Eating Honey is (Probably) Fine, Actually

Beefin’ with Bentham Bulldog’s Bad Bee Barguments

“One pump of honey?” the barista asked.

“Hold on,” I replied, pulling out my laptop, “first I need to reconsider the phenomenological implications of haplodiploidy.”

---

Recently, an article arguing against honey has been making the rounds. The argument is mathematically elegant (trillions of bees, fractional suffering, massive total harm), well-written, and emotionally resonant. Naturally, I think it's completely wrong.

Below, I argue that farmed bees likely have net positive lives, and that even if they don't, avoiding honey probably doesn't help much. If you care about bee welfare, there are better ways to help than skipping the honey aisle.

Source

Bentham Bulldog’s Case Against Honey

Bentham Bulldog, a young and intelligent blogger/tract-writer in the classical utilitarianism tradition, lays out a case for avoiding honey. The case itself is long and somewhat emotive, but Claude summarizes it thus:

P1: Eating 1kg of honey causes ~200,000 days of bee farming (vs. 2 days for beef, 31 for eggs)

P2: Farmed bees experience significant suffering (30% hive mortality in winter, malnourishment from honey removal, parasites, transport stress, invasive inspections)

P3: Bees are surprisingly sentient - they display all behavioral proxies for consciousness and experts estimate they suffer at 7-15% the intensity of humans

P4: Even if bee suffering is discounted heavily (0.1% of chicken suffering), the sheer numbers make honey consumption cause more total suffering than other animal products

C: Therefore, honey is the worst commonly consumed animal product and should be avoided

The key move is combining scale (P1) with evidence of suffering (P2) and consciousness (P3) to reach a mathematical conclusion (P4→C) that honey causes more total suffering despite individual bees mattering less than larger animals.

Where I agree with Bentham’s Bulldog

It might surprise you to learn that I agree with BB that there are a lot of bees. (Claude for example estimates 5 trillion bees in 100+ million hives). I also agree that conditional upon being able to suffer, many bees plausibly suffer greatly. Finally, I agree that many bees suffering greatly, if true, is a big deal.

Where I disagree

(Un)fortunately, I disagree with pretty much everything else. Working backwards:

1. I disagree with the (implicit) assumption that you should always avoid things that cause a lot of suffering. This is primarily because you should look at the balance of suffering against positive moral goods like pleasure.

2. I disagree, with low credal resilience, with the claim that bees suffer at 7-15% the intensity of humans.

3. Conditional upon bees having the ability to suffer greatly, it is likely that they are also capable of feeling pleasure greatly.

4. On balance, I think it’s likely that farmed bees have net positive lives.

   1. I think many EAs take it as a given that insects have net negative lives. I think this is a mistaken inference drawn from swapping intuitions of K-selected species onto the actual experiences of r-selected species.

   2. For eusocial insects like bees in particular, evolution ought to incentivize them to have net positive lives as long as the hive is doing well overall.

   3. Beekeepers are further incentivized to keep their hives living and healthy, which likely is a positive contributor to farmed bee welfare.

   4. Bees are not locked down and have exit options like swarming. Thus, revealed preferences point towards them preferring to be in managed hives over wild ones.

   5. Finally, the empirical evidence for welfare is both limited and mixed but in my opinion points mildly towards farmed bees having net positive lives, or at least better than pollinators in the wild.

5. Conditional upon farmed bees having net positive lives, I believe under most common philosophical frameworks eating honey is obviously morally permissible.

   1. The main exception I think might be strong suffering-focused or asymmetry views, where you may believe that causing a lot of suffering is bad even on consenting individuals who overall feel much more pleasure than suffering, and want to be alive.

6. If you instead believe that farmed bees have net negative lives (or alternatively, you reject the net-positive vs net-negative framing), it is still not obvious whether you should avoid eating honey. At minimum, you have to consider the counterfactual.

   1. Not eating honey may not do much to reduce the honeybee population, as many beekeepers are paid more from pollination than from honey.

   2. Successfully reducing the honeybee population significantly will likely lead to a (much) higher number of wild pollinators, which may have lives that are overall worse than in managed honeybee populations.

To briefly elaborate on #1, bubonic plague caused significant suffering and death. Therefore, bubonic plague is bad. However, the agricultural revolution has also caused significantly suffering and death (for example, because the agricultural revolution has led to bubonic plague). This does not by itself mean the agricultural revolution is bad, at least according to classical utilitarians.

Why? Responsible utilitarians look at the overall balance of costs and benefits, instead of only paying attention to one side of the moral ledger. In the case of the agricultural revolution, it led to many good things like sanitation and the internet, not just bubonic plague. Similarly, even if some bees suffer greatly in managed hives, many other bees (or in some cases the same bees!) might enjoy their lives and prefer living in managed hives greatly to not existing.

BB, as a classical utilitarian, ought to agree with this basic point, so I will not elaborate further.

I also disagree with the “bees suffer at 7-15% the intensity of humans” estimate. BB links to the moral weights reports written by my former colleagues at Rethink Priorities, which have been both praised and criticized strongly. The top critique on LessWrong expresses incredulity at the numbers, with some accompanying discussion.

I do not believe BB’s core arguments are particularly sensitive to the orders of magnitude involved here. Further, I think much ink has already been spilled on this point, and I don’t think I have much novel to add here, or be persuasive. So I’ll punt on this question further. Instead, my post will focus primarily on the sign (net positive or negative) of bee welfare, and secondarily on the sign of anti-honey interventions. I will leave the discussion of magnitude (whether the intensity of bee suffering/pleasure, or the importance of bee suffering/pleasure, is 7% of humans, or 0.000000000000000007%, or 7000%) to others.

Conditional upon believing that bees are likely to suffer intensely, you should probably also believe that bees are also capable of feeling pleasure intensely. This is not necessarily the case; for example, you might believe that suffering is an earlier primitive than pleasure in evolution, or that honeybees have uniquely evolved to suffer greatly without being capable of great happiness. However I don’t think the balance of considerations should lead you to believing this. (If you disagree, feel free to comment here or DM me and we can discuss cruxes)

The rest of my post will focus on points 4-6.

Evolutionary Considerations for Bee Welfare

When someone asks 'honey in your bubble tea?', the natural response is obviously to first consider the welfare biology implications of the evolutionary divergence between r-selected and K-selected reproductive strategies.

It's useful to have priors. I’ve considered two different evolutionary frameworks for thinking about insect/bee welfare.

Argument from r-selection

I initially thought r/K-selection might be an useful explanatory framework here. The argument roughly goes: r-selected species (many offspring, low parental investment) expect most offspring to die, so evolution might favor relatively small amounts of suffering but strong rewards for successful survival behaviors. This contrasts with K-selected species where each death represents massive lost investment.

However, multiple readers pointed out an important mistake: honeybees aren’t r-selected! Or at least, they are in somewhat of a middle ground (queens live 2-5 years, workers receive extensive parental care as larvae, and colonies maintain stable environments. However worker bee lifespans are measured in weeks rather than years), and biologists primarily consider them more K-selected than r-selected.

I still think understanding the welfare biology implications of r- vs K-selection is a key piece of the puzzle for understanding insect welfare in general (as many other insects are r-selected), as well as earlier mistakes in reasoning EAs have made with regard to wild animal welfare. However, it just isn’t that directly relevant to honeybee welfare.

(Original argument in footnote¹)

Which brings us to a more relevant framework: eusociality.

Eusociality considerations

Source

Eusocial insects represent evolution's most extreme form of cooperation. In honeybees, workers are sterile females who share 75% of genes with their sisters (due to male bees being haploid). A worker bee is evolutionarily "trying" to propagate genes through the queen's offspring, not its own.

This creates fundamentally different selection pressures on their welfare systems. Consider first the negative welfare side: For solitary insects, the worst evolutionary outcome is individual death - and crucially, most will experience the lead-up to this outcome (predation, starvation, disease). Their suffering systems evolved to signal threats to personal survival and reproduction. For eusocial workers, individual death is also common, but the worst evolutionary outcome is colony collapse - something the vast majority will never experience. Worker bees live ~6 weeks in summer while healthy colonies persist for years or decades. This means most workers go through life without experiencing their species' equivalent of maximum evolutionary failure, and the vast majority of them will not reach their maximum capacity for negative welfare, or anywhere close.

On the positive welfare side, eusociality creates additional sources of reward beyond those available to solitary insects. While both solitary and eusocial insects likely find survival activities (foraging, eating) rewarding, eusocial insects have likely evolved to also find fulfillment in social activities like nursing larvae, building comb, grooming sisters, performing waggle dances, and other colony-beneficial behaviors. The colony structure creates diverse roles and social interactions that simply don't exist in solitary insect life.

Crucially, evolution should align worker experiences with colony success. Over the timescale of 100+ million years, bee neurology has likely been sculpted to make hive-beneficial work inherently rewarding. A worker that suffered while foraging or nursing would be less effective at these crucial tasks. The alternative (bees suffering while doing hive-beneficial work) seems logically possible but likely less evolutionarily stable.

As an intuition pump: when a honeybee stings you and dies defending her hive, do you imagine she experiences agony and regret? My intuition suggests otherwise; to the extent bees have complex emotional experiences, sacrificing for the colony likely feels fulfilling rather than painful. Evolution has had ample time to align individual experiences with genetic interests, which for workers means colony survival.

When trying to understand animal suffering, some degree of anthropomorphization is unfortunately unavoidable – until we have a clear gears-level neuronal basis for suffering and happiness, relating animal suffering to human suffering is one imperfect tool among several to help us ground what it’s like to be a bee. But when anthropomorphizing, you have to be very careful; just thinking “would I like to be in that situation” is insufficient if the backdrop of your situation and biology is overall very different from that of another being.

I consider the priors here to be among the strongest arguments, not because I think they're rock-solid but because I think reasoning about animal suffering in general is hard, especially so for insects. So the theoretical arguments here are relatively stronger just because the other lines of evidence are so weak.

Non-Evolutionary Evidence for Bee Welfare

Beekeeper incentives

Bees are used for their labor (pollinating crops), and for the fruits of their labor (honey). Beekeepers want to maximize worker productivity. Usually this is not accomplished via (the equivalent of) depressed bees in pain, under parasite load, etc, but by doing their best to make sure bees are sufficiently healthy for work.

In practice, beekeepers do take actions they’re economically incentivized to do (e.g. keep bee hives warm during the winter season, feed juvenile hives, combat parasites and other diseases, etc).

In addition, many beekeepers are amateurs, and apiarists report great pride in keeping their hives healthy and well-maintained. This to me is weak but nonzero positive evidence that beekeepers do in fact keep their hives healthy and well-maintained, and in turn that bees have good lives in captivity.

Further, as Mr.Bulldog’s post points out, bees are capable of displaying complex behavior, including social behavior. Such complex behavior is also necessary for bees in captivity, to manage complex workloads of foraging, pollination, hive management, etc. (Note that this makes the case for bees importantly different to farmed pigs and chickens, see footnote²) If we assume that complex cognitive behavior benefits from healthy and mentally sound bees, this should further predispose us to believe that there are incentives to make bees happy.

One must imagine Sisyphus happy

Exit rights

(You can just fly away)

Further, as others have pointed out, bees have exit options. If a bee colony really doesn't like their hives, they can "vote with their wings" and fly out.

This behavior is known as “swarming” or “absconding”, where some or all bees choose to leave their existing hive. In practice, swarming is difficult and sometimes dangerous. However, unlike chickens and pigs and other farmed animals, the fact that the bees can leave at any point provides a lower bound to how bad their lives can be, at least compared to the wild.

Analogously, some countries are better than others. Some countries abuse their citizens, give them little rights, have poor economic opportunities, force undergrads to learn about power poses, etc. But importantly, some countries like North Korea prevent their citizens from fleeing, while others do not. For citizens of countries with free exit rights, immigration can be difficult, risky, and financially and socially costly. Nonetheless, the ability to leave a horrible situation has a very strong option value.

In this (mediocre) analogy, bees are much more in a “normal country” situation where they can leave at any point (they don’t even need to apply for visas) and much less like North Korea.

In addition, to the extent that you have nonconsequentialist moral intuitions in favor of animal rights, you should consider that respecting bee autonomy may entail respecting their choices to stay in human-managed hives.

Empirical considerations.

The Rethink Priorities managed honey bee welfare report, which Mr. Bulldog references repeatedly, presents mixed but (on my read) slightly positive evidence for managed bee welfare:

Positive factors:

• Beekeepers actively manage nutrition, disease, parasites, and climate protection

• Juvenile mortality (~30%) is relatively low for honeybees, compared to other insects.

• Managed hives show similar behavioral repertoires to wild colonies

• Economic incentives align with bee health (sick bees don't pollinate well)

Negative factors:

• Transport stress during pollination contracts

• Unnaturally high colony densities

• Honey removal requires supplemental feeding

• Exposure to agricultural chemicals

Crucially, the report's intervention recommendations focus on improving management practices, not reducing honeybee populations. They explicitly note uncertainty about whether wild pollinators have better or worse lives than managed bees. (An important consideration here is that managed bees will be replaced not just by other eusocial wild bee hives, but also by solitary bees and other non-eusocial insects, which plausibly have worse lives than eusocial insects).

So to summarize, the theoretical cases broadly point to bees having net positive lives, while the empirical evidence is more ambivalent or lean slightly positive. Overall I think we should have moderate confidence that bees have net positive lives, and Mr. Bulldog’s arguments to the contrary are quite weak.

Implications of honeybees having net positive lives

If I take a value pluralism approach. I think the strongest arguments against honey come from utilitarianism (classical or negative), rather than rights-, virtues-, or relational- based approaches. Though I’m not well-versed in those arguments, I think the standard non-consequentialist arguments against factory farming mostly do not apply to beekeeping practices.

In particular, unlike factory farmed chickens or pigs, a bee’s autonomy and rights (including critically exit rights to swarm away) are de facto preserved in managed honeybee settings. Beekeeping plausibly breeds a virtuous relationship with bees and other animals in a way that butchering animals for their meat en masse likely does not, and the relationship between beekeeper and beekeepee is mutually beneficial and consensual.

For a more neutral take (and references), see my two conversations with different Claude models exploring this topic.

If you lend sufficient credence to my arguments above, you should likewise believe that classical utilitarianism should not argue against you eating honey (if anything it should leave you wanting to eat more honey).

You may however think my arguments for bee happiness are bunk (boo!), or alternatively you have strong suffering-focused or asymmetry views, where you may believe that causing a lot of suffering is bad even on consenting individuals who overall feel much more pleasure than suffering, and want to be alive.

I think the practical considerations of classical utilitarians who think bees have net negative lives and negative utilitarians should de facto be similar on this question, which brings me to my next section:

Should I stop eating honey if I think bees have negative lives?

Answer: No (or at least, uncertain).

RP’s honeybee report lists 6 broad categories of promising potential interventions for people interested in reducing honeybee suffering:

1. Build Effective Animal Advocacy in Asia: Develop advocacy capacity in China/India, which have ~22 million hives (more than Europe + Americas combined).

2. Reduce Pollination Demand: Lower demand via self-fertile crops, mechanical pollination, wild pollinator increases, or reduced almond consumption.

3. Increase Access to Natural Forage: Plant diverse flowers near hives or move hives to better forage areas to prevent malnutrition.

4. Manage Pesticide Risk: Improve applicator training, use dropleg nozzles, coordinate spraying schedules, and restrict neonicotinoids.

5. Reduce Homing Errors: Paint hives different colors and vary heights/orientations to prevent 40% drift rate and disease spread.

6. Minimize Harvest Disruption: Use Flow Hives to extract honey without opening hives, though regular inspections are still needed.

Noticeably missing from the list is “stop eating honey and/or convince other people to stop eating honey.”

Even if you think honey bees have net negative lives, and that this is important enough to worry about at scale, it does not follow that it is consequentially prudent for you to avoid honey. To make that case, you also have to be right on both economics and applied population ethics:

Economics: Reducing honeybee consumption may not meaningfully reduce honeybee populations

When considering whether to stop eating honey, we need to examine the actual causal chain. BB claims "1kg of honey = 200,000 bee-days," but this assumes that reducing honey consumption proportionally reduces bee populations - a prima facie implausible claim that requires empirical validation.

Reality is more complex. Beekeeping revenue comes primarily from two sources: honey sales and pollination contracts, with the split varying dramatically by region and operation type. California almond pollinators might earn primarily from pollination fees, while small local operations might depend heavily on honey sales. These are essentially two different markets with different demand structures that happen to share a production input (managed hives).

The relevant economic question is the honeybee population's elasticity to honey demand - how much would bee populations actually decline if honey consumption decreased?³ This depends on complex factors including the cross-subsidization between honey and pollination revenues, the elasticity of pollination demand to price increases, and the availability of substitutes (for both honey and pollination).

If you’re interested in studying this further, the two things I’d first look at is price sensitivity of managed honeybee hives to revenue, and the demand elasticity of pollination. As a template for how to do this research, you might benefit from looking at methodologies of Norwood and Lusk’s earlier agricultural economics work on beef, pork, chicken meat, dairy, and eggs, as well as more recent Rethink Priorities work (though I’m not aware of a specific unit economics report that laid out its methodology as cleanly as Norwood and Lusk).

Population ethics: Reducing honeybee population may result in more wild pollinators, which may be bad.

Suppose you are confident about both the welfare biology and economics here, and think that reducing honey consumption by X can decrease honeybee population by Y. Does that mean you’re now safe to advocate to reduce honey consumption (modulo politics/backfire effects)? Nope.

Successfully reducing the honeybee population significantly will likely lead to a (much) higher number of wild pollinators, which may have lives that are overall worse than in managed honeybee populations. As previously noted, it is not at all clear that wild pollinators have better or worse lives than managed honeybees.

Even if you believed (controversially, in my view) that wild honeybees have better lives than managed honeybees, reducing managed honeybees will de facto replace them with a mix of wild honeybees and other animals, including solitary insects which probably have overall worse lives than eusocial insects. As argued earlier, eusocial insects have evolved over 100+ million years to find colony-beneficial work inherently rewarding, while solitary insects lack these evolutionary pressures for positive welfare. The RP report notes that eusocial bees receive dedicated care from nurse bees and have ~30% juvenile mortality, while many solitary insects face much higher mortality rates with no social support systems.

For people interested in diving further, Eukaryotes writes more considerations on their blog.

Conclusion

Bentham's Bulldog makes a mathematically tidy argument: many bees × small-to-moderate suffering = large problem. But this calculation rests on questionable assumptions about the sign of bee welfare.

Evolution gives us strong reasons to expect eusocial insects lead net positive lives when their colonies thrive. Managed bees exhibit natural behaviors, have exit options unavailable to other farmed animals, and work for beekeepers incentivized to maintain their health. While the empirical evidence remains mixed, it weakly leans positive rather than negative.

Even if you believe managed bees suffer on net, avoiding honey likely achieves little, and can be actively harmful. The economics of pollination mean bee populations won't dramatically decline, and wild replacements might fare worse. If you want to help bees, support animal advocacy movement building in Asia or additional research into welfare improvements, not boycotts.

The sweetest conclusion? If you enjoy honey and care about animal welfare, you can probably have your honey cake and eat it too.

1. K-selected species (like humans and elephants) invest heavily in few offspring; each death represents massive lost parental investment. Things that increase evolutionary success for K-selected species(eg, good food, sex) are much smaller evolutionary goods than death is bad. r-selected species (including most insects, though confusingly, perhaps not honeybees) in contrast, produce many offspring with minimal investment, expecting most to die quickly. This creates fundamentally different evolutionary pressures on their experience of suffering and pleasure.

For r-selected species, evolution should likely favor: (1) immediate and relatively small per-instance amounts of negative reinforcement/suffering (since e.g. the benefits of second-order pain are muted if an insect can’t learn from pain in its lifetime), (2) strong positive reinforcement for successful behaviors that lead to survival/reproduction, since so few individuals make it. An insect that reaches adulthood has "won" against tremendous odds - its accumulated positive experiences likely outweigh the brief suffering of its many dead siblings. Tomasik et. al's inference that "most insect life is dying, therefore insects have net negative lives" (which I believe Mr. Bulldog’s opinions implicitly draws from) incorrectly applies K-selected intuitions where individual death is catastrophic to r-selected species where it's statistically normal.

If you’re interested in learning more, Zach Freitas-Groff gives a more detailed argument including other theoretical considerations (paper, talk).

2. The case for farmed bees is dissimilar to the case of (e.g.) farmed broiler chickens or pigs. Because farmed chickens are used only for their meat, the incentives of the farmers are to cram them with as much food as possible and for the chickens to grow as fast as possible. They do not need to be happy (unless happier animals taste better, and I think there is little to negative empirical evidence of this). The pain-pleasure signaling mechanisms are almost completely irrelevant to caged animals since their display of complex behavior is incidental to their use as farmed animals, while for bees it's critical.

3. It’s actually more complicated than that if you’re thinking of individual (or organized group) reduction of honey consumption rather than a magical writ from the God of Economics. If all of Mr. Bulldog’s readers (1000? 10,000?) reduce their honey consumption so they collectively consume 3 less tons of honey, this does not mean that 3 less tons of honey are consumed, even in expectation. This is because someone consuming less honey means honey is (in expectation) cheaper which means other people may choose to consume more honey. The degree that organized boycotts have an effect demands a lot on demand vs supply price elasticities (whether consumers or suppliers are more price sensitive).