What would Peter Singer Do?

A note from Elizabeth:

I asked David to write a blog about the refugee crisis because I felt I didn’t know enough about it to do so myself. He replied, ‘I don’t know enough about it.’ We then simultaneously started writing blog posts that included the phrase ‘What would Peter Singer do?’ So, you know, snap.

I’m going to use his, and not mine, because he has a far greater readership and it will reach more people.

You see, the reason I wanted to write something is because I want to share this link – an overview of how you can help, by donating and so forth. (Although after reading David's post maybe it should be this link instead.) It seemed disingenuous – as my colleague Anna told me – to just post the link to our Facebook page, without offering some sort of explanation of our ‘stance’.

I know she is right in this, and I know, furthermore, I am a victim of my ‘cognitive biases’, exactly as David outlines below.

I’ve been prompted into action by the picture of the dead child. He looks just like my boy looks when he’s asleep, you see.

I turned to Peter Singer for this; what he ‘told’ me is more simplistic than what he ‘told’ David (possibly because I don’t have the kind of technical mind to think through all the implications, as per below). What he told me was that we have to use our ‘cognitive biases’, our ethical weakness – in this case, to care for those like us, and to ignore those who are different – to our best advantage. We have to know ourselves, and use that as a basis, a starting point to reach a higher place of empathy and generosity.

So I feel bad about my ‘cognitive bias’, and not bad, all at once. I was asleep to that suffering, and then I woke up.

I was right, I don't know enough about the refugee crisis. Why would she ask me to write about it? I live two kilometres from where I was raised. The most adventurous trip I undertook as a child was to the caravan park next to Mona, for a four-day stay. So I said no, after running in circles muttering Monty Python-esque quips about ‘bravely avoiding confrontation’.

Peter Singer recently published The Most Good You Can Do, wherein he advocates effective altruism, the idea that it is incumbent on all of us (at least anyone with the opportunity to read a blog) to live inexpensively, and to benefact causes that spend the donations in ways calculated to do the most good. In the case of human-centric charities, that means saving the most lives (getting the least bangs for your buck, in the case of war charities). One of the meta-charities he supports suggests that saving a life for less than $5,000 is money efficiently spent.

Among Singer's assumptions are that every life is equal. He introduces a concept known as Quality-Adjusted Life-Years (Qaly). A Qaly is a year lived with no disease burden. Many people would opt for a shorter life rather than suffering. If you would choose to live half as long able-bodied as bed-ridden, for example, then you are giving being ‘bed-ridden’ a Qaly rating of 0.5. Although this kind of analysis of contentious (not least because in a hypothetical scenario people overestimate how many bed-ridden years they would surrender), it does provide a quantitative method for assessing the value of ‘doing good’ Singer advocates (I think) trying to achieve the most Qalys for your donated dollar. I have a long list of issues this potentially raises. For example, effective altruists are directly manipulable by rich sociopaths: I could tell Singer that if he doesn't start eating meat – a significant abrogation of his principles – I will withdraw donations that achieve more than he achieves by being vegetarian. My reservations (and sociopathic status) notwithstanding, his model is the best presently available, and I've tentatively accepted it. Hypocritically, I haven't accepted it as my life process: sold my museum, and downsized my life to avail me the opportunity to do the most good I can do.

Some of the people Singer lauds in his book choose to pursue potentially unethical (or neutral) careers so as to maximise the money they have at their disposal, with which to be effective altruists. I've inadvertently done that in reverse: having made money gambling I felt guilty and tried to do some good. Not much good, Singer would say. A few years ago we asked him to write an essay about Wim Delvoye’s Cloaca (the poo machine). The tone of his reply made it obvious he thought money spent in this way lay somewhere between frivolous and reprehensible; he demanded a ridiculously high fee – which he intended to donate to an effective charity – for his essay. By the seamless logic of his life he had little choice: he had to work out the expectation of his fee (money multiplied by probability of achieving it) such that he maximized the donation he could make. In the event, Delvoye checked him out and vetoed the essay.

In The Most Good You Can Do, Singer analyses the very act of building an art gallery (or at least a wing on one) to see if it's ethical. Unsurprisingly, he finds it wanting. Let me paraphrase his argument (his text is too long to quote; he would be ethically bound to sue me for copyright infringement because he could do good with the money). But first, a digression:

A gambling collaborator of mine worked on the Deepwater oil spill settlement. The compensation process required people to assess their loss of utility for not being able to go to the beach. There is an assumption that they lost something even if they never go to the beach, because they have forfeited the benefit of being able to contemplate going to the beach. The assessors might go somewhere they see as equivalent, and conduct a survey, asking, ‘How much would you need to be paid a year to give up your right to go to the beach?’ For those who never go to the beach, the amount is, of course, not zero. And the average amount enables an assessment of harm done. This approach caused a few problems for those with a superficial overview. The upshot was that there was more economic damage done to people in Florida, where many beaches were damaged by sludge, and there are lots of people, than in Louisiana, where some suffered a huge amount, but fewer through these indirect modalities. In assessing whether galleries are effective, Singer used this type of assessment, and the same caveats – that people overestimate an assessed value to them in a survey – apply.

I built Mona at a cost of around $75m (ignoring the art). It's quite possible that Mona will eventually be profitable, and I could use the profits to fund effective charities. Even if it never becomes profitable, Access Economics have assessed its net benefit to the Australian community to be around $70m per year; those making that money could be donating it to charity, although the portion would be tiny. And the visitors spent their money here, when they might have donated it. So we will proceed, as Singer did, to analyse Mona's contribution as if it were only a benefit to those that visit (and also not those who benefit from contemplating visiting).

Around 350,000 people visit a year. They have all made the choice that it is worth visiting, and some 80 per cent of them actually like it. Many people visit many times a year – they must actually like it a lot. You are a reader of my blog, and are therefore more likely than most to be a fan. You plan to visit this year. But, it turns out, one person a year will be blinded for visiting the gallery (bold art intervention? Likely terrorist attack using chemicals? God hates degenerate art? Peter Singer's rational intervention to maximise charitable donations?). Would you still go if your chance of losing your sight was one in 350,000? Probably you would: you drive (or are driven), and a life is lost every 250 million kilometres on the road. That costs you about a half a year of life expectancy, but you still do it. If you ride a motorcycle everywhere, you cost yourself about a quarter of your life expectancy. A few of you do that. Smoking – and, astonishingly, there are still some smokers – costs you ten years.

But that was one in 350,000. What if one person per day lost their eyesight as a result of a visit to Mona? That's about one in 1,000. I'd say you probably wouldn't take the chance at those odds. So, forty years (say you're halfway through an eighty year life expectancy) of eyesight is worth more than 1,000 visits to a (potentially) good gallery. The capital cost of Mona ($75m investment annualised plus losses) is about $15m a year. So you think that: a year’s eyesight is worth (15m/(40*350)) or more than $1,000. That's kind of obvious. But you also think that the opportunity to visit Mona is worth less than $1,000 a year. That might be obvious, too. Another way to look at it: you think that the benefit that 1,000 people derive from visiting the gallery is less than the harm inflicted on one person being blinded for forty years. That assumes, and Peter Singer does assume, that a benefit to one is exactly 1/1000 of the same benefit to 1,000. Sometimes benefits accrue in a non-linear way. Cities become more innovative a lot faster than their population grows. A thousand people visiting an art gallery are clearly more likely to collaborate than one such person, and – who knows? – they might find a cure for blindness. A doctor visiting Mona noticed that the Rafael Lozano-Hemmer artwork that measures heart-rate could be used to construct a test that was a lot cheaper than the existing one, and he launched it as a product, expecting it to prevent many heart attacks. But let's proceed with Peter's assumption of linearity, for the sake of clarity, and computability.

It's hard to reduce all the potential ways to help and harm to numbers. And that's what you are probably thinking now. This is all so reductionist, and doing good is good, however inefficient it is. As I said, most likely you are a fan of Mona, and that means that you think Mona is good for you, and good for society. Peter Singer's approach, which is the best mechanistic approach we have, draws the opposite conclusion. Are you prepared to put your wishy-washy emotions up against his elegantly reductionist logic? Am I? If you are, can you expect others' morality to be congruent with yours? One thing I can say for sure: it can cost less than $1,000 to cure certain types of blindness. And that's not just for a year, it's for a lifetime. A WHO study of trachoma treatment in Nepal reached this conclusion:

The societal cost of mass treatment per one percentage point decrease in prevalence among 5,200 children screened was 32,400 NPR (ca US$600).

That was in 1998, so in Australian dollars, now, that might be $2,000. The kind of trachoma that likely results in blindness (intense inflammatory trachoma) has an incidence of about 4.3 per cent in Nepal, or 223 cases that will result in blindness. A one per cent reduction prevents 2.23 cases of blindness, so curing blindness comes to – voila – 2000/2.23, or around $900.

$1,000 for a one fifth of a life. Less than $1,000 for sight. Or a good time for nineteen people (0.8*350000)*(1000/15000000)? Should I close Mona down and give the cash to Peter Singer to do with as he will?

If you have been paying careful attention, you may recall that this blog is supposed to be about the refugee crisis. And it is. All this posturing was to create a framework that allows some sort of assessment of what's going on, and how to make sense of the way we react to these appalling events.

Compare: As many as seventy migrants have been found dead inside a parked truck on a highway in Austria, according to police.

With:

Is it the image that made the world react to the plight of refugees in Europe? There were no images from the truck in Austria, but photographs can't capture noxious odours. The last person to die in that truck: is his or her life worth as much as this child's? The reaction to media coverage would suggest the answer is no. Why? (And why, if you were at the beach, is your first reaction to take a photograph?)

From the beginning of 2014 to mid April 2015, 254,000 refugees made it to Europe, while 5,100 died trying. So 2% of attempted entries died (ignoring returnees). So if your Qaly was less than 0.98, then it is worth the attempt. And in war-ravaged Syria, how could that not be so? Further, about 200,000 of 22m people have died in Syria; pretty close to 1%, which means that the damage to your life expectancy attempting to find a better life in Europe is less than one year. Before I did the research I assumed that desperation was the driver of refugeeism, but those seeking a better life in Europe are completely rational. Here's my entrant in the most contrived statistic competition: since about 43% of Syrians smoke, but 27% of Europeans, if the Syrian refugees acquire European smoking habits, this alone would compensate for the risk they engaged by becoming refugees (and therefore, of course, they should really come to Australia, since smoking prevalence is lower than in Europe).

But that doesn't explain why we care about the picture.

I was just talking to my mate Mohammed, and he told me he was going to the pro-immigration rally in Sydney's Hyde Park. I checked out the rally Facebook page and they led with story of Aylan Kurdi, the child in the photo. I'm sure the promoters know what works. But why do we care more about one than one million? And why does Peter Singer's carefully reasoned abject objectivity curry very little favour with the broader community?

When we were first trying to win on horse races I found that our models, taking into account form and jockeys and tracks and breeding and lots of other stuff, did not outperform the public, even though the gambling public is just a bunch people voting with their pocket and then being aggregated (it is, incidentally, a rousing endorsement of the democratic process). It was only when we included the public assessment in our models that we could win. Essentially, the public is better at calculating the odds than we are, except that they make consistent mistakes that we can exploit. One such consistent error is called hot hand bias, which is exemplified by the fact that when a basketball player makes a few three-point shots in a row, everybody thinks he is going to keep making them. He won't; he will make his career average. There is no such thing as a streak. But we believe there is. And the reason we believe there is is evolutionary. One hundred thousand years ago in the African Savannah our ancestors foraged. Foragers were better off returning to where they found food than searching at random, and that provides selection pressure in favour of hot hand bias. Bees also exhibit hot hand bias. It's a good thing near the hive, or in the Savannah, but it's a bad idea when betting, or thinking. Mistakes of this type are called cognitive biases. And our appalling treatment of refugees, I think, results from some of these cognitive biases (the conflicts that cause people to flee may also have their genesis in some of these biases).

Not all cognitive biases once conferred an advantage. They are heuristics, short cuts which allow speedier processing of complex data. Mark Changizi, in his wonderful The Vision Revolution, points out that the more complex our environment, and the more novelty we face, the more compromises our neural processing has to make. We live a little bit in the future because the brain takes time to do its processing; but if we live too much in the future, our neural forecasts are more often incorrect (that's when we are deceived by an illusion). So we want to be as quick as possible to make our future forecasting not too distant (it seems to be about half a second) and short-cuts, mental rules-of-thumb, are required to get the job done on schedule. These approximations, in my opinion, account for most of our cognitive biases. The need for dealing with novelty also accounts for our large brains. We could be smart more slowly, and do it with smaller brains, if we didn't have to deal quickly with situations that we have not, hitherto, encountered.

So let's look at a few of these things, these biases that fuck us up, that make difference repellent, and bigotry and selfishness attractive. If it could be demonstrated that these traits really are the result of cognitive bias and thus induce systematic error, then correcting for them might set us on a better path, in the same way that, when betting on a horse race, correcting for biases allows a more accurate assessment of the odds.

We identify with those in our in-group, and often reject or even despise outsiders. Prior to the evolution of speech the maximum number of individuals that could cooperate was about 150. Cooperation is useful: a pack of hyenas can bring down a lion, but coordination of large groups is difficult. Chimps attack other chimps as a group, and bonobos in-group activities are well known. Speech enabled larger groups, but that means there may be too many members of a group to remember. An un-counterfeitable way to recognise friends and enemies is needed. Race provides an easy one. It's hard to fake the colour of one's skin (or gender, but in-groups and out-groups often have the same gender distribution). Religious affiliation, and political persuasion, and sports affiliation, and parochial leaning are easier to falsify. Unless we commit to beliefs that are so ridiculous or heinous no one would voluntarily fake them (virgin births, Nazi atrocities, team song bonding, Australian flag bikinis). All this means that we commit to the in-group at the expense of the out-group. That might be where headlines like this come from:

Refugees in Europe: Christians welcome - Muslims keep out...

Another bias, out-group homogeneity bias, is relevant here. The name says it all. We tend to assume that the groups we know are diverse, but that outsiders, within their group, are all the same. A few Islamic terrorists make stereotyping easy, provided we see them as all the same. In fact, Islamic countries often have low homicide rates (Iran lower than the U.S., Saudi Arabia lower than Australia). This type of misperception recently gave Donald Trump the ammunition he needed to stigmatise undocumented Mexican immigrants in the U.S.

If you are sceptical about in-group identification and out-group demonization, consider some studies conducted by Henri Tajfel. He assigned random subjects in his studies to groups, in some cases by coin toss. Those involved quickly started accepting the group they were assigned to as objectively superior. This, despite the fact that the experimental subjects knew they were allocated to their group at random.

I'm starting to bore myself. But in the unlikely event that anyone is still interested by this point I'll plough on.

Another significant group bias is the ultimate attribution bias, which Wikipedia explains so well that I will be lazy (as I say, I'm starting to bore myself) and lift the text:

Ultimate attribution error arises as a way to explain an out-group's negative behaviour as flaws in their personality, and to explain an out-group's positive behaviour as a result of chance or circumstance.

Relatedly, just-world bias is the view that those suffering fortune or misfortune brought it on themselves because the world is fair. Amongst other things, this yin and yang view enables rich fuckers like me to avoid donating to charities by believing that they worked hard for their money and they deserve it. The poor, of course, also deserve their lot (and that's not a lot). It also contributes to explaining why seventy-one corpses in the back of a truck doesn't cause uproar.

But it doesn't explain the photo.

Most cognitive biases allow us to get through analysis quickly, but some also allow us to avoid the uncomfortable state known as cognitive dissonance. When we have multiple stimuli that are contradictory, we quickly assume that the information which are most aligned with our personal biases is correct. When exposed to suffering, externalising by locating it in an undervalued out-group is easy. Except when that stimulus is baldly biological, and triggers those protection mechanisms that evolution has amplified within us. We need large brains, in part because we evolved intelligence, and in part because of the required speedy response time to novelty. But kids need to pass through the birth canal, and that means their heads can't be too large. Fifty days ago I watched my daughter, Sunday, emerge for the first (and only) time, and it reinforced my notion of how risky birth is. But small heads mean undeveloped babies, and that means a long childhood, and a great need of nurturing. Children are cute. It must be so, because they are hard to keep alive, and so we need incentives. A suffering child, therefore, sets up a very strong cognitive dissonance. And a dead child doesn't offer mental exclusion as a solution because the dominant biological impetus is protection. There is no way out but remorse. One dead child makes us all responsible.

What about a million anonymous dead children? The advertisements for charities tell you how many children die each minute of preventable causes, and despite the good intentions of Peter Singer and others like him, we do nothing. It isn't that the problem is too large. The lack of direct exposure allows us plausible cognitive denial. We can resolve our dissonance by ignoring the stimulus. The children aren't right there, so our biases can be employed to save us from suffering. But that only makes those who we could have helped suffer more.

In the ten minutes it has taken you to read this far, seventy-five children have had their lives ended by preventable causes. And three hundred people have become refugees. Now let's have a beer, or watch reality TV, while our biologically biased brains decide that it ain't so. See if you can remember these numbers tomorrow.

Hansie Cronje, the South African cricket captain who fell from grace after taking bribes (and who later fell from space), had WWJD tattooed on his knuckles. This stood for, ‘what would Jesus do’. I doubt Jesus would have taken the cash. Let's contemplate, for a moment, WWPSD (what would Peter Singer do)? The strict application of his principles might suggest he would ignore the plight of refugees, since it's cheaper to save lives that are more directly threatened by disease, or starvation, or nature. It costs more than $5,000 to save a refugee. The off-shore detention centres (prisons?) that Australia employs as staging posts to sending the suffering home to suffer more cost more than $100,000 per year, per person (inmate?). Of course, those who get into the community probably pay their way. In fact, the very people with the balls to take on such high-level risk might be the ones who could get things done in a community. Perhaps it is cheaper to allow refugees in than to send them back. If each new resettled refugee contributes to the community, does it matter if it opens the floodgates? Factoring the long-term return on investment, allowing refugees to settle reduces costs to below the Singer criterion of efficiency. That is: the money must do the most good it can do.

James Newitt, a Tasmanian artist, gave people in the streets of Los Angeles a dollar for their story. He got his money’s worth. Here’s one:

I left Africa because I wanted to go to Europe, because I had dreams. So I went. I never had enough money to leave Cameroon directly to the US, so I left Cameroon and went to Nigeria – the neighbouring country – and worked there for a couple of months, and from there I went to the next country – Niger – and I worked there for a couple of months, and to Morocco, and from Morocco I went to France. From France I saved enough money to finally come to America, my final destination. I've been here for three-and-a-half years now, so you count the three-and-a half years back and I was doing that journey.

I used to think that maybe it was different, you know, money-wise. I know I can make money but it seems to be more competition, you know? Not that I'm discouraged, you know I'm still just working hard to make it like everybody and it's just a matter of time, I just have to keep working hard.

We are all machines processing stimuli. But we are slightly lop-sided machines, and we pick things more easily on the side (geographically or socio-politically) nearest to us. (This is not just a metaphor: right-handed bias is prevalent. The word ‘sinister’ comes from the Latin for ‘left’, and dexterous, from ‘on the right’.) Our short-cut cognition leads to error, and that error leads to persecution of those most different from us, even if the difference is arbitrary. With effort, we can counterbalance and correct. Most of what I've presented in this unwieldy blog I'm not too sure about. But I know this: with effort comes understanding, and with understanding, tolerance.

Sincere apologies for the mum-and-pop psychology.