The good, the bad and the effectual: From cost-efficiency to cause-effectiveness in the fight against malaria
In my last blog post, I made the case for making the ‘fight against malaria’ a key global priority, even in troubled times. Generally, it is among the deadliest and most neglected yet still solvable causes of premature deaths, globally. Along with several ‘effective altruists’, I promulgate the work of the Against Malaria Foundation (AMF) as one of the most effective charitable causes to this end, through their procuring, distributing and monitoring the use of preventive bed nets, and urged everyone to support us in this endeavor (primarily through donations to the AMF).
Still, in the process of doing research for the campaign and reading up on the issue of mosquito-borne parasites and other diseases, have caused a few sleepless nights (and plenty of hours of procastrination), making me reflect more critically on the overall approach of ‘giving effectively’, so successfully promoted by the Effective Altruism (EA) movement.
Overall, I’ve come to the conclusion that the key concerns around zoonotic and mosquito-borne disease are not (causally) rooted in animal behavior, or even evolution.
It’s us.
And not only does this relate to how we are neglecting ecosystems and their ecological services, but it’s rooted in the way the EA community has approach these ‘ecologically-relevant’ challenges.
(Two caveats at this point: I do not intend to criticize the EA Community, to which I’m not formerly affiliated or familiar, and where discussions have progressed through other (internal) channels. Also, this is not an argument against the overall mantra and principles of ‘giving what we can’ for ‘doing good better’. Rather the text is better thought of as a truthful attempt at constructive dialogue for improving and addressing some serious clouds on the long-termists’ horizon.)
To me, the question isn’t as much whether there are more effective measures than those practiced by the AMF and promoted by GiveWell or GWWC, the Global Priorities Institute or the EA community (we can assume that, in theory, there are always more effective ways of doing things, e.g. through Bayesian updating of our credence and cumulative, experiential learning). The question is rather why the obsession with doing the greatest good very efficiently in the short run seems to lead us to consistently recommend a class of measures that are ineffective in the long run.
To paraphrase the American environmentalist and design pioneer Bill McDonough: ‘If you do the wrong thing efficiently, it’ll turn bad effectively’.
Throughout this blog post, I will use the case of malaria prevention as an illustration, yet my points have implications for a range of ‘global priorities’ and world problems, particularly those pertaining to (the state of) our natural environment. Towards the end (after some initial references to the science behind mosquito-borne disease-prevention and a lengthy thought experiment), I will seek to provide some lessons to be drawn for the EA community and argue for how protecting wildlife could be thought of as an effective lever for protecting future wellbeing.
The priority and the cause: Preventing premature deaths from malaria
Anyone with just a toe inside the EA-movement will be well aware of the AMF, its mission and how it works to reduce malaria-related deaths, particularly in Sub-Saharan Africa. It has remained one of the most talked about and promoted charity and charitable cause for more than a decade.
Its historic legacy speaks for itself: The AMF estimates that without anti-malaria bed nets the annual death toll from malaria would be around 1.2 million people (many children below the age of 5), compared to the roughly 600 000 today.
Furthermore, the work of the AMF is considered one of the by far most effective and cheap means of preventing (premature) deaths. For instance, GiveWell (that provides the most trusted vetting of charitable causes for the EA movement) concludes that “it costs between $3,000 and $8,000 to save a life in countries where GiveWell currently supports AMF to deliver ITN campaigns.” This means this measure is nine to 23x as cost-effective as providing direct cash transfers (GiveWell’s benchmark for comparing different programs). In their summary assessment, they explain the cost effectiveness of AMF arise “because ITNs are cheap to purchase and distribute (roughly $4 to $6 per net) and significantly reduce malaria mortality, among other benefits.”
While the work of the AMF has had other beneficial (multiplier) effects as well, such as popularizing and researching effective means of distributing insecticide treated bed nets (ITNs), that have been crucial for protecting thousands of vulnerable people in some of the most impoverished communities of the world, the question remains: Why is the problem of malaria getting worse, globally, and what’s the role of ITNs in reversing the trend?
As I laid out in the previous blog post, malaria fatality has been going down since 2000, over the last few years at least in no small part because of the distribution of ITNs from organizations such as AMF.
However, as mosquitoes are becoming better (evolutionary) adapted and resistant to treatment, fatality has been going up, and the mere fact that the mosquitoes are carrying the disease to new and more populated areas, is threatening to undermine the rapid progress made over the last 20 years. Etnomologists at the University of Oxford have assessed that a particular new species of malaria-carrying mosquito, Anopheles stephensi, continued expansion into urban areas and countries interior puts an additional 126 million people at risk of malaria. Public health experts describe this as a particularly malaria-vulnerable population, as people in urban dwellers don’t have immunity from repeated prior exposure and may therefore be more prone to severe illness. Further, urban health care workers are often inexperienced in diagnosing malaria-related illnesses.
What is more, bed nets are not really a means of protection from this kind of mosquito. As the stephensi mosquito prefers artificial breeding sites, such as drainage ditches, water tanks and trash bins, exposing humans not only in bed, but also during day-to-day activities. Further, compared to other species, the stephensi is accustomed to human husbandry, forming habitats in goat, chicken and cow sheds, exposing farmers and rural inhabitants outside during the day.
Djibouti, one of the first countries to discover the presence of the new species, was close to declaring malaria eradicated in 2012, when there were just 27 cases identified. The year after the first stephensi cases were recorded in the country, the amount of cases rose to nearly 1,700. The number has kept going up each year since, and in 2020, more than 70,000 cases, and 190 deaths, were recorded.
Finally, and most crucially, the currently used insecticides and widely available pesticide products won’t stop this species of mosquitoes. This new breed of insects are largely resistant to the insect-killing chemicals currently on the market.
Necessary, yet insufficient
While many scientists have been sounding the alarm bells for some time, this development might work as just another ‘call to action’ for effective altruists. It might lead some in the EA community push for AMF and other (Western) charities to do more, produce better, and distribute new types of ITNs across ever wider regions. Once more research has established the anatomy of the new mosquito species, new insecticides could be devised to treat nets that will repel also this species. Nets can be redesigned to protect larger areas, such as cow sheds.
After all, if you just look at the historical dollars-per-life save performance and its proven success at preventing (excess) deaths, it seems only reasonable to assume this is the most effective measure also in the future.
Such an effectiveness-oriented base instinct might have been true if only we lived in an ergodic world. Alas, we don’t.
As someone who has supported the work of GiveWell, EA, the AMF (and therefore, indirectly, the against-malaria cause) for a range of years already, these new developments have made me question our approach more fundamentally. Without criticising the important work of the AMF, doubting the need to keep implementing preventive measures (such as mosquito nets), or ending up in a kind of pessimistic nihilism (as do so many EA critics), I believe we need to update our map as the terrain has shifted (metaphorically shaking the very ground we’re standing on). Let’s recall one of J.M. Keynes’ classic quotes (and indeed a basic Bayesian principle):
“When the facts change, I change my mind — what do you do, sir?”
I believe there is due reason to go back to the drawing board and re-assess our approach to ‘cost-effective’ means of defeating malaria. To me, the question isn’t simply: why are not things getting better (faster)? It is rather: Could what we do somehow make things worse? And if so, why?
Protector of the realm
Let’s start with a thought experiment that may allow us to appreciate the problem anew. Imagine you were called on from another planet to help resolve humanity’s greatest challenges and was told for the first time, just having arrived at Planet Earth, that a small mosquito-borne parasite, known in earth-tongue as malaria, was the main culprit of premature deaths among children on earth. How would you respond? That a seemingly miniscule creature such as the mosquito could cause such harm to the creature seemingly at top of the food chain, would undoubtedly perplex you. Next, you would probably start by figuring out why, or at least how come?
Once you are up to speed on the historical context of malaria-induced disease and death among earthlings, particularly among the youngest population that have little to no immunity, you turn to the question of why it has been worsening over the last years. A scientist specializing in this kind of mosquito-borne disease could tell you that the proximate cause for a growth in malaria-related disease is the movement and evolution of malaria-carrying mosquitoes. At once, you would realize that the problem is not only tied to the anatomy or lifestyle of this particular family of insects, but also related to the lifestyle of humans.
The proximate cause of the movement and evolution of new species of mosquitoes is in fact human-induced. A new class of anthropogenic stress seem to have become the main driver of ‘natural selection’ on earth, affecting not only the mosquito genera but almost the entire animal kingdom. Today, across a wide range of localities, the main cause of evolutionary selection pressures are human-induced pressures leading to greater phenotypic plasticity (genetic adaptation to new environmental conditions). The consequence of such a view of evolution (or evolutionary trajectories) may, highly simplified, imply that even short-term changes in living conditions (i.e. environmental change) may result in irreversible changes to behavioural patterns over the long-term.
In the case of mosquitoes, the ability of the new breed (stephensi) to reside and thrive in and among urban settlemens, may cause an irreversible change to their habitat, mating and (cruically) eating patterns.
Today, most scientists, from climate scientists to biologists and (ecological) economists agree that primary (root) cause of this increasing (evolutionary) stress largely pertains to human activity, such as the urban sprawl, intensive land-use and forestry, as well as the encroachment of natural ecosystems by extractive industry.
Even if this is perhpas perplexing (or even mundane) to an other-wordly alien, let’s assume you accept the basics of earthly evolution and natural selection. To fulfill your role as a true ‘alien protector’, you next move to the question of how this trend may be reversed, you might again turn to the scientific community. Or rather, the range of scientific communities dealing with subjects relevant to this issue. You would soon discover that different communities would provide you with different, most likely divergent (policy) recommendations.
We may compare their advice from the perspective of risk, namely the risk of harm from mosquito-borne disease. After all, the presence of mosquito is not a problem in itself. Neither is the transmission of bacteria, parasites or disease as long as there exists a general level of immunity and/or effective treatments. We may consider risk as resulting from a configuration of varying degrees of exposure (to harm), vulnerability (for harm) and hazard (severity of harm).
Some policy experts may recommend you get to the direct ‘source’ of the risk. Here, you would target the immediate interjunction, where disease is transmitted from mosquito to humans. I.e. they recommend you target policy interventions at the people who are in contact with mosquitos. You try to either make humans avoid the mosquitoes directly or protect humans indirectly, e.g. through protective nets, when they are exposed. In this scenario, humans are still exposed to mosquito-borne disease, yet less vulnerable.
Others will recommend you target the mosquitoes’ (freedom of) movement. To avoid human-mosquito interaction, you would try to minimize such intermingeling; you could try to clear an area of mosquito or, once the mosquito breaches a physical (or metaphysical) boundary implement drastic mosquito control measures. In this scenario, humans become less exposed to mosquitoes, yet still highly vulnerable to the disease.
Yet another group will recommend you target the mosquitoes’ anatomy. Here you would be advised to utilize advancements biotech and chemical sciences to intervene with the mosquitoes genetically. This means that the mosquitoes either become immune/resistant to the disease-infected parasites, or otherwise avoid becoming carriers. One particular method called ‘gene drive’ has already been experimented with in labs, as presented here.
In this scenario, the mosquito becomes less hazardous, even if humans are equally exposed and vulnerable to the mosquitoes themselves.
Still, most public health experts would caution that there are few quick fixes available to protect people from stephensi; the most effective measures available at scale — better housing and infrastructure, and more efficient municipal government — require significant investments, commitment and time.
Let’s say next, after consulting these medically oriented scientists and public health experts, you turn to an economist. Economists, after all, are the stewards of public purses, budgets, and accounts, which can tell you something about the cost-effectiveness (i.e. what each measure costs per life saved) and financial impact of your interventions.
Any economist with even just the elementals of microeconomics in her course portfolio (surely beyond any alien to question) will tell you that the intervention with the greatest benefits (typically denominated in utility per unit of prevention measure) at the lowest cost (typically denominated in dollars per unit of prevention measure), should be prioritized. The economist will probably start by telling you that these measures have a great spread in terms of relative cost-effectiveness.
For instance, it is fair to assume costs (in dollars per unit of measures) to be greater ex post than ex ante (your alien translator might intervene at this point to explain that this particular, economist-mused Latin prose indicates that an investment made before an event where people are in desperate need of aid, e.g. a natural disaster, will be cheaper than investments made once everyone are in acute need of aid). Humanitarian logisticians typically talk of the 1/6 rule, or the observation that costs (e.g. of logistics and supplies) typically increase six times after a disruptive event. So, the economist first advice will be to try to anticipate what detrimental effects will happen, before it happens.
Next, she tells you that the ultimate decision-maker in human societies are individuals. They have a set of preferences, such as not dying and staying healthy, and an ability to calculate different utility maximizing choices based on these preferences. The individual will therefore know what to do and, in the aggregate, do what is required to protect him/herself. So, the second advice is to avoid investing in broad or macro-level structures, and rather target individual behaviour or prioritise interventions the individuals can themselves implement.
Third, the economist might tell you that in human consciousness, the present is worth more than the future. In purely monetary terms and accounting language, this is indicated by the time value of money, or rate of discounting, and asset loss of value, or rate of depreciation, both directly correlated with time. This implies that real assets are to be depreciated year by year, and that the (far) future value of investments made today, are discounted (reduced) to present value. The same principle may be applied to risks or costs: current costs and hazards are more important than future downside risks and vulnerabilities.
So the economist’s third advice and conclusion: If measures pay off first long into the future, their benefits should be discounted (though not ignored) to present value metrics and weighted according to present-day concerns.
At this point, fellow earthly economists might alert the alien that this represent particular and highly stylized advice, not representative of practicing economists advising these kinds of (non-profit maximizing) projects. Still, an observant alien will notice that this approach is what is thought to young economists (and future project advisors) at almost any university across the world and promoted to governmental advisory bodies by prominent academic economists, such as the much cited Nobel laureate and climate economist William Nordhaus.
(Side Note: There are indeed great debates ongoing within a broad tent definition of Economics, many of which have led to consequential disagreements and even distinct branches of the discipline. Two highly consequential debates that might be mentioned in this context is that between Norman Borlaug and William Vogt within agroeconomics, as brilliantly documented by Charles E. Mann, and that between Friedrich Hayek and Gunnar Myrdal in development economics, as brilliantly documented by Bill Easterly.
These debates and academic quarrels may seem pretentious and irrelevant in a discussion of the most effective and worthy causes. Alas, to paraphrase John M. Keynes, the world is ruled by little else!)
The effective altruist movement (and a lot of other movements, for that matter) has drawn heavily on econ101-style simplifying assumptions to allow them to arrive at precision-calculated effectiveness. Indeed, most EA and GiveWell analyses of malaria preventive campaigns, such as those provided by the AMF and the Malaria Consortium, largely agree with the advice provided by the economist above.
Still, the Global Priorities Institute has been admirably open about the moral and epistemological uncertainties arising from the ‘indirect effects’ and ‘externalities’ of the interventions they promote (in stark, historical contrast to many prominent academic economists). It further highlights the critical uncertainty in how to determine the value of important normative parameters, e.g., the social rate or pure time preference” (see page 3–5 of this good GPI reflection note for detail). GiveWell further acknowledges the possibility of ‘negative consequences’ and admits it has not considered whether the nets will “remain impactful in the future”, or the cost-effectiveness of alternative means of distributing bed nets, such as market incentives.
Overall, the EA or GiveWell program advisor might add that:
· As direct investments typically give more ‘bang for your buck’ compared to investments managed through an intermediator (e.g. due to commissions or outright arbiter arbitrage) investments and programs that target individuals or an affected population (e.g. people affected by malaria) directly are preferred.
· Direct measures, typically being less costly per unit of intervention (e.g. an insecticide-treated net), should be prioritised when they provide direct benefits (e.g. mosquito-bite protection).
· Investments that provide second-order benefits should not be prioritized if these come at the expense of the primary cause’ cost-effectiveness. In such cases, the two benefits should be targeted separately to improve the (cost-effective) precision of the measure(s).
We owe the planet to the future
So far so good. You have evaluated different causes, harmful effects and counter-measures and weighted them according to the best available heuristics and scientific advice. So as a wealthy alien philantropist (drawn from rich deposits of kryptonite or the like), you may decide to ‘put your money where your moth is’ (wherever alien mouths may be) and invest heavily in preventive bed nets. This after all is the hithereto best documented and dollar-per-life effective measure out there.
Still you may be cautious and ask your human collaborators to elaborate: Given that distribution of intesticide-treated bed-nets have been the predominant intervention for more than a decade, and the disease seem to be spreading faster than before, what is the long term effectiveness of bed net distribution?
If I were your adviser, I would encourage you, as I just did my friends and peers, to take historic evidence seriously and weight highly the documented effect of measures such as those represented by the AMF: ITNs do indeed have documented effect and are proven highly effective.
To date.
I would hasten to add that, just because something has worked historically, does not mean it will work under present or future conditions. Climate change and changing evolutionary pressures are shifting the ground. So far, we have not been good at analysing this kind of change (mostly because it is unprecedented at this scale, but also because of the ‘change blindness’ that follows from slow, incremental change). Yet, there is a more fundamental ‘blindness’ that has hampered our efforts at analysing effectiveness.
I have chosen to call this ‘systems blindness’.
What is missing from our current conception of both future value and risk is an appreciation of system dynamics. In particular, the possibility of feedbacks, non-linear impacts and tipping points. Feedbacks (sometimes also referred to as cascade or domino effects) implies that an intervention directed at one problem (system effect) may impact several problems simultaneously, may amplify a (seemingly) unrelated side-stream effect, or more typically, may itself become a cause for other (unintended) effects. Non-linearity and tipping points are related and indicates that dose-response effects are correlated in unpredictable ways. These dynamics implies that risk or impact distributions are skewed or have so called ‘thick tails’. Events on this end of the spectrum represent rare, yet catastrophic events that are disproportionately costly (damaging) or shift the ‘state of affairs’ (system state) all together.
Therefore, in the next stage, you are encouraged to consult someone with a background in ecological or biological science. Or perhpas to economise on time, you refer directly to an ecological economist — trained in making both environmental and economic assessments. Biologists, ecologists, and (especially) ecological economists do of course come in many shapes and sizes, and may in no way provide sufficient answers to the question posed above. Yet they possess a rather unique earthly ability; they have been trained to think in systems.
The ecological economist will again point out that the root cause of the uptick in malaria cases is the rapid adaptation of an aggressively expanding mosquito species (Anopheles stephensi). Yet what causes this evolutionary process. Let’s not go all the way to Charles Darwin (who never got to anywhere near explaining ‘origins’, as alluded to in his famous book’s title), and his principles of natural selection. Rather, suffice to say that the present evolution of certain species of mosquito are all but natural. Today, across a wide range of localities, the main cause of evolutionary selection pressures are so-called phenotypic pressures. These may, simplified be referred to as short-term induced change in long-term behavioural patterns.
In our case of malaria-carrying mosquitoes, it can be traced back to the mosquitoes being pushed out of its natural habitats or sources of nutrition, or at least being pulled by the new (reproductive) opportunities presented by urban settlements (and husbandry).
Commenting on the (macro/micro) economist’s advice above, the ecological economist would caution you: While probably sound and appropriate for the pursuit of private investments or savings purposes, the principles are highly problematic for assessing long-term and ecological system effects. In short, these principles are not cause-effective when taking on causes that are systemic (systems-wide) and richly joined (interacting).
So what would such investments provide in the future? More opportunities? More pay-offs? No, the question is not one of ‘what’s the expected rate of return’ — but rather ‘what is an acceptable rate of reduced downturn’!
In fact, we already owe the future. At present rates we’re consuming natural resources on future credit, and already owe a considerable dept to future generations. Put in simple terms: for the planet to provide the basis for thriving and happy lives, ‘we owe the planet to the future’.
Better than efficient, smarter than effective: Dancing with (ecological) systems
In short, being cause-effective implies playing on the strengths (and effectiveness) of complex systems, rather than trying to reduce them to simple, mechanistic systems.
In the case of insect-borne or zoological diseases we have mostly taken the reductionist and mechanistic approach. This implies targeting individual ‘problematic’ animals and trying to divert the immediate risk (exposure) ‘at source’.
While such a design approach might work (relatively) well when working with human-devised systems (such as government programs, artificial intelligence, or animal farms), it will rapidly unravel in naturally evolved systems, such as ecosystems.
Rather, an ecological and systems-supportive approach would try to understand system effects, such as feedbacks, non-linear impacts and tipping points (for a great primer to this approach, refer to this great intro on dancing with systems, by the ‘system thinking’-pioneer Donnella Meadows). In many ways, this means breaking with a few of the ingrained, largely tried and tested economic principles for policy design. In the following, I lean on systems thinking and ecological principles to suggest a way forward.
For example, the principle of anticipating detrimental outcomes, ex ante, is largely futile, at least highly improbable, when such events occur within systems displaying high interconnectivity and non-linearity. Rather than trying to device interventions that target present conceptions of ‘disasters’, the policy planner should aim efforts at understanding and scaffolding the conditions and support functions that make society more resistant — or in biologist parlance, resilient — to adverse effects.
This would seriously impede the first recommendation to try to protect human populations (indefinitely) from interactions with malaria-carrying mosquitoes, because this might reduce both humans adapted resistance to the disease they carry, and ecosystems’ fine-tuned resilience in the face of sudden changes to life-conditions in mosquitos natural habitats.
Second, the principle of solely targeting affected populations (or again in biologist parlance, ecosystem components) directly, might lead to unintended consequences or adverse effects. This is typically because the behaviour of one (eco)system component affects the behaviour of all other components (e.g. other species). This is typically illustrated through predator-pray simulations, where a direct intervention, e.g. a reduction in the population of pray, results in unpredictable feedback effects and potential tipping points for the reproductive ability of all other populations that compete for interrelated resources.
This would impede the second recommendation of eliminating mosquitoes, as this would simultaneously eliminate (or seriously diminish) the population of insect-eating species, such as birds and amphibians. This would seriously affect the health and support functions of the surrounding ecosystem.
Third, the principle of valuing present and tangible benefits higher than future, largely immaterial benefits may lead to loosing attention to support functions, operating in the background. While prioritizing interventions at present generations and trying to reduce suffering real suffering in peoples real lives today is very much justified, the problem comes once only these near-term, symptomatic effects are targeted. This makes it harder to direct financial and mental capacities towards the not-so-visible support functions and ecosystem resilience that, if neglected, might cause severe suffering and unknowable consequences in the medium term.
I conclude that to be truly effectual, the Effective Altruists and Giving What We Can movements need to pay more attention to the present drivers of suffering in sentient beings and the pre-conditions for life to flourish — both here and now and into the (not-so-distant) future.
Because the conditions for thriving life (and lives) we formerly took for granted are under serious threat. Without these preconditions for life — not least for life to thrive — we cannot expect to make much progress on any development indicators we care about.
By present drivers I mean the first-order harmful activities that we presently cause natural ecosystems, that might lead to second-order detrimental effects at a later stage. I may describe it as targeting the environmental footprints that work, metaphorically, to saw off the branches that support life.
By pre-conditions I mean the conditionals and priors (in Bayesian parlance) that allow for the (cost-effective) implementation of measures or that enables an intervention to become effective. A visceral example is the prevalence of floods or forest-fires in an area already ravaged by ‘global priority’ causes such as epidemics, extreme poverty or malnourishment. Another example is failing state bureaucracies that severely hamper implementation feasibility.
As things stand we know of one livable planet — Planet Earth (i.e., in the foreseeable future there is no Planet B). And the planet that my and future generations inherited is struggling. This view should provide two major implications. First, the worsening state of and conditions for Earth’s life-support systems (such as the biosphere, hydrosphere and lithosphere) is existential for the billions of lives — and livelihoods — of all future generations. Second, the drivers for this deterioration, must and can only be solved by present generations living in, and enacting the cause for worsening conditions.
Finally, based on my commitment to and worry about the decades long, albeit resurfacing fight against malaria, I draw three lessons for the EA community that are crucial for actually Doing Good Better:
· The floor sets the limit: Funding campaigns and projects cannot be cause effective in the medium-term while ignoring underlying drivers and ecological foundations.
· Target long-term, second-order sources of suffering: Global priorities need to target the suffering deriving from the worsening conditions of ecosystems and their support functions as well as the (first-order) suffering in (worst of) sentient beings
· Value the future, finally and formally: While some of these measures might appear costly and cost-ineffective in the short-to-medium term, the counterfactual scenario (i.e. the irreversible loss of support functions or ecological foundations) would prove much more costly in the mid-to-long term. This is not captured in most cost-benefit-style assessments, primarily because of its myopic instincts (foregrounding present value), and our inability to (at least highly inapt) measure immaterial benefits.