by Brian Tomasik
First written: 4 Nov. 2016; last update: 22 Oct. 2017


The net impact of vegetarianism on animal suffering is very unclear. Such an analysis involves many uncertain components, including the net impacts of climate change, crop cultivation, fishing, fish feed, and memetic side-effects. In this piece, I focus on just two impacts of vegetarianism that have less uncertainty:

  1. Preventing the suffering of land farm animals raised for meat
  2. Plausibly increased invertebrate populations due to less cattle grazing of grasslands.

Given certain assumptions, the farm-animal suffering prevented by vegetarianism is in the same ballpark as the wild-springtail suffering created as a result of vegetarianism. This analysis underscores the difficulty of evaluating the net impact of vegetarianism on wild-animal suffering, but it should not be taken as a complete analysis of all relevant factors.


  • Let ΔN be the number of animals who are created or prevented per year due to a typical American omnivore going vegetarian.
  • Let D be the lifespan of an animal in days.
  • Let S be the average intensity of suffering of the animal over the days of its life. This intensity is expressed relative to the typical intensities of emotion that animals of that type can experience, i.e., it doesn't count the sentience of the animal. For example, below I'll set the typical intensity of suffering of beef cowsa as -1 and of broiler chickens as -5. That means I think broiler chickens experience roughly 5 times as much suffering per day as beef cows do compared against the typical levels of sentience of each respective animal. This average includes the painfulness of death at the end of life. Shorter-lived animals will tend to have higher average intensities of suffering per day because a bigger fraction of their lives is spent enduring the agony of dying.
  • Let R be the sentience of the animal relative to the sentience of a springtail. This factor is subjective and depends on your moral values. In the below table, I've made up values that roughly represent my intuitions, but feel free to use your own numbers instead.

Comparison table

The following table summarizes some main impacts of going vegetarian. I'll explain it in more detail in the next section.

Animal ΔN D S R Total suffering = ΔN * D * S * R (in millions of springtail-days)
Beef cow -0.07 402 -1 100,000 3
Pig -0.3 183 -3 200,000 33
Chicken -14.9 42 -5 20,000 63
Springtail 3,500,000 37 -1 1 -130

The net impact on suffering is 3 + 33 + 63 - 130 = -31, i.e., suffering increases by 31 million equivalent springtail-days as a result of going vegetarian. This calculation doesn't count non-springtail invertebrates, so the increase in suffering on the part of wild invertebrates overall might be even higher. On the other hand, I'm less certain that beef production reduces wild-invertebrate populations on balance than I am that factory farming causes immense suffering to farm animals. And there's lots of noise in these numbers, such that the sign of the net impact is very unclear. This calculation is just one simple illustration of the kinds of analysis that can be brought to bear on this question.

More explanation of the above table


In the above table, values of ΔN are taken from Table 2 of this blog post by Harish Sethu and are adjusted by multiplying by what Animal Charity Evaluators calls the "cumulative elasticity factor" (CEF). CEF values by animal product can be found in this spreadsheet. As an example, Sethu estimates that a typical American omnivore eats 23.7 chickens per year. The mean CEF estimate for chicken is 0.63. So ΔN = 23.7 * 0.63 = 14.9. And I write this as a negative number (-14.9) because it's a decrease in the number of chickens farmed. The mean estimates of the CEFs for beef and pork are 0.67 and 0.76, respectively.

The value of ΔN for springtails is based on an estimate that a beef cow prevents in the ballpark of 5 million springtail-years over its grazing lifetime. Hence, preventing 0.07 beef cows by going vegetarian creates 0.35 million springtail-years. I'm assuming that springtails live roughly 0.1 years = 37 days each, which gives ΔN = (0.35 million springtail-years) * (10 springtails per springtail-year) = 3.5 million springtails. This page says: "Springtails reproduce rapidly and a single life cycle may take just 3-5 weeks from hatching to maturity."

The sign of the net impact of cattle grazing on grassland invertebrate populations is not clear, and there's a decent chance that grazing actually increases total invertebrate abundance on pastures. So the springtail ΔN estimate should be taken with a grain of salt. Cattle farming seems especially likely to increase invertebrate populations when it's done in dry regions requiring irrigation, since irrigation seems likely to increase plant productivity and thus heterotroph numerosity.


Values of D are taken from the table here.


Beef cattle on Eefie Hill - - 236732I made up values of S based on my own guesses about how bad various lives seem. Most people agree that beef cows probably suffer less per day than more confined farm animals. I gave springtails the same amount of per-day suffering as beef cows (not adjusted for differences in sentience between the two types of animals) because both animals spend most of their time grazing outdoors. One could argue that springtails suffer more than beef cows due to spending a greater fraction of their lives dying, but springtails aren't branded/castrated/etc. the way many beef cows are.


I made up the values of R, though they're partly informed by brain-size comparisons.

I'm uncertain how many neurons a springtail has, but below are some considerations:

  • Fruit flies have ~250,000 neurons and are 3-4 mm long (average = 3.5 mm). Meanwhile, springtails "are normally about 1 mm long." Assuming fruit flies are ~3.5 times longer than springtails in length, width, and height, fruit flies are ~(3.5)3 = 43 times as big. If neuron count scales linearly with size, springtails would then have 250,000 / 43 = 5800 neurons.
  • Caenorhabditis elegans is "about 1 mm in length" (about the same length as a springtail), but it's far more narrow than a springtail. C. elegans has 302 neurons. Given that even small springtails are probably several times larger than C. elegans, springtails probably have at least a few thousand neurons.

Meanwhile, this page gives neuron estimates for cows and chickens. Combining this information gives the following table:

Animal Approximate number of neurons R calculated using linear weighting by neurons R calculated using square-root weighting by neurons
Cow 3,000,000,000 500,000 700
Chicken 221,000,000 40,000 200
Springtail ~6000(?) 1 1

Examples of how I calculated the last two columns:

  • For linear weighting by neurons, a chicken has the sentience of 221,000,000 / 6000 = ~40,000 springtails.
  • For square-root weighting by neurons, a chicken has the sentience of sqrt(221,000,000) / sqrt(6000) = ~200 springtails.

I gave pigs more weight than their brain sizes would suggest, because pigs are reputed to be extremely intelligent, perhaps more so than cows.

Selectively reducing chicken

The obvious answer to the tradeoff posed in this piece is to encourage people to eat less chicken but not necessarily less beef. This also makes sense on the grounds that chickens produce much less meat per individual than bovines do.

Unfortunately, most ways to promote vegetarianism at scale involve promoting abstention from all forms of meat at once. Indeed, some people give up red meat and eat more chicken in response to animal-welfare, environmental, or health concerns.


  1. In this piece and in general, I use "cow" in a gender-neutral way as a singular form of "cattle".  (back)