Former software engineer and manager now researching and writing about EA-related topics in harm reduction. Interested in global health and mental health more broadly.
Hi Joel, thank you for the detailed and thoughtful comment!
(1) I agree that the post mischaracterized the report. My apologies for that. I've updated the corresponding section to make it clear that the adjustments were made due to the surveyed group being unrepresentative of the intervention's target, not because of circular reasoning.
(2) I agree that autonomy shouldn't be privileged a priori above life, health, or any other values, but this is the more committed harm reductionist view the post is describing, not defending. What I am arguing for is something like the Nuffield Council on Bioethics' intervention ladder which takes the least intrusive option as the default and puts a rising burden of justification on each more coercive step. In other words, reaching for coercion should require first showing the less intrusive option won't do, rather than letting a large enough health number outweigh the autonomy cost directly.
The justification for this is that case-by-case pricing of autonomy is unusually unreliable in this specific domain, because the people building the models often don't share the preference being restricted, so there's a greater bias in one direction. This is consistent with data on acceptability of paternalistic interventions (unaffected people support them more). Externalities may explain a part of that (e.g. second hand smoke), but that wouldn't apply to basically self-regarding cases like sodium or helmet laws.
Another example with a similar structure is a dishonesty prior. A consequentialist doesn't re-run the cost-benefit each time they're tempted to deceive, not because honesty is privileged for its own sake, but because the case-by-case exceptions are exactly the ones motivated reasoning corrupts. I'm claiming autonomy in this domain is the same kind of case.
This is awesome! Claude does have an incognito mode so I tested your queries there and made a copy of your doc with its responses along with my ratings included:
In 1633, Murad IV banned all tobacco in the Ottoman Empire - the policy was reversed by his successor 10-15 years later
The longest ban in history was implemented in Muscovy Russia in 1634 and lasted just over fifty years until Peter the Great repealed it
A smoke-free generation law similar to what one SMA incubatee has argued for was passed in New Zealand in 2022 and repealed less than two years later
Jurisdictions where cigarettes are currently taxed higher than most of the market will bear, or where noncombustible products are banned, also offer some evidence about the feasibility of keeping smuggling down:
A study examining littered cigarette packs in New York City found less than 20% of them had tax stamps showing they were legally sold
Seizures of e-cigarettes in Brazil, where vaping is completely banned, have been growing since 2019 and reached 1.3 million units in 2023
This article from one of their incubatees describes their strategic approach:
Governments spend billions on healthcare and millions more subsidising policy campaigns to fight industries that thrive on harm, trying to protect us from tobacco, ultra-processed food, alcohol and fossil fuels to name a few. Yet year after year, these industries continue to expand, exploit, and interfere with regulation. They privatize profits while socializing costs. The result: fragile health systems, sick populations, and taxpayers footing the bill for preventable disease.
But here’s the thing: we already have the laws to stop them. What we lack is the will, and the funding, to enforce those laws through the courts.
That’s why we launched SHIFT, a catalyst funder for strategic litigation against health-harming corporations. Our aim is straightforward: enforce existing legislation, hold corporations accountable, and stop industries of harm from thriving. Litigation, after all, doesn’t persuade — it compels.
They appear to see the tobacco industry as an emblematic example of industries that are net harmful, and legal action as a specifically effective way to fight them:
We believe the fight begins with tobacco because it’s the model for every other harmful industry. It is the deadliest consumer product in history, killing more than 7 million people each year. It is also the most regulated, thanks to the Framework Convention on Tobacco Control (FCTC), the world’s only health-specific treaty.
That treaty, and the precedents already set in tobacco control, give us a unique opportunity. If we can enforce existing laws against tobacco companies, we can create a “halo effect” — precedents that ripple across industries, forcing others to meet higher standards of corporate accountability.
One of their other fellows authored a report for the European Respiratory Society that specifically advocates for and "endgame" strategy for eventual full prohibition via generational sales bans:
The power to introduce a generational sales ban, where it will never be legal to sell tobacco products to people born after a certain date, lies firmly within the competence of the Member States. There is no impediment, under EU law, to introducing such an endgame policy on the grounds of public health, given that it can be demonstrated to be both a proportionate and necessary measure to achieve a legitimate objective.
It isn't necessary to conclude that the benefits of tobacco are not worth the costs for any user, or that no user would continually make a free choice to consume if fully apprised of the facts, to conclude that the tobacco industry should be eventually abolished.
Agreed. However, the post as I read it isn't arguing that stricter regulation of some or all tobacco products is inherently unjustified. It's postulating that our confidence in the claim that it's net positive is shakier than the current models make it seem, because SMA and other organizations favoring such regulation are ignoring a major factor when modeling its costs and benefits, namely the enjoyment people get from using the products. It's one thing to disagree about the extent to which this matters and what weight to assign it in a model; it's another thing to fail to include it altogether.
Of the 1/3 of smokers who did not express a desire to quit, I wonder how many of them would express a desire to return to smoking if we were to wave a magic wand that cured addiction and associated cognitive distortions and then asked them to make an actually free choice about whether to resume smoking. Cognitive dissonance is quite a drug! I don't put that much weight on the fact that each smoker decided to start smoking at some prior point in time as evidence of their preferences in a non-addicted state. Unless the smoker decided to partake the first time that cigarettes were a possibility, they also made decisions not to smoke in the past.
IMO this is a crux of the disagreement the OP has with SMA's approach. The concept of addiction is doing a lot of work here in justifying treating smoking as a special case in a way that isn't applied to less socially stigmatized risks like a dangerous hike or regretted choices the OP mentions like those about social media use and diet. I'd be interested in understanding what you mean by "curing addiction" in this context - would it be something like never having known the pleasure of smoking, adding a moment of mindful reflection before each cigarette, or something else? And how can we tell whether someone is making the choices they are making because they are addicted vs. because they have weighed the pros and cons sufficiently for us to believe they are making a free choice?
Of the 2/3 of people who smoke and did express a desire to quit despite not having done so, I also wonder how many of them respond in this way at least in part because of the stigma surrounding smoking (this seems to happen in at least some subpopulations). Social desirability bias is also a hell of a drug.
There is no known realistic way to abolish only the part of the industry that sells to people who would prefer to be free of their addiction.
Do you think there's a known realistic way to abolish the industry as a whole when more than a billion people in the world smoke daily? If yes, what's a tractable path given the track record of prohibition of psychoactive substances more generally, and tobacco specifically (e.g. recent bans and subsequent reversals in South Africa and Bhutan)? If not, why is this a reasonable goal for an evidence-driven organization to set, compared to educating the public about the risks of smoking and assisting those who desire to quit, as the OP suggests?
Excellent post. I would add that SMA's approach to reducing smoking-related harm has another defect that doesn't require one to have a position on the question of how to model the benefits of smoking, which is their position on much safer, noncombustible forms of tobacco and nicotine use like vaping, nicotine pouches, and snus.
There's plentiful evidence that all of these products hugely diminish risks to longevity, are the most effective known way to stop smoking, and have a major measurable impact on population-level disease incidence when widely available. An Our World in Data post from a few weeks ago summarizes the bulk of it, and I've also written a sequence on this for the EA forum. SMA describes vaping as a "way the tobacco industry is dodging the rules," puts scare quotes around "safer" when discussing (and dismissing) its health benefits to people who smoke, and their incubatees appear to be lobbying against regulation that would make it more available while ignoring the countervailing evidence when confronted with it publicly.
I don't have any reason to think SMA has anything but good intentions in helping people achieve better outcomes (health-related or otherwise) for their lives, but from the tone of their communications it appears that they are conflating hurting the tobacco industry with helping people that suffer because of the defects of their products. The idea that an industry that created a problem that causes so much death and suffering may, if properly regulated and incentivized, also be part of the solution to that problem appears to be beyond their moral imagination (or ambition, one might say) even though this approach seems to me pretty standard in other causes EA is interested in, like working with farmers and the meat industry towards progress in animal welfare or with fossil fuel companies to mitigate climate change.
Hi Brad, thanks for reading and commenting. The School for Moral Ambition and its incubatees are the EA-associated organizations I mentioned above. They appear to be strongly opposed to promoting vaping to people who smoke ("Vaping: a healthy alternative to smoking" appears as an item in a "Big Tobacco Bullshit Bingo" card used in one of their trainings), have lobbied for restrictions on e-cigarette use in places where smoking is banned, and have spread misinformation on both the risks of vaping and the evidence for its usefulness in smoking cessation.
I think they have good intentions and may be doing helpful work with other tobacco control efforts, but when it comes to vaping and other noncombustibles, they appear to be contributing to the problem Hannah describes rather than trying to be part of the solution.
Great post. Do you know whether GiveWell (or any other org doing evaluation) accounts for the extent to which NGO's are doing this when assembling recommendations? While being aware of ecosystem thinking seems important, having incentives for acting in accordance with it could be even more so.
I think there's some complexity (and a lot of confusion) in the tobacco story as well. The industry did try to trick consumers and regulators in the 1970's that light cigarettes were less harmful when they themselves had evidence that that wasn't the case. And now, the opposite is happening as misinformation has resulted in majorities of doctors believing nicotine causes cancer and majorities of consumers believing vaping is as harmful or more harmful than smoking.
Great post, and rhymes with some of the challenges I've observed in my own area of interest, tobacco harm reduction, particularly this part:
As far as I can tell, the movement responded to this by calling it humanewashing, and pushed harder for the BCC. But if McDonald’s actually did do all the things they said they would, the impact on the broilers would likely be more impactful than the BCC.
A number of large tobacco companies have developed products they claim are significantly less harmful than cigarettes and can serve to transition smokers away from them. If this is true, it's a boon for public health, but many organizations in the field, including the WHO from which many regulators in LMICs take their cue, support banning them altogether despite criticism from non-industry affiliated experts who see their potential in reducing death and disease from smoking.
I get the impression that pragmatists in drug policy and animal welfare have quite a bit to learn from each other in advocating for sensible policies that avoid letting the perfect be the enemy of the good.
Hi Joel, thank you for the detailed and thoughtful comment!
(1) I agree that the post mischaracterized the report. My apologies for that. I've updated the corresponding section to make it clear that the adjustments were made due to the surveyed group being unrepresentative of the intervention's target, not because of circular reasoning.
(2) I agree that autonomy shouldn't be privileged a priori above life, health, or any other values, but this is the more committed harm reductionist view the post is describing, not defending. What I am arguing for is something like the Nuffield Council on Bioethics' intervention ladder which takes the least intrusive option as the default and puts a rising burden of justification on each more coercive step. In other words, reaching for coercion should require first showing the less intrusive option won't do, rather than letting a large enough health number outweigh the autonomy cost directly.
The justification for this is that case-by-case pricing of autonomy is unusually unreliable in this specific domain, because the people building the models often don't share the preference being restricted, so there's a greater bias in one direction. This is consistent with data on acceptability of paternalistic interventions (unaffected people support them more). Externalities may explain a part of that (e.g. second hand smoke), but that wouldn't apply to basically self-regarding cases like sodium or helmet laws.
Another example with a similar structure is a dishonesty prior. A consequentialist doesn't re-run the cost-benefit each time they're tempted to deceive, not because honesty is privileged for its own sake, but because the case-by-case exceptions are exactly the ones motivated reasoning corrupts. I'm claiming autonomy in this domain is the same kind of case.
This is awesome! Claude does have an incognito mode so I tested your queries there and made a copy of your doc with its responses along with my ratings included:
https://docs.google.com/spreadsheets/d/1Qlq9hTLQ4U7BPdMNN7Rsxxo4A8GLDv0tfYjll4MKkW4/edit?gid=0#gid=0
It did a bit better than ChatGPT and Gemini but still below Grok. I'm pretty impressed with how "EA-pilled" it was overall.
It seems to me there's a lot of available data to help make an informed guess of the likelihood of success of the kind of measure SMA advocates for.
Tobacco bans are almost as old as the global tobacco trade itself, and all of them have been eventually reversed. Some examples beyond that of Bhutan:
Jurisdictions where cigarettes are currently taxed higher than most of the market will bear, or where noncombustible products are banned, also offer some evidence about the feasibility of keeping smuggling down:
This article from one of their incubatees describes their strategic approach:
They appear to see the tobacco industry as an emblematic example of industries that are net harmful, and legal action as a specifically effective way to fight them:
One of their other fellows authored a report for the European Respiratory Society that specifically advocates for and "endgame" strategy for eventual full prohibition via generational sales bans:
Agreed. However, the post as I read it isn't arguing that stricter regulation of some or all tobacco products is inherently unjustified. It's postulating that our confidence in the claim that it's net positive is shakier than the current models make it seem, because SMA and other organizations favoring such regulation are ignoring a major factor when modeling its costs and benefits, namely the enjoyment people get from using the products. It's one thing to disagree about the extent to which this matters and what weight to assign it in a model; it's another thing to fail to include it altogether.
IMO this is a crux of the disagreement the OP has with SMA's approach. The concept of addiction is doing a lot of work here in justifying treating smoking as a special case in a way that isn't applied to less socially stigmatized risks like a dangerous hike or regretted choices the OP mentions like those about social media use and diet. I'd be interested in understanding what you mean by "curing addiction" in this context - would it be something like never having known the pleasure of smoking, adding a moment of mindful reflection before each cigarette, or something else? And how can we tell whether someone is making the choices they are making because they are addicted vs. because they have weighed the pros and cons sufficiently for us to believe they are making a free choice?
Of the 2/3 of people who smoke and did express a desire to quit despite not having done so, I also wonder how many of them respond in this way at least in part because of the stigma surrounding smoking (this seems to happen in at least some subpopulations). Social desirability bias is also a hell of a drug.
Do you think there's a known realistic way to abolish the industry as a whole when more than a billion people in the world smoke daily? If yes, what's a tractable path given the track record of prohibition of psychoactive substances more generally, and tobacco specifically (e.g. recent bans and subsequent reversals in South Africa and Bhutan)? If not, why is this a reasonable goal for an evidence-driven organization to set, compared to educating the public about the risks of smoking and assisting those who desire to quit, as the OP suggests?
Excellent post. I would add that SMA's approach to reducing smoking-related harm has another defect that doesn't require one to have a position on the question of how to model the benefits of smoking, which is their position on much safer, noncombustible forms of tobacco and nicotine use like vaping, nicotine pouches, and snus.
There's plentiful evidence that all of these products hugely diminish risks to longevity, are the most effective known way to stop smoking, and have a major measurable impact on population-level disease incidence when widely available. An Our World in Data post from a few weeks ago summarizes the bulk of it, and I've also written a sequence on this for the EA forum. SMA describes vaping as a "way the tobacco industry is dodging the rules," puts scare quotes around "safer" when discussing (and dismissing) its health benefits to people who smoke, and their incubatees appear to be lobbying against regulation that would make it more available while ignoring the countervailing evidence when confronted with it publicly.
I don't have any reason to think SMA has anything but good intentions in helping people achieve better outcomes (health-related or otherwise) for their lives, but from the tone of their communications it appears that they are conflating hurting the tobacco industry with helping people that suffer because of the defects of their products. The idea that an industry that created a problem that causes so much death and suffering may, if properly regulated and incentivized, also be part of the solution to that problem appears to be beyond their moral imagination (or ambition, one might say) even though this approach seems to me pretty standard in other causes EA is interested in, like working with farmers and the meat industry towards progress in animal welfare or with fossil fuel companies to mitigate climate change.
Hi Brad, thanks for reading and commenting. The School for Moral Ambition and its incubatees are the EA-associated organizations I mentioned above. They appear to be strongly opposed to promoting vaping to people who smoke ("Vaping: a healthy alternative to smoking" appears as an item in a "Big Tobacco Bullshit Bingo" card used in one of their trainings), have lobbied for restrictions on e-cigarette use in places where smoking is banned, and have spread misinformation on both the risks of vaping and the evidence for its usefulness in smoking cessation.
I think they have good intentions and may be doing helpful work with other tobacco control efforts, but when it comes to vaping and other noncombustibles, they appear to be contributing to the problem Hannah describes rather than trying to be part of the solution.
Great post. Do you know whether GiveWell (or any other org doing evaluation) accounts for the extent to which NGO's are doing this when assembling recommendations? While being aware of ecosystem thinking seems important, having incentives for acting in accordance with it could be even more so.
I think there's some complexity (and a lot of confusion) in the tobacco story as well. The industry did try to trick consumers and regulators in the 1970's that light cigarettes were less harmful when they themselves had evidence that that wasn't the case. And now, the opposite is happening as misinformation has resulted in majorities of doctors believing nicotine causes cancer and majorities of consumers believing vaping is as harmful or more harmful than smoking.
Great post, and rhymes with some of the challenges I've observed in my own area of interest, tobacco harm reduction, particularly this part:
A number of large tobacco companies have developed products they claim are significantly less harmful than cigarettes and can serve to transition smokers away from them. If this is true, it's a boon for public health, but many organizations in the field, including the WHO from which many regulators in LMICs take their cue, support banning them altogether despite criticism from non-industry affiliated experts who see their potential in reducing death and disease from smoking.
I get the impression that pragmatists in drug policy and animal welfare have quite a bit to learn from each other in advocating for sensible policies that avoid letting the perfect be the enemy of the good.