Thanks Michael, this deserves a longer response but a few thoughts on PTC (additional to the discussion in the original post):
I do agree with Jacob / Rethink’s nudge that the alt proteins space needs to explore other theories of change—priorities 3 and to a certain extent 4 in this RFP rely less on this model being true (and more on what it would take for food manufacturers to adjust their formulations away from animal products).
That said, I think there are perhaps two reasons one might underweight the review slightly:
Taste is underspecified IMO. The operationalization (to my memory) is mostly blind or informed hedonic tests on finished products. These are useful inputs but arguably underpowered for the question they’re being asked to answer, and don’t account for population heterogeneity in taste perception. A broader operationalization using analytical chemistry (eg. OAV profiles) suggests current products have made real progress but are nowhere near equivalence to target meats. So “taste-competitive PBM doesn’t displace meat” is plausibly too strong a claim, using the evidence on current products.
On price, I’m less sure cross-price elasticity is the right lens: It’s intuitively appealing, and the Rethink paper uses it sensibly, but I haven’t found a historical food transition where x-price elasticity estimates were good forward predictors of how consumption shares would change with price changes. Chicken vs beef in the latter half of the 20th century is a plausible example—the x-price elasticity estimates tell a pretty confusing story, even though most people’s intuitive read seems to be that chicken at least limited beef’s growth. I think this reflects something about how the framework handles novelty and norm shifts at low penetration.
Ultimately predicting shifts in tastes or technological change over long timescales is hard, and having good-tasting alt meat available seems like a decent bet to have placed under uncertainty.
Writing in my personal capacity. Abhi and I have also discussed the possibility of working together on this topic.
I appreciate your engaging with my review of alt proteins. I agree that taste is somewhat underspecified there. One challenge I ran into is that proponents of improving taste don’t often specify exactly how they think taste equivalence should be operationalized. I tried to cover the range of operationalizations for which I could find evidence, but ended up focusing primarily on human taste tests rather than analytical chemical evidence. My reasoning was twofold. First, taste tests measure taste experience most directly and address the extent to which differences can be perceived. Second, in practice they capture not only gustatory taste, but also texture, aroma, and other sensory characteristics that are often bundled together when people talk about “taste equivalence.”
I’m curious about your comment that these studies are “arguably underpowered.” Do you mean that, for example, the confidence intervals in the NECTAR taste tests are too wide to support meaningful conclusions about hedonic taste equivalence? Or are you pointing to something else?
Having only recently started looking more closely at the analytical chemistry literature, I have a few tentative thoughts:
Some operationalizations in this space seem quite demanding. Meat contains an enormous number of compounds (likely millions/billions), so demonstrating equivalence across all (or even most) of them may be difficult. Restricting attention to flavor-relevant compounds above perceptual thresholds would help, though that also raises the issue of measuring those thresholds for many compounds.
More broadly, I’m not yet convinced that this level of chemical equivalence is the right target. It seems plausible that there is a fairly large space of flavors consumers would find highly appealing, even if many do not closely resemble conventional meat. This seems especially plausible given that consumers themselves have only sampled a small fraction of the possible flavor space even within animal products.
In any case, analytical chemistry would also need to be complemented by measures of texture and mouthfeel. That’s not necessarily an objection, but it does make the overall evaluation more involved.
On cross-price elasticity, I agree it isn’t generally intended to predict long-run changes in market share unless the underlying demand structure remains stable while prices change—a fairly strong assumption. As you say, predicting these dynamics is hard. That said, given the current evidence on cross-price elasticities for alt proteins—including some apparently conflicting findings about whether lower prices increase or decrease meat consumption—I’m not sure I have strong intuitions about how a long-run prediction follows from that data. For beef/chicken cross-price elasticity estimates, is there a particular reference you have in mind?
Overall, I’m glad to see this RFP: as you know, I agree there’s substantial value in better measuring and improving alt protein taste, and I’m looking forward to seeing more work in this area.
Those are all good points, and to be sure, I do expect that the development of non-animal protein that satisfies PTC will reduce meat consumption. By just how much, I agree that we can’t predict, as it depends on many other factors. I strongly support the development of good alt-meat.
But I think some other historical analogies are instructive. For example, with aquaculture and wild-catch fishing and with low-carbon energy and fossil fuels, there have been major rebound effects that made these substitutes much less effective as replacements than one might have predicted in advance. I expect a similar thing to happen with alt-meat.
Thanks Michael, this deserves a longer response but a few thoughts on PTC (additional to the discussion in the original post):
I do agree with Jacob / Rethink’s nudge that the alt proteins space needs to explore other theories of change—priorities 3 and to a certain extent 4 in this RFP rely less on this model being true (and more on what it would take for food manufacturers to adjust their formulations away from animal products).
That said, I think there are perhaps two reasons one might underweight the review slightly:
Taste is underspecified IMO. The operationalization (to my memory) is mostly blind or informed hedonic tests on finished products. These are useful inputs but arguably underpowered for the question they’re being asked to answer, and don’t account for population heterogeneity in taste perception. A broader operationalization using analytical chemistry (eg. OAV profiles) suggests current products have made real progress but are nowhere near equivalence to target meats. So “taste-competitive PBM doesn’t displace meat” is plausibly too strong a claim, using the evidence on current products.
On price, I’m less sure cross-price elasticity is the right lens: It’s intuitively appealing, and the Rethink paper uses it sensibly, but I haven’t found a historical food transition where x-price elasticity estimates were good forward predictors of how consumption shares would change with price changes. Chicken vs beef in the latter half of the 20th century is a plausible example—the x-price elasticity estimates tell a pretty confusing story, even though most people’s intuitive read seems to be that chicken at least limited beef’s growth. I think this reflects something about how the framework handles novelty and norm shifts at low penetration.
Ultimately predicting shifts in tastes or technological change over long timescales is hard, and having good-tasting alt meat available seems like a decent bet to have placed under uncertainty.
Writing in my personal capacity. Abhi and I have also discussed the possibility of working together on this topic.
I appreciate your engaging with my review of alt proteins. I agree that taste is somewhat underspecified there. One challenge I ran into is that proponents of improving taste don’t often specify exactly how they think taste equivalence should be operationalized. I tried to cover the range of operationalizations for which I could find evidence, but ended up focusing primarily on human taste tests rather than analytical chemical evidence. My reasoning was twofold. First, taste tests measure taste experience most directly and address the extent to which differences can be perceived. Second, in practice they capture not only gustatory taste, but also texture, aroma, and other sensory characteristics that are often bundled together when people talk about “taste equivalence.”
I’m curious about your comment that these studies are “arguably underpowered.” Do you mean that, for example, the confidence intervals in the NECTAR taste tests are too wide to support meaningful conclusions about hedonic taste equivalence? Or are you pointing to something else?
Having only recently started looking more closely at the analytical chemistry literature, I have a few tentative thoughts:
Some operationalizations in this space seem quite demanding. Meat contains an enormous number of compounds (likely millions/billions), so demonstrating equivalence across all (or even most) of them may be difficult. Restricting attention to flavor-relevant compounds above perceptual thresholds would help, though that also raises the issue of measuring those thresholds for many compounds.
More broadly, I’m not yet convinced that this level of chemical equivalence is the right target. It seems plausible that there is a fairly large space of flavors consumers would find highly appealing, even if many do not closely resemble conventional meat. This seems especially plausible given that consumers themselves have only sampled a small fraction of the possible flavor space even within animal products.
In any case, analytical chemistry would also need to be complemented by measures of texture and mouthfeel. That’s not necessarily an objection, but it does make the overall evaluation more involved.
On cross-price elasticity, I agree it isn’t generally intended to predict long-run changes in market share unless the underlying demand structure remains stable while prices change—a fairly strong assumption. As you say, predicting these dynamics is hard. That said, given the current evidence on cross-price elasticities for alt proteins—including some apparently conflicting findings about whether lower prices increase or decrease meat consumption—I’m not sure I have strong intuitions about how a long-run prediction follows from that data. For beef/chicken cross-price elasticity estimates, is there a particular reference you have in mind?
Overall, I’m glad to see this RFP: as you know, I agree there’s substantial value in better measuring and improving alt protein taste, and I’m looking forward to seeing more work in this area.
Those are all good points, and to be sure, I do expect that the development of non-animal protein that satisfies PTC will reduce meat consumption. By just how much, I agree that we can’t predict, as it depends on many other factors. I strongly support the development of good alt-meat.
But I think some other historical analogies are instructive. For example, with aquaculture and wild-catch fishing and with low-carbon energy and fossil fuels, there have been major rebound effects that made these substitutes much less effective as replacements than one might have predicted in advance. I expect a similar thing to happen with alt-meat.