Hey David! This is great, thanks for sharing, I'll take a more detailed look. For now, some comments:
I agree that a more detailed Monte Carlo than the one I did is desirable.
I also think it is reasonable to present a trio of scenarios (neutral, optimistic, pessimistic), though science is somewhat tricky in that what matters is "what is reasonably achievable", not the average of efforts. Eg, we do not measure AI progress by the average AI model, but the best. Thus, my bias towards optimistic figures and towards considering Pasitka's numbers as more relevant than Humbird's. Additionally, the scaling issues also seem more relevant for the bioreactor than for the media, so I don't think a lack of large-scale commercial proof is strong evidence against Pasitka's numbers being achievable. In general, my best guess is that $0.3/L to $0.2/L is likely to be achievable for the media, even if it is not easy. And I think that is ultimately the interesting question to answer.
Hey Bruce, thanks for such a detailed comment. Having also read the book, I very much agree with it.
for consumers, “same meat, less likely to make you sick” or contain drug residues or mercury seems like it could be a pretty compelling message
I think that could be true in case of some contaminated meat scandals, but if people are already used to a perceived level of risk they may not automatically adjust their behavior. It is the same as telling people that motorbikes are so much more dangerous than all other transportation methods; I think most people don't change their behaviour based on that. That is also why I think there is no strong pull incentive for cultivated meat, unfortunately. However, the same applies to EV and yet they seem bound to succeed, so I think the government support strategy is likely correct.
Consumer Motivation 2: animal suffering
I think this is likely a severely underestimated retention effect and also probably (IMHO) the biggest lever we have to cause a societal change: once the societal, health and taste pressure on people severely decreases, people will find it less and less acceptable to eat meat. My theory of change is that it might be hard to obliterate factory farming just on price and other features, but cultivated meat should catalyse a societal moral change by eroding the justifiability of factory farming.
Plant-based meat has failed because almost no one understood the assignment:
I think it makes sense, but I think there is also a possibility that people psychologically distinguish meat from plant-based meat, and just follow inertia and custom to avoid social pressure. There are a lot of people who do not want their identity to be vegetarian. I do not have strong priors here. However, I model this on the webpage https://pabloamc.github.io/Cultivated_meat/interactive.html and let people assign value to meat being identical (including made of animal cells), or even to meat being produced from animals -- which I find significantly more unlikely and is different from neophobia. Significantly more unlikely because they'd have to explicitly put value on killing animals, which seems to me like a less natural psychological separation than "just being meat". In any case, my goal was partly to allow the EA community to see how different priors would lead to different market shares.
I think you miss the much stronger arguments that can really move policymakers: GDP and food security, which are my focus in chapters 9 and 10, respectively (chapter 10 walks through the incentives for China, India, the US, Brazil, Europe, and small tech forward countries (Japan, S. Korea, Israel, Singapore).
Probably you have much better access to information here. The only thing that worries me is that every small field (like quantum, where I work) claims "national security" relevance, and I am not sure the politicians can calibrate things accurately. For example, quantum computing, where I work, is largely not a national security issue, as long as one uses some well-understood cryptographic standards. Also, it so happens that AI can be a blessing for the scientific and technological component of our endeavour, but it will almost certainly eclipse all other topics for the next few years to come. Again, no strong priors; you probably know better.
I think most cuts should be something that we could replicate without significant price difference. Or perhaps scaffolding ends up being a challenge. Also probably cultivated gives you more freedom to explore conventionally infeasible options.
Are you saying "factory farming" an actual axis of variation people will seek? Or are you arguing that since some species are much more challenging to achieve price parity with, people will keep eating those species too? I think the latter makes sense to me.
I have a few other thoughts on your post that I'll try to share later today, re: government incentives to make this happen faster than 10-20 years.
I'd love to hear so. The model I proposed and that resulted from those many years is based on Bass diffusion dynamics with typical parameter values, but of course, I'd love to supercharge the change.
Hey Bruce! Thanks for the reply. This is indeed one point where I was sloppy, though I was not trying to challenge your number. Let me explain my current understanding:
Lewis Bollard argues (and I think it is right) that 2kg of feed converts into approximately 1kg of live animal, of which 2/3 is edible meat.
You argue (and I also think it is accurate) that a chicken needs to eat 9 calories of grain to produce 1 calorie of meat. This is not incompatible with the above because grains contain much less water, which you noted above.
Beyond caloric content, we should also analyse amino acids, which are also used for cell buildup and I did not take into account. I think the amino acid content is the binding constraint to set up a cost floor, and sits at ~20% for both dry feed and wet meat (mostly coincidence).
This is how I think of it:
Let the chicken's FCR be ~2 kg feed/kg live weight (Bollard). Dry feed is ~20% protein (typical broiler ration).
So 1kg of chicken consumes roughly: 2 kg feed×0.20=0.4 kg feed protein per kg live weight
Meanwhile, a kg of (wet) live weight yields ~0.65 kg edible meat at ~20% protein = ~0.13 kg protein in product.
So the chicken's protein conversion is roughly: 0.13 kg protein out / 0.40 kg protein in ≈33%
So I think if we care only about the caloric content, you are right (and I was wrong to use Lewis' number); but if we care about sourcing amino acids from broadly the same plants used for chicken feed, probably there is a cost floor for cultivated meat around 1/3 the cost of chicken meat.
I think I did not do a good job of framing this post. My goal was not to criticise the people working in animal welfare organisations, quite the contrary! I think they are doing very necessary work, and I am really grateful to them, especially given how weird this looks to most people. In other words, I am not claiming that these feelings are right, only that they exist.
The goal of the post was instead to reflect on something I feel (and presumably others feel too), which may be dragging donations to those organisations. In my case, this applies to politics too.
Hey David! This is great, thanks for sharing, I'll take a more detailed look. For now, some comments:
With regards to the bioreactor, it seems progress has been happening there too, (see eg https://parima.bio/blog/parima-demonstrates-ton-scale-cultivated-meat-production-at-viable-unit-economics). Interestingly, they claim to have used a 22k L bioreactor, which is roughly the large scale that Humbird and Pasitka had envisioned. Again, not an insider, but it seems we live in the "optimistic" world.
Last: my $15 to $30 estimate was an eyeball from the results I had, it was just meant to be an anchor for people's minds.
You can also use the forum chat.
Hey Bruce, thanks for such a detailed comment. Having also read the book, I very much agree with it.
I think that could be true in case of some contaminated meat scandals, but if people are already used to a perceived level of risk they may not automatically adjust their behavior. It is the same as telling people that motorbikes are so much more dangerous than all other transportation methods; I think most people don't change their behaviour based on that. That is also why I think there is no strong pull incentive for cultivated meat, unfortunately. However, the same applies to EV and yet they seem bound to succeed, so I think the government support strategy is likely correct.
I think it makes sense, but I think there is also a possibility that people psychologically distinguish meat from plant-based meat, and just follow inertia and custom to avoid social pressure. There are a lot of people who do not want their identity to be vegetarian. I do not have strong priors here. However, I model this on the webpage https://pabloamc.github.io/Cultivated_meat/interactive.html and let people assign value to meat being identical (including made of animal cells), or even to meat being produced from animals -- which I find significantly more unlikely and is different from neophobia. Significantly more unlikely because they'd have to explicitly put value on killing animals, which seems to me like a less natural psychological separation than "just being meat". In any case, my goal was partly to allow the EA community to see how different priors would lead to different market shares.
Probably you have much better access to information here. The only thing that worries me is that every small field (like quantum, where I work) claims "national security" relevance, and I am not sure the politicians can calibrate things accurately. For example, quantum computing, where I work, is largely not a national security issue, as long as one uses some well-understood cryptographic standards. Also, it so happens that AI can be a blessing for the scientific and technological component of our endeavour, but it will almost certainly eclipse all other topics for the next few years to come. Again, no strong priors; you probably know better.
I think most cuts should be something that we could replicate without significant price difference. Or perhaps scaffolding ends up being a challenge. Also probably cultivated gives you more freedom to explore conventionally infeasible options.
Are you saying "factory farming" an actual axis of variation people will seek? Or are you arguing that since some species are much more challenging to achieve price parity with, people will keep eating those species too? I think the latter makes sense to me.
Awesome, thanks, Bruce,
I'd love to hear so. The model I proposed and that resulted from those many years is based on Bass diffusion dynamics with typical parameter values, but of course, I'd love to supercharge the change.
Hey Bruce! Thanks for the reply. This is indeed one point where I was sloppy, though I was not trying to challenge your number. Let me explain my current understanding:
This is how I think of it:
Let the chicken's FCR be ~2 kg feed/kg live weight (Bollard). Dry feed is ~20% protein (typical broiler ration).
So 1kg of chicken consumes roughly: 2 kg feed×0.20=0.4 kg feed protein per kg live weight
Meanwhile, a kg of (wet) live weight yields ~0.65 kg edible meat at ~20% protein = ~0.13 kg protein in product.
So the chicken's protein conversion is roughly: 0.13 kg protein out / 0.40 kg protein in ≈33%
So I think if we care only about the caloric content, you are right (and I was wrong to use Lewis' number); but if we care about sourcing amino acids from broadly the same plants used for chicken feed, probably there is a cost floor for cultivated meat around 1/3 the cost of chicken meat.
Would you agree with this analysis?
Fully agree with everything you point out.
I think I did not do a good job of framing this post. My goal was not to criticise the people working in animal welfare organisations, quite the contrary! I think they are doing very necessary work, and I am really grateful to them, especially given how weird this looks to most people. In other words, I am not claiming that these feelings are right, only that they exist.
The goal of the post was instead to reflect on something I feel (and presumably others feel too), which may be dragging donations to those organisations. In my case, this applies to politics too.
I think they already do it, under the premise of supporting local and rural farmers.