but TypeScript doesn’t know we’ll only ever use it with a string. And, for that matter, nor do we or our colleagues or our future selves. That’s why TypeScript has all these rules in the first place.

In short, whereas variables are covariant (a variable of type A can be assigned to a variable of type A|B, but not vice-versa), functions are contravariant (a function of type A|B can be assigned to a function of type A, but not vice-versa).

Example

type Animal = {
  species: string;
};

const dog: Animal = {
  // You can assign a more specific value to the species property
  species: "Dog"
};

type Person = {
  introduction: (name: string) => void;
};

const alice: Person = {
  introduction: (name: "Alice") => `Hello. My name is ${name}`
  // Type '(name: "Alice") => string' is not assignable to type '(name: string) => void'.
  //  Types of parameters 'name' and 'name' are incompatible.
  //    Type 'string' is not assignable to type '"Alice"'.
};

Playground Link

Solution 1: Use generic types

type Person<T extends string> = {
  introduction: (name: T) => void;
};

const alice: Person<"Alice"> = {
  introduction: (name) => `Hello. My name is ${name}`
};

Playground Link

In this solution, we use a generic type to specify the type of the name parameter in the introduction function. This way, we can specify the type of name when we create the Person object.

Solution 2: Use a looser type for the function

type Person = {
  introduction: (name: never) => void;
};

const alice: Person = {
  introduction: (name: "Alice") => `Hello. My name is ${name}`
};

Playground Link

In this solution, we use the never type for the name parameter in the introduction function. Because nothing is assignable to never with covariance, everything is assignable to never with contravariance. This way, we are assigning a more general function to a type that describes a more specific function, which is allowed with contravariance.

Conclusion

I’ve found a couple of ways to work around contravariance in TypeScript. I’ve used both in my code, and I’m happy with the results. There are probably other ways to work around contravariance, but these are the ones I’ve found so far.

If I run into more easily-write-up-able problems like this, I’ll write more blog posts about them.

Notes

And how we can get there

Rutger Bregman [Elizabeth Manton trans.] (2017)

Utopia for Realists offers a general argument that far-reaching visions are a requirement for real progress alongside three specific candidates for such visions: universal basic income, a fifteen hour workweek, and open borders. It’s lively, readable, and engages with criticism of its ideas in a clear and frank way.

The first of the candidate visions is for a universal basic income. As the name suggests, this is the idea that some reasonable amount of money, say £15,000/year,¹ is given to everyone to do with as they wish. Proponents argue that the approach is cheaper, more dignified, and more effective than the alternatives. And proponents aren’t just liberal firebrands: one of the most compelling chapters in the book explores the story of Richard Nixon’s Basic Income Bill which could have set the mould for Western governments. The empirical evidence for basic income is promising, and set to grow considerably in the next few years as trials which are currently running start turning in data.²

The second candidate vision is less work, for a specific definition of work. Perhaps it is better phrased as ‘less drudgery’: the kind of work Bregman wants less of is the kind you’re not doing when “if you find a job you love, you’ll never work a day in your life.” The argument is the same one that has been made since the industrial revolution: efficiency is an increase in output (stuff) for a given input (time), so why not use increases in efficiency to reduce the input rather than increase the output? The short answer, at present, is because most of us get paid for providing the input, not for providing the output. Bregman’s assertion is that we are able to produce enough³ to spread the necessary drudgery out thinly between us all and spend the rest of the time in leisure.⁴

The final major candidate vision in the book is for open borders. Of all of the counter-intuitive topics in the book, open borders is perhaps the most counter-intuitive. Even for those who can see why giving free money to everyone makes everyone richer, it can seem bizarre that sharing our resources with strangers can make us better off. It seems bizarre because we intuitively think of these resources as pools with a limited capacity: e.g. that there is a fixed amount of meaningful work to distribute among those who are capable of doing it. But in many instances this is not true; additional people create additional demands for food, housing, schooling, leisure facilities, etc., all of which mean there’s more work for someone to do.⁵

Behind all these candidate visions is a condemnation of a politics which sees government as an optimisation problem: how do we allocate resources and deploy programmes to maximize X, where X is employment, education, the economy, housing, healthcare, or anything else acknowledged to be within the purview of government. Some of these seem to be no-brainers,⁶ but others deserve discussion about whether X is the right sort of thing to maximize in the first place. Bryan Caplan has argued that education is not something we should be seeking to maximize in The Case Against Education.⁷ Bregman offers the same treatment of employment. In both cases the argument is that we’re maximizing the wrong things not because the policies are ineffective, but because they’re affecting the wrong things. In yet another reprise of the same old script, we’ve let our measures become targets: we’re maximizing % of university-educated population rather than capacity for people to be effective and self-governing individuals, citizens, and employees; and we’re maximizing hours worked rather than tangible value⁸ created (and shared!).⁹

Ultimately, I’m wary of accepting the findings in the book too uncritically, especially because I agreed at the outset with the potential merits of universal basic income and obviating drudgery (and was agnostic regarding open borders). I’m not convinced that I will dive into Politics (with a capital ‘p’) to support these ideas in the way Bregman advocates,¹⁰ but if I do I suppose it is books like this one which will be partly responsible. More than any of the specific utopian ideas put forward, the book is a call for the reinvigoration of value-based Political debate, and it will be interesting to see whether that call can garner enough of a broad and diverse base of support to become a reality.¹¹

Notes

The Secret Life of the Brain

Lisa Feldman Barrett (2017)

We feel ways about things because things matter to us.¹ That is not as tautological as it sounds: individual instances of things make us feel things because, in general, over evolutionary time, those things had a meaningful impact on our survival. The genes of creatures who felt those things about those sorts of things reproduced at greater rates than the genes of creatures who didn’t feel those things, or felt different sorts of things in those circumstances. In the debate over the nature of emotions, the vast majority of scientists agree on this point.

What separates them is the question: what kind of things can we feel? For essentialists, there are some limited number of true emotions which are evolved specifically to motivate particular behaviours in particular kinds of circumstances: we feel fear to encourage us to run away, happiness to encourage us to engage, etc. Perhaps at any given moment we might feel a mixture of them, but, as in the animated film Inside Out,² they’re real and distinct parts of our minds which are dominant at different times. For constructivists, emotions are generated more dynamically in response to context, blurring into one another at the edges and demarcated by socially determined rather than discovered categories.³ Constructivists see emotions more like the rainbow: it may appear to be made of bands of colour but only because we put our colour categories onto it - in the rainbow itself there is only a continual smooth variation in electromagnetic radiation wavelengths. Fledman Barrett’s theory of How Emotions are Made is firmly in this latter camp.

What distinguishes How Emotions are Made from other constructivist theories of emotion is that it has predictive processing as its foundation. Predictive processing considers the brain’s role to be predicting its sensory inputs, which it does through a combination of learning (making its internal model match the world) and acting (making the world match its internal model). Feldman Barrett’s view of emotion is that it is a conceptual framework imposed on this basic regulatory system. As sensory input (and the brain’s own continuously evolving state) prompts learning and action, the accommodations come with affective consequences - they feel like something.⁴ This affective flavouring forms the basis of our experience, and onto it we project a socially-constructed array of categories, turning a continuous, featureless landscape of ‘ways we can feel’ into an atlas of ‘emotions’ which have boundaries, classic features, and ostensive evolutionary purposes. For Feldman Barrett, the essentialist perspective mistakes this atlas for the territory it represents, assuming that because emotions can be put into categories, those categories must represent something fundamental about the nature of those emotions.

The story is complicated, however, by a feedback mechanism in which the categories taught to us by our culture inform our classification of the emotions we feel, and, in turn, the emotions we are able to feel. We cannot, on Feldman Barrett’s view, experience emotions we don’t have categories for, even if our internal state is the same as someone who can and does experience those emotions. In teaching children to identify their emotions, for example, we turn people who lack emotional concepts and consequently cannot feel particular emotions into people who have those concepts and can feel those emotions. If we were to teach them different concepts, they would experience different emotions.

A futher complication is the idea that emotions are not diagnosable in any special way by the person who is experiencing them: they are just as readily interpretable by someone else, perhaps with different emotional concepts with which to categorise them. Moreover, someone else’s interpretation of your emotion has as much truth to it as your own. This is perhaps the most counter-intuitive inference Feldman Barrett draws from her predictive account of emotions: the idea that the subject of the emotion has no special authority over what emotion they are feeling. If I say, for example, that I feel sad, I am basing my identification of sadness on very different information than when I say that you feel sad. In the first case, I have my internal state, as well as my external behaviour. In the latter case I have your behaviour, though perhaps your behaviour includes helpful communication such as saying “I’m feeling sad.” While Feldman Barrett does focus a good deal on the role of interoception (our ability to detect internal sensations like how tense or relaxed we are) in emotion categorisation, this extra input is not allowed to constitute a fact-of-the-matter. The reasons for this are never adequately explained, but it is presumably due to emotions being created in the process of categorisation. If I categorise my own emotion as ‘sad’, that act of categorisation means that I genuinely am sad,⁵ and, because your emotions are in part constructed in response to mine, if you categorise me as being ‘sad’ then I’m sad in a real enough sense for your response to be generated.⁶

Because of the grounding in a predictive processing approach, emotions are successful or unsuccessful insofar as they allow better prediction of incoming sensory information (and perhaps our own and others’ behaviour). For Feldman Barrett, having a large and varied emotional vocabulary, and having the ability to categorise one’s own emotions effectively using that vocabulary (i.e. having good interoceptive sensitivity), allows us to predict ourselves and others better. A child who is learning about their emotions might only have very basic categories for identifying (and hence constructing) emotional instances: they may only be able to tell if they are angry. An adult teaching them, however, may construct instances of more nuanced emotions for the child: perhaps they’re vengeful, or overwrought, or jealous, or envious, or frustrated, or maybe just tired or hungry. Over time, as they become more adept at wielding emotion concepts, the child becomes better able to predict and understand their own emotional responses, and to produce emotional responses which are culturally and situationally appropriate.

In the tradition of popular science books offering a comprehensive theory of X, How Emotions are Made illustrates how widespread adoption of its theory of emotion will lead to a fairer, happier world. This, as is often the case, is the weakest part of the work: many of the changes apparently suggested by the theory do not require the theory to be made, and often do not even clearly follow from the theory at all. The other major sources of weakness in the book are a philosophical over-simplicity (which is perhaps necessary for popular science), and the caricaturing of opposing arguments to which it leads. At no point are opposing arguments really allowed to shine with any lustre or motivate any intuitions: a pity given how rueful Feldman Barrett is over the long history of strawmanning in emotion science.⁷ Lastly, in undermining robust distinctions between widely-recognised emotions such as joy and other products of the conceptual system (such as ‘things you could use to spread butter’), How Emotions are Made can leave the reader wondering whether emotions are worthy of study in their own right at all.

Overall, the book has some broadly persuasive ideas, although much of the persuasion is in the rejection of essentialist categorisation of emotions, which is (at least to my mind) relatively easily accomplished because essentialist thinking is appealing but almost always wrong. The broad point, that emotions are best understood in relation to affect-generating disturbances to our bodies energy distribution, is well made, though it is not clear this point necessarily implies the predictive account offered. Predictive accounts of neuroscience and psychology are in vogue, and have a broad appeal, and the account of emotion sits well within it.

Notes

A Brief History of Capitalism

Yanis Varoufakis (2017)

Talking to My Daughter is a critical history of capitalism from the perspective of someone who is acutely aware of its more pernicious effects. Despite being an academic, and deeply sceptical capitalism and of the political deification of the economy, Varoufakis has first-hand experience of macroeconomic policymaking after serving as Finance Minster for Greece towards the end of the government debt crisis in 2015.¹ The book offers a warts-focussed analysis of the roots of capitalism, exploring how each of its components arose and how each has shifted the balance of resources further from the many towards the privileged few.

Markets are the place where the story of capitalism starts. Early markets traded in commodities - goods produced for exchange. It was possible for people to do very well for themselves trading commodities, especially when reliable sea trade routes connected areas with vastly different natural resources where supply-and-demand principles could be harnessed to turn profit into more profit.² While commodities were traded to mutual benefit,³ commodities were a distinct from goods, including raw materials, land, and labour.

The next step is commodification: the extension of the class of commodities to almost all goods. Land and labour are commodified first in Britain through the process of enclosure,⁴ wherein previously-communal (or feudally owned but communally managed) land was portioned into parts and leased to the farmers who farmed it. Now, rather than subsisting on the land they once occupied, peasants had to either take on debt to acquire and manage an enclosure, or sell their labour to someone who had.

Enclosure produced both commodification of labour and the key engines of capitalism: debt and profit. Debts necessitate profit because, interest is needed to make a loan worthwhile for the creditor, meaning the amount to be repaid outweighs the amount borrowed, which in turn means the money borrowed must be used to produce more than its own value so the debt can be paid.⁵

The essential ingredients of capitalism in place, Varoufakis spends the rest of the book examining capitalism’s quirks: bubbles, crashes, automation, politicization, and environmental devastation. The same theme recurs again and again in these chapters: that capitalism is unworkable given human nature.⁶ Banking, particularly the delegation of the ability to create money to private banks, is shown to foster greater wealth creation and opportunity, but also to cause financial crashes through (apparently inevitable) irresponsible lending. Crashes are made deeper and more disastrous by decreasing the ability of consumers to purchase goods, which decreases manufacturers’ ability to pay employees whose wages will be spent on consuming those goods. These effects are magnified by automation: while goods produced through automation are cheaper, the process also pays fewer people meaning there are fewer people earning wages which can be used to purchase those goods. Finally, Varoufakis argues that the inherently political nature of economics should not mean that political decisions are made using economic machinery, arguing strenuously against the market solutions to climate change.

The real strength of Varoufakis’ work is to bring everything back to human conscious experience.⁷ He cares little for the abstract beauty of model economic systems, and sees value only where it eventually translates into experiential value for a conscious mind: a thrill of pleasure or the alleviation of a want. Economic activity, for Varoufakis, is the flow of experiential value through space and time, albeit abstracted into money for much of its journey. A sibiling of this phenomenal-experience-first perspective is the insistence on keeping in view the psychology of the economic actors, especially those with great power. The intimate and cyclic relationship between those with political and those with economic power is blamed for much of the devastation which capitalism produces: banks are left unchecked by government regulation in good times and bailed out by intervention in bad times, supporting an enrichment which pays for political donations and creates seats on directorial boards for retiring government officials.

There are weaknesses to the book, although it should be borne in mind that it is not necessarily intended to be a thorough academic work. There are arguments or examples scattered throughout the book which don’t quite work, and in some places it seems that other explanations are given a shorter shrift than they deserve. It doesn’t really address globalization or the vast improvements in the quality of life which have accrued to the majority of humanity in the last century or so (a phenomenon covered excellently in Factfulness). Finally, the book is almost entirely devoid of proposals for solutions to the catalogue of capitalistic woes it documents, satisfying itself instead with illustrating how things go awry.⁸

Talking to My Daughter is not a balanced book. It is written in a plain and engaging style, and the love and pain of a parent trying to parent from halfway around the globe is palpable. It introduces other (i.e. mainstream economics) ideas, arguments, and perspectives, but more in isolation than as a coherent objection to the framework which the book espouses. If you are going to read one book on capitalism, you’re better off reading something more even-handed, but if you’re going to read a few you can do a lot worse than include Varoufakis’ earnest and cogently argued effort in the mix.

Notes

Putting Ancient Wisdom and Philosophy to the Test of Modern Science

Jonathan Haidt (2006)

The Happiness Hypothesis is an attempt to fuse insights from psychology with theories from philosophy and religion’s long histories of enquiry into the components of ‘the good life’. While the book settles for the unsurprising and rather bland conclusion that treading a path between extremes is likely to be best for most people most of the time, it does, especially in the earlier chapters, manage to live up to the promise of holding philosophy’s feet to the empirical fire.

The key psychological points the book makes concern set points, free will, and sources of meaning. The key philosophical ideas are stoicism, deontology, utilitarianism, and virtue ethics, while the discussion of religious ideas is mostly couched in terms of a dimension of sacredness.

In keeping with several ancient dichotomies between instinct and rationality, urges and self-control, etc., Haidt introduces ‘the rider and the elephant’, the model of the self which lies at the heart of the book and which is broadly supported in psychology. The ‘elephant’ represents instincts/urges/habits/emotional responses/unconscious thought, etc. and is similar to Kahneman’s notion of System One processes. The elephant is essentially responsible for most decisions, especially ones which are made without ever attaining the status of ‘decision’. The ‘rider’ is the conscious part, the bit of you you think of as you, and as the analogy is supposed to suggest, its control over the elephant is incomplete at best. Haidt leaves it as a rather open question quite how much the rider can steer the elephant and how much it has to settle for coming up with explanations why the direction they ended up taking was just where they were hoping to go all along…

The second major piece of bad news for those hoping to come out of The Happiness Hypothesis with a recipe for the hypothesised happiness is the notion of set points. Affective neuroscience and psychology have demonstrated that each person has an idiosyncratic basic level of optimism and happiness, and that this is only perturbed briefly by life events, pharmacological intervention, etc.¹ This psychological insight is somewhat less secure than others - it’s easy to think that because something happens in the brain, or because the effects of some interventions are short-lived, that this means the underlying mechanism must be immutable. The brain is continually changing in response to itself and its environment, and it reflects experience as much as it does genetics, meaning there is a lot more to do to claim that set points cannot be shifted than pointing to homeostatic regulation of administered neurotransmitters or adaptation to life events.

The promise of the book starts to emerge when, building on the two previous ideas that you’re not especially in control of what you do, and you’re not going to be able to improve things a whole lot for yourself even if you become an expert elephant-wrangler, it introduces stoicism and relates the stoic (and Buddhist) philosophy of controlling responses to uncontrolled events to the basis of Cognitive Behavioural Therapy. A healthy dose of endurance for things you cannot control, and curation of considered responses to them, is justly applauded, but Haidt follows with a well-aimed criticism of stoicism’s lauding of detachment as a way of minimising vulnerability. Avoiding forming connections with others is an exquisitely logical way of avoiding being upset when they are no longer a part of our lives, but it ignores the vital point that connections with others are an important, almost certainly necessary, part of what brings meaning and happiness to our lives in the first place.

Haidt’s evidence for the importance of connections with others comes mostly from attachment theory, a characterisation of close human relationships which posits a few identifiable attachment ‘styles’ - collections of behaviours brought on by particular contexts such as novel situations, parting, and reunification. Attachment is thought to be established in caregiver-child bonding and form a model for significant relationships in later life. The philosophical treatment of love is notable for its brevity: the key point is that romantic love makes people behave in ways philosophers viewed as irrational and bestial, and was for that reason largely cautioned against and ignored.

Eventually, Haidt uses the importance of connectedness to others (both interpersonal relationships and embeddedness in society) to argue that a happy and an ethical life are inextricably entwined. This concept is captured by the notion of ‘virtue’ - of leading a life which is not only of benefit to others and morally praiseworthy, but also satisfying and enjoyable. To lend psychological backing to this notion, Haidt presents experimental philosophy evidence suggesting that people have broadly² similar notions of virtue across cultures, and psychological evidence that happiness and pro-social behaviour are linked (happy people are more helpful to others, and helping others makes people more happy).

Haidt is most well known at the moment for research into the association between disgust and social and political dispositions, and the background for this research is presented in a chapter on divinity. Haidt argues that the sacred is a dimension in human social organisation, alongside hierarchy (social rank/prestige) and closeness (emotional connectedness). The argument is not wholly compelling, especially given the ubiquity of the other two dimensions in human societies, but it is interesting and does demonstrate quite clearly that distinctions between the sacred and profane can be hugely important to a culture. For me, the most interesting notion in the book was the suggestion that crying when witnessing exemplary generosity, justice, etc. (as at the end of every decent children’s movie ever) was due to seeing something rising in this dimension of sacredness.³ Again, not wholly persuasive, but interesting.

Ultimately, Haidt comes to rest on the psychology of meaning: people get a sense of meaning from feeling engaged with the world - feeling like they can affect it and be affected by it. This is true of activities (work and hobbies) and of relationships (family, friends, and society). Thankfully, positive engagement with those aspects of life is both beneficial for others and rewarding for oneself. It is, in a word, ‘virtuous’. Insofar as philosophies have managed to capture or be compatible with the notion of interactive, positive engagement with the world, Haidt is prepared to grant them the psychology seal of approval. There is legitimate room to debate what constitutes positive engagement given different groups’ goals and perspectives, meaning ‘virtue’ is not an unambiguous beacon, but it is at least a useful framework for understanding the relationship between behaviour and happiness.

Notes

This post originally appeared on the ReproducibiliTea website. ReproducibiliTea is an organisation of Journal Clubs discussing open and reproducible science in universities all over the world. I organise the Oxford branch and am on the committee for the core organisation.

Last week’s ReproducibiliTea at Oxford Experimental Psychology (honourable mention to our Anthropology regulars!) went on a little longer than our usual one-hour of snacks and journal articles. We, like a few other ReproducibiliTea clubs, teamed up with the RepliCATS project and estimated the replicability of research studies for the benefit of science (and pizza).

repliCATS (@replicats) is a project aiming to understand the replicability of social and behavioural sciences and how well replicability can be predicted. There are three arms to the project: gathering researchers’ assessments of replicability of specific claims in papers across a broad range of disciplines, running replications of a subset of the assessed claims and attempting to predict replicability using machine learning techniques. The ReproducibiliTea sessions help with the first of these arms.

We spent around 30 minutes signing up to the repliCATS platform, which involves completing an interesting questionnaire about areas of expertise, metascience knowledge, and statistical knowledge, and watching the introduction video. Over the next couple of hours, we went over four claims using the platform’s IDEA approach. The IDEA protocol grounds itself on group decision-making literature which pleased me, as a graduate student studying group decision-making. Individuals make their judgements about the claim, then they have an opportunity to discuss those judgements with others. After discussion, each person makes a final decision in private. This allows for both a diversity of estimates and reasons as well as minimising groupthink effects, at least in principle. As a group, many of us were quite effusive (especially me) so there may have been some cross-pollination of ‘individual judgements’ due to facial expressions, quizzical sounds, or irrepressible comments.

Individuals judge how readily each claim can be understood, whether it appears plausible, and how many of 100 direct replications of the claim are likely to show an effect in the same direction (with alpha = .05). This seems pretty straightforward but, in practice, we often found it quite tricky…

For some of the claims, we found that the claim extracted by the platform for assessment bore little resemblance to the inferential test put forward to test it. For example, one claim was that people in condition A would out-perform those in conditions B and C. This claim was ‘tested’ by a t-test of condition A performance vs chance. These discrepancies sometimes happened because the authors didn’t do a good job of testing the claim, but in others the platform selected the wrong test. Also, in some cases, the N supplied for the test didn’t match the actual N reported in the paper or the test’s degrees of freedom. These issues made it difficult to work out whether we should assess the replicability of the claim or the inferential test.¹

For other claims, the tests seemed entirely misguided. The authors of one paper seemed to divide teachers into two groups based on whether their teaching was improving. They then used an inferential test to show that the ‘improving’ group improved relative to the non-improving group. That test seemed pretty pointless in terms of a meaningful claim, and it wasn’t very clear what we should assess. We could assess the probability that for any random set of 6 teachers you’ll be able to assign them to improving vs stable groups, or we could assess the claim that tautological t-tests should come up significant.

Of the four claims we investigated as a group, the most sensible one was that juvenile offenders are more likely to be placed in young offenders’ institutions if their family assesses as dysfunctional. We were pretty happy with the inferential test for that claim and it seemed intuitive. The only reason we came up with to doubt its replicability related more to the claim’s generalisability across time than the robustness of the study. The claim seemed to show a behaviour of the courts which could be subject to political correction which would make it hard to find the same effect if courts used in future replications behaved differently (perhaps as a result of the original study).

The whole exercise brought to light a good many considerations of what we mean by replicability and which factors we should or should not care about. It was also like a rapid-fire journal club with pizza, which is about the best way to spend an academic afternoon! I really like the ambition of the whole repliCATS project, from the breadth and depth of the disciplines covered to the idea of (responsibly) running machine learning on the studies. It’s a very cool project and I’m really looking forward to seeing how it evolves.

Notes:

Croteam (2014)

The Talos Principle is a first-person puzzle game interwoven with a story which explores key questions in religion of philosophy of mind: what is it to be conscious, what is the basis for being deserving of moral considerations, do we have free will, what is the point of existence, and how can we determine whether other agents are conscious?

The puzzle and gameplay elements are pretty robust (reminiscent of Portal and Portal 2, perhaps the best games ever made),¹ being do-able but not trivial. Some of the solutions seem quite hacky at times,² and some game elements are frustrating, such as lethal obstacles which can force you to repeat trival-but-time-consuming initial stages of puzzles. There’s also a large number of puzzles which are solved with techniques which are unique to that puzzle, and often quite unintuitive such that you feel you’ve learned a lesson about the game’s interpretation of physics, which you never use subsequently.³ These frustrations aside, the strengths of the core gameplay - engaging puzzles, insight-based (‘ah-ha!’) solutions, and 4-d thinking - are enough to frame the story and provide a meaningful feeling of progress.

The puzzles and gameplay are not why The Talos Principle is reviewed here, of course. Its story positions the player as some kind of android in an esoteric world of ruins from various ancient civilisations, commanded to solve the game’s puzzles by a disembodied Voice of Command identifying itself as Elohim. The other major characters are a virtual library assistant terminal which conducts the player through philosophical arguments, and one of the designers of the world in which the player inhabits whose philosophical curiosity is portrayed through a series of voice notes.

Key questions foregrounded by the game are the relationship between the player’s avatar and Elohim, who issues commands and prohibitions with which the player can engage, and whether or not the library assistant is a conscious entity.

Engagement with the library assistant, where most of the philosophical debate takes place,⁴ proceeds through the selection of pre-determined dialogue options. The variety is fairly good,⁵ and although it’s readily apparent that the assistant will argue you to a contradiction or a standstill whichever option you pursue, that’s pretty much the point. If there were obvious answers to these very difficult philosophical questions we would almost certainly have found them by now.

The other components of the game, Alexandra’s voice notes and the various written information which pervades the computer terminals scattered through the world, work well to provide an impetus for the player’s philosophical curiosity and bite-sized doses of backstory/philosophy respectively.⁶

Finally, the presence of earlier androids in the game world, both in their occasional physical presence and their frequent messages embedded in QR codes pasted on walls all over the world, helps to problematize the question of the player’s avatar. In a nice touch of meta-ludonarrative synergy, the player sometimes gets to create these messages themself, by selecting from a list of pre-specified options… From my own philosophical perspective, I’d have liked to see a sense of progress in the distinctiveness and apparent consciousness of the past agents as their version numbers increase,⁷ but it’s understandable why this is underplayed.

Overall, The Talos Princple is an enjoyable puzzle game, and should please fans of puzzle games and philosophical enquiry equally, while being a special treat for those who enjoy both.

Notes

Exposing data bias in a world designed for men

Caroline Criado Perez (2019)

Invisible Women is a litany of complaint about how ignoring the needs and wants of women is ubiquitous, from basic scientific and market research to policy planning and government decisions. It’s also robustly evidenced, surprisingly readable,¹ and long overdue.

The point of the book is simple to sum up: humanity is perceived as being male by default, and this means women’s needs are not looked for or seen. Whether it’s assuming that work means driving off to work in the morning, spending the day at the office, and then driving home in the evening, or assuming that female bodies behave like smaller, lighter male bodies in car crashes, the fact that half of the population has been treated as an irrelevant afterthought is fairly astounding.²

The book offers numerous examples of areas in which women’s experiences are sidelined. The measure of GDP, for example, doesn’t take housework or non-professional caring work (both predominantly done by women) into account, despite their meeting most reasonable definitions of ‘work’. This point is also made by Doughnut Economics. Women rely far more heavily than men on public transport and walking, so transport networks and snow-clearing schedules designed around car travel disproportionately disadvantage women. There’s some evidence to suggest that drugs and other medical treatments can have different effects in different hormonal environments,³ and clothing, including safety gear, is often fitted for male as opposed to female body shapes. The list goes on.

It’s an important point, made occasionally in the book (including in the preface), that the vast majority of these issues are to do with sociocultural rather than biological differences.⁴ Women are perceived as different, and socialised to behave differently, and it is predominantly these factors which both produce the overlooked differences and account for their being overlooked. Women are taught to be silent, and then ignored because they are.⁵

The examples are both highly varied and highly specific. Sometimes this has the desired effect of painting a picture of a pervasive problem which surfaces anywhere it’s sought. At other times, it feels a bit like patchwork: as if a handful of carefully selected examples have been cherrypicked to provide a misleading impression. This is likely an artefact of style rather than substance, however: one gets the impression that similar evidence would be available in comparable countries/areas/disciplines, and the lack of data on the data gap is a product of the data gap itself.

It is perhaps arguable that the focus on gender is unhelpful, and that the data gap is better conceptualised along a different axis, perhaps individual power or cultural prominence, but subtleties like that already conceed the major point that a substantial data gap exists which it is in our collective interest to address.

Invisible Women is only important to read until you’re sufficiently persuaded by the argument to consider how your work could be adjusted to address the data gap. If that’s the subtitle, then fair enough. If it’s the whole book and a black hole of under-the-line comments all over the internet… okay. But for those who make key decisions,⁶ especially, the point should carry that we need better data collection.

Notes

Ten reasons we’re wrong about the world - and why things are better than you think

Hans Rosling, with Ola Rosling and Anna Rosling Rönnlund (2018)

It would be awfully convenient if learned things stayed learned. Or so we often feel when we just can’t quite remember that name, or that holiday, or where we left our keys.¹ Factfulness is essentially a book about the opposite problem: when our memories for things we’ve learned stay the same, but the facts change.

In a masterclass of big-picture thinking, the authors demonstrate the poor fit between the ways we’ve learned to think about the world and the way the world actually is. This isn’t really a book about cognitive biases so much as it’s about providing better ways to conceptualise the world. We can’t make good decisions with bad information, and Factfulness aims to provide that information.²

The most important conceptual overhaul the book aims to achieve is replacing the binary rich/poor, West/Rest, developed/developing etc. distinctions with ‘the Four Levels’ of income.

• Level 1 (US$0-2/day)
  • People on Level 1 (about 1 billion) look like the pictures of dire poverty from charity fundraising materials: walking long distances each day to collect water; little or no access to healthcare; sleeping on dirt floors.
• Level 2 (US$3-8/day)
  • People on Level 2 (about 3 billion) probably have rudimentary transportation, such as a pedal bike, reducing the time taken to collect water, and have some variety in their diet, but are highly vulnerable to unexpected expenses such as healthcare, repairs, or redundancy.
• Level 3 (US$9-16/day)
  • The 2 billion people on Level 3 get cold water taps, stable electricity (and therefore refrigerators and more varied diets), and can afford occasional holidays. Sudden expenses like motorcycle repair are still a threat, but now they come out of savings for sending the children to high school.
• Level 4 (US$17+/day)
  • The 1 billion people on Level 4 are the book’s (and this blog’s) audience. People with access to hot and cold water taps, cars, foreign holidays, more than a decade of education, pension plans, etc.

While the binary model was useful once (when the world really was split into developing and developed countries), this model needs to be unlearned because the world has changed. Nowadays people’s lives are more accurately described by their income level than their geographical location.³ This is neatly demonstrated by the authors’ Dollar street project, which shows visually how cultural differences are dwarfed by income differences.

Equipped with the Four Levels, it becomes easier to anticipate how the demographics of the world are changing. There is a steady transfer of people up through the levels,⁴ and with it reductions in infant mortality and childbirth rates, and increases in vaccination rates, years of education (for boys and girls), and the like. The resulting picture is of a world with a similar population to today, where most people live in Asia on a reasonable income level, including demand for electricity, travel, and consumer goods. Whether you’re concerned with how to provide power without destroying the climate, or how you’re going to make money from selling your local gadget, a fact-based world view is vital to making appropriate decisions.

The book provides a neat summary of some of the mistaken instincts we have about the world (in that sense it is about cognitive biases), but more than anything else it makes a very successful case for lifelong learning. Just because you learned something about the world once, it doesn’t mean it’ll stay true for your whole life.

Factfulness is humane, cautious where warranted, but consistently frank, powerful, and persuasive.⁵ It’s surprisingly light reading which cannot be taken lightly. Even for those who fundamentally reject the idea that useful knowledge about human flourishing can be gathered from aggregated data, the book does a better job of outlining that position than others.

Notes

Seven ways to think like a 21st-century economist

Kate Raworth (2017)

The central theme of Doughnut Economics is that the discipline of economics is badly out of step with its purpose, and that it is too concerned with theorising about what it can measure rather than what actually matters. Reading the book is like talking to a student activist: there’s a crushing indictment of the present state of affairs, a jumble sale of earnest indicators of how things might be better, and a dearth of thorough, practical, specific solutions. And as when talking to a student activist, it’s worth remembering that the absence of useful solutions isn’t grounds for dismissing a problem.

The titular case study is the doughnut, a graphical replacement for Kuznets’ GDP curve, and the Raworth’s point is that economics has largely become GDPology. If GDP were a good measure of the total economic activity of a country this would be bad enough, its failure to capture arguably the most valuable aspects of economic life mean it constitutes a fatal flaw.¹ Enter the doughnut: two circles which define an area of economic harmony; the inner edge of the ring sits atop an abyss of deprivation in which people lack the necessary resources (food, sanitation, healthcare, education, etc.), while the outside edge presses on the limits of the resources we have (global temperature, ocean acidity, ozone coverage, etc.). Between the overreach of unsustainable exploitation and the shortfall in adequate provision for human life lies a zone in which the economic system is in balance both within humanity and between humans and the environment.²

Raworth proceeds through various other examples to explore territory long ignored in classic economics, including resource management alternatives to the market. Markets are efficient systems for distributing certain resources, but they are subject to a whole slew of bugs (e.g. information asymmetry), exploits (e.g. insider trading), and implementation issues (e.g. regulatory capture) which mean they do not perform their function properly. Markets should be a tool to help people thrive, not vice-versa.

Market alternatives include commons management, which is viewed somewhat sceptically by game theoretic models of rational economic actors (see Thinking, Fast and Slow). The problem, known as the ‘tragedy of the commons’, is that everyone using a shared resource such as a common for grazing their cattle can take all of the benefits for overexploitation (e.g. grazing an extra bullock) while taking only a share of the costs (depletion of the pasture), resulting in a pressure on each person to maximise exploitation of the resource. Commons do exist, despite this neat model of their failure, and are mostly powered by strongly enforced bylaws and shared expectations throughout the community.³

The recognition that the market is not necessarily the best solution to resource allocation problems is refreshing: by now it’s pretty well established that casting things in terms of an explicit value alters the way people think about them (Kahneman notes that parents were late to pick up their children from childcare more often after a fine was introduced, presumably because the payment alleviated their guilt).⁴ Similarly, Raworth’s most exciting suggestions on tax are to target resource use, not employment, income, or wealth. Taxing resource use is enticing because it reifies the idea that the planet and its resources are jointly owned by everyone, and that when we use some of it up we should pay the owners (i.e. everyone) for doing so. It also alleviates some of the difficulties in European countries with stewardship of valuable resources - some people inherit buildings, artworks, etc. which are very valuable (meaning they have high wealth), but which do not or should not bring much income (meaning they have no money to pay wealth taxes).⁵

For me, the whole idea of the book was brought out most clearly by the chapter on growth in the economy. In the midst of the fierce assault on the ubiquitous notion of Growth Is Good by the growing movement for environmentally-sensitive economics, which argues that Growth Is Bad, Raworth takes a moment to put on the guise of a puzzled ingenu and ask “does it matter?” Raworth’s concept of ‘growth agnostic’ economies is appealing: instead of concentrating on whether or not growth is necessary for economic prosperity, or whether it is unsustainable in the face of limited resources, the focus is on ensuring the the economy does its job of providing for the needs of everyone regardless of whether or not it is growing.

Overall, the book somewhat resembles a doughnut in that there’s a fair bit to chew through but it’s empty in the middle. This is mostly a consequence of the book being a signpost to areas which are in dire need of research rather than an account of the results of that research. In many cases Raworth’s only conclusion is ‘needs more work’, but that doesn’t undermine the importance of stating the shortcomings of current economic research and suggesting areas for improvement.⁶

Notes