The Humane League Needs Your Money More Than Alt Proteins, Probably

Should you donate to animal welfare nonprofits or invest in alt proteins?  It’s a difficult question to answer because there are so many assumptions that go into any head-to-head comparison.  But like all ancient philosophers, I turned to Excel for moral enlightenment.  Specifically, I built a model and found that under most assumptions, you’re probably better off donating to animal welfare nonprofits.  I need to give a fair bit of context, and then I’ll dive into the model, which can be found here

PF vs CC Impact US.xlsx

The first bit of context is a crash course in precision fermentation.  If you like beer, you’re already familiar with fermentation, but scientifically this is what’s happening.  Yeast eat grains and “digest” them into alcohol, basically by rearranging the old atoms into new molecules.  Yogurt, kombucha, and tempeh are produced in the same way- microbes eat one food and process them into another.  That’s all fermentation is.  Precision fermentation is the same process, except the microbes are genetically engineered to produce the exact protein of interest, literally down to the individual nucleic acids in the microbe’s RNA.  In essence, we can “program” microbes to digest plant proteins into animal proteins.  (Which is exactly what animals do.)  Then we can dry that protein into a powder and sell it as whey protein, or mix it with water and sell it as egg whites, or make it mimic animal products in other cool ways.

Precision fermentation is a lot easier technologically than cultivated meat.  Microbes grow faster than animal cells, they can withstand greater stresses like vigorous mixing and temperature variation in the fermentation tank, and at least in the US, they have to jump through fewer regulatory hurdles.  A cheap cultivated chicken fillet might still be a decade (or three) away, but you can buy only-slightly-overpriced precision fermented whey protein from Perfect Day right now.

The egg industry is insanely bad, even by factory farming standards.  (Aren’t transitions fun?)  In the first tab of my Excel sheet, “Suffering from Eggs,” I attempted to quantify exactly how much suffering the egg industry causes.  Welfare Footprint Project has documented the average amount of time a hen spends in 4 different intensities of pain- annoying, hurtful, disabling, or excruciating.  There’s been a lot of debate on this forum about how exactly to weigh those intensities, so I decided to use a conservative estimate from Rethink Priorities, and a liberal estimate from Vasco Grilo (both cited in the Excel sheet.)  Then I used the same method to quantify the suffering the broiler chicken industry causes, again using weights from both Rethink Priorities and Vasco Grilo.  Under both models, the broiler chicken industry only causes about 2x as much suffering as the egg industry.  This is somewhat counterintuitive- after all there are a lot more broiler chickens- but the hens live a lot longer.  Welfare Footprint Project estimates a caged hen spends 431 hours in disabling pain, whereas a broiler chicken spends just 50.  So it shouldn’t be that surprising that the egg industry is almost as bad as the broiler chicken industry.

So compared to other industries, investing in the alt protein egg industry probably gives you the most ethical bang-for-your-buck.  The broiler chicken industry produces slightly more suffering than the egg industry, but it should take way more money to develop cultivated chicken than precision fermented eggs.  The dairy industry should be about as technologically easy as the egg industry to replace with precision fermentation, but it produces way less suffering.  The cow, pig, and sheep industries are the worst of both worlds- cultivated beef will be very expensive to develop, and it won’t offset that much suffering.  The alt protein egg industry is probably the most efficient, meaning the least amount of investment will probably offset the most suffering.  So, I decided to model just the egg industry.  I also restricted my model to the US just because the USDA keeps good data.

One last bit of context, and then I promise I’ll explain my model.  Over the last ~9 years, The Humane League, Mercy for Animals, and many other nonprofits have been working tirelessly on hen welfare corporate campaigns and legislation.  They’ve seen an incredible amount of success- moving the American egg industry from barely cage-free in 2014 to almost half cage-free today.  Using the same methodology I explained 2 paragraphs ago, I used data from Welfare Footprint Project and weights from Rethink Priorities and Vasco Grilo to find that a cage-free hen is only 36-37% as miserable as her caged counterpart.  That’s a huge welfare improvement, and if we’re utilitarians, it means taking 2 hens out of the supply chain completely is about as good as taking 3 hens out of battery cages.  Ok, that’s enough context.  Let’s dive in!

Because the egg industry is such a good alt protein investment, my original question can be boiled down to “Will $100k offset more suffering if invested in precision fermented eggs or donated to an effective hen welfare nonprofit?”  I tried to answer that question by first predicting what the egg industry will look like counterfactually (if you don’t use the $100k to help hens at all.)  Even without the $100k, hens will continue to move from battery cages to cage free aviaries, and eventually both will start to be replaced by precision fermentation.  Visually, if red is caged hens, yellow is cage free hens, and green is precision fermented “hen-equivalents,” it should look like this:

(I’m defining a hen-equivalent as an amount of precision fermentation capacity that replaces 1 hen.  For example, if a hen lays 320 eggs, and a fermentation tank produces 3200 eggs, the tank is 10 hen-equivalents.  It’s a weird unit but, then again, this is the EA forum.)

You could reasonably object that you disagree with the shape of my graph.  Maybe you expect precision fermented eggs to never fully replace eggs from hens, maybe you expect them to be scaled / accepted by consumers faster or slower than I’ve modeled, or maybe you expect battery cages not to drop off so suddenly.  Below there are 4 graphs that show other possible curve shapes.

But conceptually, we’d expect these graphs to at least be in the right ballpark.  Demand for eggs will continue to rise gradually.  Battery cages will continue to give way to cage-free aviaries, both of which will eventually give way to precision fermentation.  The exact shape of the graph doesn’t affect my model too much (though it certainly has some effect), but this is how I modeled everything.

The 2 interventions I’m proposing- funding cage-free initiatives or precision fermentation- can be visualized on the graph.  Funding cage-free initiatives will cause a little downward jump in proportion of hens living in battery cages, and funding precision fermentation will cause the adoption of precision fermented eggs to accelerate slightly.  Graphically, here’s what those look like.

Purple is funding advocacy nonprofits, blue is funding precision fermentation.  (And I apologize for my lack of image editing prowess- I’m really more of a numbers guy :)

So in order to quantify the impact of each intervention, we basically just need to figure out the shape of each region of the graph, the shift caused by each intervention, and compare.  Most of the rest of this post is a description of how I did that.

The first step in figuring out the shape of the graph is defining the top curve- basically how many hens (or hen-equivalents in precision fermentation capacity) will exist each year.  We can multiply the US population by the per-capita egg consumption to determine the total number of eggs eaten in any year, and then divide them by the number of eggs a hen lays over her lifetime to figure out how many hens are needed to produce those eggs.  It makes sense to separate per-capita consumption from population because population trends can change suddenly, whereas cultural ideas about food usually don’t.  In the second tab of the Excel sheet, I found that American egg consumption per capita has trended up very slightly in the last few decades, from ~250 eggs per person in 2000 to ~280 eggs per person in 2024.  I just ran a linear regression and extrapolated, figuring even if it eventually levels off, the error should be pretty small.  Then I pulled projected US population data from a UN database for every year until 2100.  (Obviously this isn’t an exact science, but the UN is probably going to be more accurate than anything I could do in Excel.)  Multiplying eggs per capita by population in each year, I found the expected total number of eggs America will eat every year between now and 2100.  Then dividing by the number of eggs each hen lays over her lifetime, I found the number of hens in the US over time should approximately follow this curve.

(This is the number of hens needed to lay all of one year’s eggs, not the number of hens alive at any one time.)  But this curve is the upper bound of the graph: green + yellow + red.

Next, we need to predict how many hens will continue to live in battery cages counterfactually (without a donation to an effective hen welfare nonprofit.)  This defines the red region of the graph.  In the third tab of the Excel sheet, I pulled data from a Market News Report on the percentage of American hens who were cage-free in each year from 2007 - 2018.  More updated data must exist somewhere but I was having a hard time finding it.  From 2007 - 2014, the percentage of hens who were cage-free bounced around between 3-6% with no clear trend.  Then, starting in 2014, when more nonprofits began effective cage-free advocacy, the percentage shot up and kept rising.  From 2014 to 2018, the data follows a clear linear trend with an increase of a few percent each year.  I ran a linear regression and found the percentage of hens who are cage-free increased 3.34% per year with an R^2 value of 0.98.  But can we really extrapolate data from those 5 years far into the future?  If we extrapolate the trendline out, we find we’d expect to hit 40% cage-free about a third of the way through 2024.  And in fact, in April of 2024, HSUS reported that 40% of American hens were cage-free.  So it’s certainly not an exact science, and the trend could deviate in either direction in the future, but based on the data we do have, it seems fair to think the data will keep following that linear trend, with 3.34% of hens per year moving out of battery cages.

How can we understand this data when so many factors affect whether a company or state will ban battery cages?  One way to think about it is there are a lot of reasons we’d expect the data not to be linear, and they might cancel each other out.  For example, on the this-should-probably-be-accelerating side of the equation, nonprofits become more skilled at corporate campaigns and ballot initiatives as they do them more, funding for cage-free initiatives has increased over time, and producers who sell to multiple states or companies might become entirely cage-free if only some of their customers demand cage-free eggs.  On the this-should-probably-be-decelerating side of the equation, nonprofits may have already targeted and beaten the easiest companies, some states don’t allow ballot initiatives, and inflation may make consumers more unwilling to pay higher prices for cage-free eggs.  It’s kind of surprising that these trends would all cancel each other out, but that seems to fit the best with the data we have.  So, I basically just graphed a straight line, predicting the entire American egg-laying hen population will be cage-free by 2042.

With the total number of hens in each year defined, and the percentage of hens who will be cage-free in each year defined, the next step is determining the number of hens who will be replaced by precision fermentation each year.  This is probably the least precise part of the model because it’s just so difficult to predict how a new technology will mature and sell.  But (at least according to Google), the Bass Diffusion Model is the standard way to predict the spread of a new technology.  Mathematically, it’s a differential equation with a solution that looks like this:

But on the off chance that the inspiring F(t) function didn’t evoke a vivid mental image in your mind’s eye, here’s what’s going on.  New technology spreads in an S curve that looks like this (source: Wikipedia)

There are 3 constants that define the shape of the curve: M, p, and q.  M defines the maximum number of people willing to adopt the new innovation.  For example, if 10% of Americans play baseball, a new baseball bat will have an M of 10% of Americans, or ~33 million people.  Even if everything goes perfectly, adoption of the new technology will never exceed M.  The constant p is the coefficient of innovation, or the proportion of M that will spontaneously adopt the new product in each year.  Maybe only a small percentage of baseball players are willing to buy the new bat when none of their teammates have it.  And q is the coefficient of imitation, or the proportion of the remaining population that will adopt the new product if their friends / teammates have it in each year.  Maybe a larger percentage of baseball players are willing to buy the new bat once they’ve seen their teammates use it.  Plugging those 3 numbers into the above equation yields an S curve that models adoption of a new product.

So I needed to use M, p, and q, plus the year precision fermented eggs will even begin to be sold at scale, to define the shape of the precision fermentation curve.  The EVERY Co seems to be the precision fermentation egg company that’s the furthest ahead.  They are currently just fermenting egg whites and mixing them with plant-based yolks to make a full egg replacement, and they’re not able to sell at scale or price parity yet.  But they do have FDA approval to sell their egg whites, and they are selling to a few companies in very limited quantities.  So they seem like the likely frontrunner to commercialize precision fermented eggs at scale.  Now this isn’t even close to a proper citation, but I remember hearing on a podcast that EVERY plans to launch their first full-scale plant in 2028.  I’ve spent an embarrassing amount of time trying and failing to figure out where I heard that, and I don’t remember if they were planning to begin construction or finish construction in 2028.  (An important difference :)  But I picked 2028 as the first year where a significant number of precision fermented eggs will be sold.  I set M to 100% of the US population, which is necessarily an overestimate.  And after a lot of unhelpful google results, I asked Chat GPT when it thinks half the American egg industry will be precision fermented rather than laid by hens, to which it said “A 50-50 split could realistically be reached between 2040 and 2050, assuming… [a buncha stuff].”  So I set p and q such that the 50-50 split is reached in ~2046.  And after all that fun, the graph looks like this:

I will be the first to admit, there are a lot of things that could be wrong with this model.  Intuitively it seems like precision fermented eggs make too much progress too fast- I don’t think they will basically have finished scaling up by 2070.  Consumer acceptance is too high- it will almost definitely not be 100%.  And it’s very reasonable to think cage-free progress will speed up or slow down at some point between now and 2042.  That said, my goal in making this model wasn’t to exactly predict the exact trajectory of the egg industry, it was to compare the relative benefits of precision fermented eggs and cage-free advocacy.  And as I will show in the next few paragraphs, I found that even after making very generous assumptions about precision fermentation, cage-free advocacy is probably still a better use of resources.  But if you disagree, please feel free to download my Excel sheet and play with the numbers.  I think you will see what I saw- that almost no matter how you set up the problem, cage-free advocacy is still better.

Now we have a baseline model of the number of caged and cage-free hens in the egg industry over time.  It’s time to evaluate the impact of putting $100k towards each intervention.

In the fourth tab of the Excel sheet, “BI Cost,” I estimated the marginal impact of each dollar spent on cage-free advocacy.  Corporate campaigns seem to be more effective than ballot initiatives (https://docs.google.com/document/d/1p7xqop2FlIF8Kw45za0NnJPwvUA70Mb1UzjijMRKRr8/edit?tab=t.0#heading=h.cb26clwxwwow), though they may be slowing down.  So a strategic nonprofit could run corporate campaigns only until they become less effective than ballot initiatives, and then switch to ballot initiatives.  In this way, ballot initiatives establish a “floor” or lower bounds of advocacy effectiveness.  And only 2 ballot initiatives, CA Prop 12 and MA Question 3, explicitly mandated that hens be cage-free.  So I used Laura Duffy’s / RP’s report to evaluate the impact of CA Prop 12 and MA Question 3, and I modeled only the less effective one, MA Question 3.  Because $4.7M were spent on MA Question 3, which liberated ~4M hens per year from battery cages, the expected impact of a $100k donation is ~85,000 hens per year.  And again, this is almost definitely an underestimate, because that money would probably liberate even more hens from battery cages if it went to corporate campaigns or more effective ballot initiatives like CA Prop 12.

There is also probably a lag between when you donate and when factory farmers phase out battery cages.  Most corporate campaigns ask for compliance by some date a few years in the future, and ballot initiatives usually run on a 2 or 4 year cycle.  So in the second to last tab of the Excel sheet, “Full Model- BI,” I assumed a donation now would not have an impact until 2028.  Then, I modeled a one-time jump in the graph, where ~85,000 hens, or ~0.026% of the American flock, became cage-free.  This model averts ~550,000 “hen-life-equivalents” of suffering overall relative to the baseline model.  A hen-life-equivalent is the amount of suffering a caged hen experiences over her lifetime, or about 2.5x the amount of suffering a cage-free hen experiences over her lifetime.  In theory it could be converted into hours in pain, or even QALYs.  But the actual unit doesn’t matter, as long as it can be compared to the impact of precision fermented eggs.

We’re almost done, but let’s recap briefly.  We now have a model for the number of hens, caged and cage-free, who will be subjected to the egg industry for the next ~75 years.  We have a model for the impact of a $100k donation to an effective hen welfare advocacy nonprofit, which will spare about half a million hens from battery cages over the next few decades.  Now we need a model for the impact of an $100k investment in precision fermented eggs, and we can finally compare.

Rather than looking at the entire precision fermentation industry, I again decided to just focus on EVERY.  Their production strategy consists of 2 distinct steps- R&D and scaling- and there are 3 ways to think about the impact of an investment.

1. The free market will invest the right amount of money both during the R&D and scaling stages.  For example, in early R&D there’s a lot of risk, so the market will invest limited money.  As EVERY’s R&D progresses, the risk will get lower and lower, so more and more money will pour in.  If you invest any money at all, it will be above the market’s setpoint, so other investments will slow for a while until their R&D progress catches up to their budget.  Under this model, the impact of an investment is literally nothing.  Fortunately I think this model is wrong :)
2. The free market will underinvest in both the R&D and scaling stages.  The market is only optimizing for profit, but you are co optimizing for hen welfare too.  For example, if EVERY builds their first full-scale production facility and has a payback period of 8 years, a purely selfish investor might put their money somewhere with a faster payback period.  But you might invest in EVERY anyways, allowing them to build another full-scale production facility.
3. The free market will underinvest only in the R&D stage.  But once investors see the technology is derisked, they will be willing to pour money in to scale up.  Under this model, R&D investments are part financially responsible, part philanthropic.  Thus the impact of an investment is its acceleration of the R&D phase, but not of the scaling phase.

Obviously the first model doesn’t help us, and I talked to 4 different people with advanced degrees in econ to try to figure out which model I should use, but none of them agreed.  So, in the spirit of being bullish on precision fermentation, I decided to model both and pick the most optimistic one.  In the fifth tab of the Excel sheet, “PF Cost,” I did just that.

1. Impact is 0.  Sad times.
2. I tried to estimate EVERY’s total expected spending between when they were founded in 2014 and when, according to my model, they will have replaced 99% of the American egg industry in 2069.  I used a report from Lever VC to estimate the cost and productivity of one precision fermentation facility, which should cost $150M and replace a full ~1% of the American egg industry.  I also ignored all R&D and operating costs because they’re likely to be small relative to capital expenses for scaling, and I assumed the cost of egg yolks, water, and packaging to be 0.  Both of these assumptions should make the model overestimate the impact of the investment.  Then, I used a simple spending to time ratio to figure out the benefit of investing $100k.  If $15B of precision fermentation capacity replaces the egg industry in 55 years, apparently $15.0001B of precision fermentation capacity replaces the egg industry in 54.99963 years.  Or, in a more intuitive unit, it shifts the green region left by about 3 hours.
3. I tried to estimate EVERY’s R&D spending between when they were founded in 2014 and when my model thinks they will begin scaling in 2028.  Of course these are really more overlapping processes- most companies invest in R&D well after they’ve begun scaling.  But, again in the interest of overestimating the impact of the investment, I decided to only look at R&D spending before 2028.  There’s no direct point of comparison here (no precision fermentation egg companies have commercialized yet), but I found a Pubmed study that most new pharmaceutical drugs require between ~$300M and $4B in R&D before they’re ready to scale.  This is the closest thing we have to a direct comparison, and precision fermentation is actually somewhat common in drug production, so it seemed like a good reference point.  I assumed EVERY would spend $1B, which is on the lower end of the range.  EVERY has raised $230M in investments so far, plus whatever they’ve made in profits, and R&D expenditures are usually backloaded (testing at a full scale prototype production facility is more expensive than testing at a 10% scale pilot plant, which is more expensive than testing at the benchtop scale.)  Then I did the same thing as I did in the last step, which was basically to take a ratio.  Under this model, the $100k investment shifts the green region of the graph left by about 12 hours.

For yet another reason, both 2 and 3 are an overestimate of the true benefit of your investment, because they look at average, not marginal, impact.  Most new technologies see “learning curves,” where each doubling in production results in a less-than-50% decrease in cost.  For example, Wikipedia has a long entry about learning curves for solar panels, where each doubling of scale has historically reduced costs by ~20%.  For newer technologies, the learning curve can be steeper, but if there are diminishing marginal returns, which there always are, it’s never fully linear.  This is yet another point to be added to the why-my-model-overestimates-the-impact-of-precision-fermented-eggs list.

In the very last tab of the Excel sheet, “Full Model- PF,” I used the more optimistic R&D numbers to estimate the impact of the $100k investment in precision fermenting eggs.  If the entire green region of the graph shifts left by 12 hours, it will save 166,000 “hen-life-equivalents.”

So there you have it.  Under pessimistic assumptions about cage-free advocacy, an $100k donation does as much good as saving 550,000 hens, and under optimistic assumptions about precision fermentation, an $100k investment does as much good as saving 166,000 hens.  And the mathematically-inclined reader will notice that 55 is more than 16, so you should probably donate to cage-free advocacy.  At least according to this model.

There are probably a ton of caveats that haven’t occurred to me, but here are the ones that have.  First, these are the caveats that favor alt proteins:

-Scaling across multiple countries might be easier if one country has already scaled.  For example, if EVERY does a lot of R&D in the US, it will help them scale in China too.

-Success of one kind of alt protein might help others.  For example, maybe tech that EVERY develops will help Perfect Day, or maybe a good reputation EVERY develops will help all alt protein companies.

-Effective lobbyist groups might act as a donation multiplier.  For example, maybe if you donate $1 to the Good Food Institute or Food Solutions Action, they can unlock $10 in federal funding towards alt protein.  A lot of that funding might go towards non-egg alt proteins, but it’s still clearly good.

-Unlike donations to a nonprofit, investments can actually generate a return!  Maybe an investment in precision fermented eggs has a super high expected value because there’s a small chance you get an enormous return, which you can then use for more alt protein investments or advocacy work.

And here are the caveats that might favor advocacy:

-The egg industry is fantastic for alt proteins, but it might not be the best industry for advocacy.  For example, the aforementioned Vasco Grilo estimated that the Shrimp Welfare Project is significantly more effective than hen welfare advocacy.  (source: https://forum.effectivealtruism.org/posts/EbQysXxofbSqkbAiT/cost-effectiveness-of-shrimp-welfare-project-s-humane)  If you are donating rather than investing, you might do the most good helping non-hens.

-Like technology, cage-free advocacy in one country might accelerate the transition to cage-free hens in another country.  For example, maybe Mexico and Canada are more likely to pass cage-free legislation if the US does, or maybe companies in Mexico and Canada will become cage-free so they can export to the US.

-Cage-free hens are slightly more expensive to raise, which both encourages people to buy alternatives like JUST eggs, and makes it easier for alt protein to compete.

That’s a wrap!  To my knowledge, this is the first model comparing advocacy to alt proteins.  There’s no question this is an oversimplification of reality, but if I’m not super wrong, we’re probably better off funding advocacy than alt proteins.  This model has informed my donating decisions this year, and I hope it helps inform others’ too.  Thanks for reading!