Former software entrepreneur, I now advise businesses, play the stock market, and attempt non-fiction writing
bfinn
I’m not an expert, but I assume (from a glance at the second paper) this is because the 1%-59% is a cost (opportunity cost), not a value of a life year as such; i.e. in a very poor country you can extend a life by a year for as little as $3, maybe with a vaccine or micronutrient supplement. Actually that seems an order of magnitude too low to me; but nonetheless, it’s a great deal!
Indeed, though if working from existing 0-10 life satisfaction scores I don’t think it’s plausible that those who responded below 5⁄10 thought they’d be better off dead. (Maybe those responding below say 2⁄10 would.) Otherwise suicide rates would be far higher.
(But indeed some kind of calibration of death and worse-than-death states is needed more generally. E.g. it concerns me that almost all the bad in the world may be located in extreme pain that is hugely underweighted, and so almost all efforts to improve the world may be missing the point.)
Hi Michael
Thanks for this. (I hope my summary of value of life was mostly right!)
Yes I haven’t really given any thought to what the best way of handling the situation would be, or would have been. Clearly complex given that there are sociological/political constraints too (e.g. how would the public react if x% of them die in a new dramatic way—as contrast with, die routinely from seasonal flu or traffic accidents).
It seems to me a global recession could reduce income/employment & hence quality of life without having much effect on life expectancy. For I’m not sure the last recession had much or any identifiable effect on it; growth in life expectancy has slowed since 2008⁄9, and I asked Paul F about this in a comment below one of his recent articles (which I have indeed been following), but he attributes it to other things. So I wonder if only looking at saving lives is going to miss most of the damage.
I think Paul F is effectively combining quality with quantity of life in his dollar numbers, and converting to whole lives lost as a convenient way to express it, but not completely sure. After all, dollars can be spent on quality or quantity of life.
I look forward to reading your analysis in due course!
Coronavirus: how much is a life worth?
Belatedly—thanks. I’m not sure what to make of this. That survey is quite large (30-50,000 people p.a.), so much larger than Eurobarometer, though smaller than ONS (around 150,000). Eurobarometer shows a large rise 1996-2016 (7.19 to 7.74/10), and the later-starting ONS shows a smallish but non-negligible rise 2012-2016 (7.45 to 7.67/10). Possibly again the question wording might have an influence.
But 5.2 to 5.3 is a rise, even if (statistically?) insignificant. It’s unfortunate that the paper cites other surveys (in other countries) which confirm its claim of no effect, but doesn’t cite these other UK surveys which suggest the opposite.
Since the ONS survey is much the largest, and also kind of confirmed by its findings on happiness (i.e. positive emotions), perhaps the reality is that there has indeed been a substantial rise since 2012, but only a small rise, or perhaps none, before that.
On a separate small point, I think your probability estimate for ESP is too low, for two reasons:
Firstly, it is a taboo topic (like UFOs and the Loch Ness monster), which scientists are therefore far more likely to dismiss from a position of ignorance, or with weakish arguments (e.g. ‘it lacks an explanatory mechanism’, ‘much of the research methodology is flawed’, or ‘some of the research has been on fraudsters’ - hardly disproof). Few skeptics have domain expertise, i.e. of having conducted or investigated research in the area.
Secondly, ESP covers quite a range of rather distinct phenomena. Only one has to be right for ESP to be true. And I’m not sure that all would require completely novel scientific principles (e.g. unknown physical forces); and the fact that our understanding of physics has gaps, and our understanding of consciousness certainly does, may well leave room for some form of ESP to be compatible with current science (not that that is essential).
Great article. I’m very late to the party in reading it & commenting, but I hope not too late to be of use!
I have three further reasons for epistemic immodesty in some circumstances. They all involve experts, or those who follow their advice, being overconfident about the experts’ relevant knowledge. (Though I note your comments about debunking experts; none of these arguments show an amateur is better than some other, probably small, set of experts who have taken these considerations into account.)
HIDDEN PREFERENCES
You mention that expert views aren’t relevant in matters of taste, i.e. preference. However, expert views are often based on non-explicit preferences, which some experts may even be unaware of themselves.
To start with a clear situation where preferences are involved: If I’m looking for a house to buy and trying to decide which one to choose, I may well consult experts in the field, such as an estate agent (realtor), a mortgage advisor, and an architect (if it may need building work). They may advise that I can’t afford a house more than $x, or it will cost $y to do up, etc. But even with all their expert advice, this won’t necessarily settle the matter of which house to buy, because I also have to *like* the house in question, want to live in that area, etc. So my decision involves both expert factual opinion and my personal preference; and I am the sole expert on the latter.
Now to take a less clear situation, currently topical in the UK: Brexit. Despite years of debate about this, which often includes discussion of experts and whether they should be trusted, I don’t think I’ve heard anyone state clearly that it too mixes expert opinion and preference. Most economists say Brexit will harm the economy, and most voters opposed to Brexit assume this simply entails Brexit is a bad thing. But of course the issue is not only about money—various other considerations are involved (e.g. self-determination) - and the trade-off between these is a matter of preference. Some people with unusual preferences may have coherent reasons to oppose Brexit (e.g. I spoke to someone who voted based on the fact that animal welfare is taken more seriously in the UK than most other EU countries, a consideration she regarded as more important than the economy). So this is an example of a ‘semi-hidden’ preference—one where many people assume expert opinion is a silver bullet—perhaps including the experts themselves—and overlook the element of preference.
A different example is government guidelines on alcohol consumption. In the UK men are advised by experts to drink no more than (I think) 14 units per week. However, this advice is based on a trade-off between health and pleasure: if you really enjoy alcohol you may be happy to exchange a risk of significantly reduced health or longevity for drinking much more than 14 units. This trade-off is a preference, which the experts have made for you. (And AFAIK the trade-off they chose is arbitrary, not even based on research into say average preferences.)
Other topics may include preferences so hidden that even the experts are hardly aware of them. An example in EA would be the use of DALYs and QALYs (disability/quality-adjusted life years) as human welfare metrics in assessing charities & interventions. Some who work with these metrics may overlook, or perhaps be unaware of, their shortcomings. DALYs and QALYs as currently defined assume that no condition is worse than death—which is inconsistent with the existence of suicide and euthanasia. When ordinary people are surveyed, their views on this vary widely—some taking the (perhaps religious) position that nothing is worse than death, and suicide/euthanasia should never be allowed, whereas others have no problem with the idea of suicide/euthanasia to escape prolonged untreatable agony, for example. So the mere use of these units involves tacitly taking a position on this, i.e. a hidden preference. A resulting expert view that X charity or intervention is better than Y is therefore partly objective and partly subjective; the expert themself may overlook this fact, or even (when involving technical philosophical issues) be unaware of it.
Other unstated assumptions are widespread in EA, e.g. that saving lives is a good thing (even though the world may be overpopulated), or that the prevention of merely potential future humans by mass extinction is a bad thing (even though contraception is fine).
In such cases, a non-expert who identifies such a hidden preference that they don’t share may well have good reason to disregard the expert opinion.
SHAKY FOUNDATIONS
Relatedly, there is the issue of core assumptions that are largely unquestioned within a scientific field. A classic example is induction: physics assumes that just because in the past things seem to have behaved in a regular fashion, they will continue to do so. This is the basis of the belief in physical laws (and other laws of nature). Philosophers have long questioned this assumption; there really may be no reason to assume the sun will rise tomorrow, or that the speed of light was the same yesterday or a million years ago; which undermines all kinds of experiments and models. I expect many physicists are only dimly aware of this, know little of the arguments involved, and perhaps regard it as a quasi-theological debate not worth serious attention.
As with DALYs and QALYs, core assumptions like induction are often shaky, and the shakiness is often only taken seriously (or even known about) by those outside the field, e.g. philosophers. Indeed, articles of faith are often left unquestioned by true believers, lest they turn out to be an Achilles heel, and (mixing more metaphors) the whole edifice is built on sand. To question foundational beliefs may be heresy.
So an amateur outsider may well be more aware of such problems than an expert in the field; and may therefore be justified in using them to dismiss expert opinion, or at least, to take it with a big pinch of salt.
NARROW EXPERTISE
Many experts are only expert in an extremely narrow field, yet may be assumed to have a broader range of expertise (and some experts may also believe this themselves).
Apologies, but the clearest example I can think of is myself! At one time I was one of just a handful of world experts in an extremely narrow field—the music notation software industry. (As I owned a company in this field.) My knowledge was extremely in-depth—I had spent years coding this kind of software, knew endless obscure feature requirements, knew all about the market, wrote manuals and brochures, etc. Yet in other respects I knew less than many amateurs. I had never used (and hardly even seen) any music notation software other than my own company’s. I knew even less about other types of music software (e.g. sequencers), used by millions of people, often my own customers. So I was a world expert in a very narrow field, yet an ignoramus both in aspects of my own field, and in very close fields.
The same is presumably true elsewhere. Amateurs may know as much as a world expert who is only slightly outside their very narrow field, or even on topics within their specialism. And at least occasionally, experts are unaware of their ignorance on these things. That is, they may make the same false assumptions as others do about the breadth & depth of their expertise.
(An example: the book The Oxford Companion to the Mind is an encyclopedia edited by the eminent psychologist Richard Gregory. Some of the entries in the original edition are by Gregory himself, despite dealing with philosophy of mind & metaphysics, topics evidently outside his expertise. They are amateurish, making confusions that would embarrass a philosophy undergraduate. Even the blurb on the cover jacket casually conflated ‘brain’ and ‘mind’ in ways only an ignoramus would do. When I was a philosophy student I was so astonished by this I almost wrote a letter to Gregory suggesting he get someone with domain expertise to rewrite his entries.)
Actually there has been one change in method—in 1998 it was made illegal to sell large quantities of paracetomol, to make casual suicide harder. The suicide rate has been falling since but there was no sudden drop, so I’m not sure we can attribute much effect to that.
For comparison, when converted to a 0-10 scale, the Eurobarometer survey shows a rise by 0.7/10 between 1999 and 2019.
Thanks for this.
The ONS suicide data is from 1981, showing a decline by about a third by 2007 - pretty big, and I’m not aware that any popular suicide method became less available in that time.
(Unlike e.g. coal gas used in ovens, once a popular suicide method but it was phased out in the 1960s and 1970s leading to a suicide reduction—I don’t think coal gas was available thereafter as the last plant closed in the late 1970s. And guns have never been generally available in the UK.)
The methods used have apparently changed popularity in recent years; hanging/suffocation/strangulation and poisoning are the most popular:
But I’m not sure why they have changed other than ‘fashion’. It could be the case that some of these methods are significantly more effective than others which could affect the statistics, but I doubt by this much.
Also I’m not aware that suicide has changed in acceptability in the UK in recent decades. It was never considered acceptable (unlike say in Japan).
So I’m still inclined to regard suicide as a better proxy of extreme mental health problems than anything else. (That said, I’m not an expert at all in this area.)
Oddly, Britain has never been happier
Thanks, very useful. The World Happiness Report data (from Gallup World Poll; I’d seen the figures before but couldn’t find more info) does show a rise of about 0.25/10 over the period—about one-third of the rise in the other ONS and Eurobarometer results (when Eurobarometer converted to a score out of 10).
I suspect the difference is in the wording of the question, which defines 0 as ‘the worst possible life for you’ and 10 as ‘the best possible life for you’ and asks where they are now. (It doesn’t mention the word ‘satisfaction’). Whereas the ONS question is ‘Overall, how satisfied are you with your life nowadays?’ where 0 = not at all satisfied and 10 = completely satisfied (Eurobarometer has similar wording AFAIK). They’re rather different, and personally I find the ‘life for you’ wording (which I’ve come across before) a bit confusing, as does 10 mean ‘doing as well as I could given my abilities & circumstances’ or ‘having the best life I can imagine (with no restriction)’?
[ADDED] To put it another way, the Gallup question seems to be asking people to compare with some unclear external scale of what’s possible for their life (in the real world? in a fantasy world in which they could be a rock star or Bill Gates?), rather than how satisfied they feel about their life (a more internal scale of feelings). If they’re comparing where they actually are with what might be possible in a fantasy world, it’s not so surprising it doesn’t go up much, because reality rarely approaches fantasy.
Without deciding which survey has a ‘better wording’, if any of them shows a substantial effect then it suggests something is going on in whatever that question is measuring.
There is a lot of happiness data available, including from the ONS, but there is a tendency now to prefer life satisfaction because it’s more stable (happiness varies with the weather and day of the week) and more all-encompassing. So I didn’t look into it. Though the ONS happiness data show the same trend over time as satisfaction.
(On a lesser point, I don’t know how large the Gallup poll is but I imagine, like Eurobarometer, it’s a few thousand people per country. The ONS is over 150,000 so very reliable. That said, aggregating multiple years removes the sample size problem.)
Thanks. It’s a very cool site. I couldn’t figure out how to find the source, but never mind. The figures for England (the vast majority of the UK population) do indeed show a slight decline so at least that’s consistent with my hunch, though not explaining much of the misery fall (and it shows anxiety as completely flat). Possibly there was a bigger mental health improvement in the 1970s/1980s. More likely these figures measure presenting or treatment rather than pure prevalence, so hard to conclude much.
On the aging thing I’ve just done a rough estimate. The median age increased about 3 years between 1999 and 2019, and as it’s around 40 which is before the mid-life crisis (life satisfaction being U-shaped as you age, the lowpoint around 50), an age increase of 3 years would if anything lower happiness. (Of course it would be a mix of some going down and others up depending on their ages, but the overall effect would presumably be down. Incidentally the bottom of the U-shape hasn’t noticeably got older as the population aged over this period.)
But the effect is small anyway—the median, getting 3 years older, would lose about 0.07/10 points life satisfaction (when expressed as a score out of 10), which is only about 10% of the 1999-2019 change, as well as in the wrong direction.
More consistent with the blue line falling due in part to better treatments. (If they’re significantly better; may well be more widely prescribed anyway.)
Do you know how this data was gathered? The prevalence of mental health problems is presumably hard to determine (see my footnote 7). I’m inclined to believe the falling suicide stats as a proxy, as they’re objective.
The aging did occur to me, but without doing the numbers I doubt it would have an effect that fast on happiness, and certainly not that fast on misery.
Re immigration, research shows immigrants mostly take on the happiness of the country they move to, partly retaining their previous happiness. And as most recent immigration has been from less happy countries (eg Eastern Europe), I’d expect the effect to be a small fall in happiness not a rise. Though again without doing the figures I doubt it would be big enough to affect the general shape of the graphs.
Good article. Various things you mention are examples of bad metrics. Another common kind is metrics involving thresholds, e.g. the number of people below a poverty line. Since they treat all people below, or above, the line as equal to each other, when this is far from the case. (Living on $1/day is far harder than $1.90/day.) This often results in organisations wasting vast amounts of money/effort moving people from just below the line to just above, with little actual improvement, and perhaps ignoring others who could have been helped much more even if they couldn’t be moved across the line.
Very good article. Re marriage, and also the Easterlin Paradox:
Re fig 1, marriage, more recent research shows the apparent reversion of satisfaction to singlehood level (or less) a few years after marriage actually just results from not controlling for age properly. Specifically, from the U-shaped happiness curve during the life course - people tend to marry before middle age, so get less happy as they approach middle age, not because of the marriage. (I think this may be mentioned in the recent Origins of Happiness book.)
Re fig 3 for UK, actually if you look at a longer time-series e.g. Eurobarometer (since 1973), UK life satisfaction has gone up significantly—from around 7.1 in the 1970s to 7.7/10 in 2018 (after transforming to 0-10 scale); and rather more since the only older survey I could find, 5.7/10 in 1948.
Also re China, you say elsewhere ‘it’s SWB seems to have gone down been 1990 and 2015, even though per capita GDP increased by 5 times’. I haven’t read the report you refer to, but I the World Value Survey on happiness (as opposed to life satisfaction) shows the proportion of Chinese who are happy rising from 67% in 1993 to 85% in 2014. (You can check it on the second chart here: https://ourworldindata.org/happiness-and-life-satisfaction# )
So maybe the Easterlin paradox does not exist, or at least, is limited.
These are important topics IMO.