I don’t know what most of the biology terms mean. Maybe you should link this to ‘a primer on the biology of cultured meat etc’?
I’d also like to better understand the methods used in these analyses in terms of the statistics/data sci/economic stuff. Like
what are these ‘9 parameters that become 4 factors’ in the third paper?
How do their estimates and ranges work? Are they incorporating uncertainty in Bayesian/Fermi-Monte-Carlo-ish ways?
Future work could be done in the context of an explicit overall calculation/guesstimate, with interactive visualizations to help us understand how each factor affects the total cost estimate
What would be nice would be to be able to have a sort of ‘production function’ for cultured meat depicted and explained somewhere.
Which are fixed costs
Which things add
which things multiply
which are substitutes in the process
what are the physical constraints, etc.
Some of this may be out of the scope of an EA forum post, but this question is so important that we should not be limited by this format IMO.
I don’t know what most of the biology terms mean. Maybe you should link this to ‘a primer on the biology of cultured meat etc’?
I’m not aware of any really good primers. For what it’s worth, I’ve usually found the first page of Google search to be sufficient most of the time, and if not, pinging my biologist friends with targeted questions is sometimes useful. Of course, this might be too high a bar to reasonably expect for someone who’s only reading our summaries; Neil and I will consider whether it’s worthwhile to make a good vocabulary list.
I still don’t know what all the terms in cultured meat means, and sometimes in practice what ends up happening is that I treat terms like what Scott calls “crenulate their zeugma”as placeholders and I just cross my fingers and hope it isn’t actually that important. (I genuinely suspect they aren’t, for what it’s worth).
A bigger problem with a lack of understanding, to me, isn’t the terms but the underlying scientific details. cf Feyman’s quote. Unfortunately I don’t have a good way around this.
I’d also like to better understand the methods used in these analyses in terms of the statistics/data sci/economic stuff. Like
I’ll let Neil answer about the Risner (third) paper specifically, but the short answer is that I don’t think the treatment of statistics or probability in any of these papers is particularly impressive.
Future work could be done in the context of an explicit overall calculation/guesstimate, with interactive visualizations to help us understand how each factor affects the total cost estimate
Yeah agreed. I think a really good TEA that would be very useful for EA folks would require someone or a team of people to have the following:
A deep understanding of the scientific, engineering, and economic problems involved
A commitment to truth-seeking, high rigor, and no “axe to grind”
A reasonable formal or intuitive understanding of probability and statistics.
Unfortunately, the existing analyses so far only has 0-2 of the above. Neil and I are likewise missing #1 on this list (I’m sure other people may contest whether we have #2 or #3 as well!)
Re
What would be nice would be to be able to have a sort of ‘production function’ for cultured meat depicted and explained somewhere.
Yeah I can see how this will be valuable. The Hughes review also complained about under-specificity. The main problem with such an approach is that it’s hard to have a very precise production function for a futuristic technology (which GFI also complained about), though Humbird is the best treatment we’re aware of on that topic.
I’d also like to better understand the methods used in these analyses in terms of the statistics/data sci/economic stuff. Like
I’ll let Neil answer about the Risner (third) paper specifically, but the short answer is that I don’t think the treatment of statistics or probability in any of these papers is particularly impressive.
Makes me think your pessimism might be premature, in the absence of clear and strong evidence?
Main comments/suggestions:
I don’t know what most of the biology terms mean. Maybe you should link this to ‘a primer on the biology of cultured meat etc’?
I’d also like to better understand the methods used in these analyses in terms of the statistics/data sci/economic stuff. Like
what are these ‘9 parameters that become 4 factors’ in the third paper?
How do their estimates and ranges work? Are they incorporating uncertainty in Bayesian/Fermi-Monte-Carlo-ish ways?
Future work could be done in the context of an explicit overall calculation/guesstimate, with interactive visualizations to help us understand how each factor affects the total cost estimate
What would be nice would be to be able to have a sort of ‘production function’ for cultured meat depicted and explained somewhere.
Which are fixed costs
Which things add
which things multiply
which are substitutes in the process
what are the physical constraints, etc.
Some of this may be out of the scope of an EA forum post, but this question is so important that we should not be limited by this format IMO.
I’m not aware of any really good primers. For what it’s worth, I’ve usually found the first page of Google search to be sufficient most of the time, and if not, pinging my biologist friends with targeted questions is sometimes useful. Of course, this might be too high a bar to reasonably expect for someone who’s only reading our summaries; Neil and I will consider whether it’s worthwhile to make a good vocabulary list.
I still don’t know what all the terms in cultured meat means, and sometimes in practice what ends up happening is that I treat terms like what Scott calls “crenulate their zeugma”as placeholders and I just cross my fingers and hope it isn’t actually that important. (I genuinely suspect they aren’t, for what it’s worth).
A bigger problem with a lack of understanding, to me, isn’t the terms but the underlying scientific details. cf Feyman’s quote. Unfortunately I don’t have a good way around this.
I’ll let Neil answer about the Risner (third) paper specifically, but the short answer is that I don’t think the treatment of statistics or probability in any of these papers is particularly impressive.
Yeah agreed. I think a really good TEA that would be very useful for EA folks would require someone or a team of people to have the following:
A deep understanding of the scientific, engineering, and economic problems involved
A commitment to truth-seeking, high rigor, and no “axe to grind”
A reasonable formal or intuitive understanding of probability and statistics.
Unfortunately, the existing analyses so far only has 0-2 of the above. Neil and I are likewise missing #1 on this list (I’m sure other people may contest whether we have #2 or #3 as well!)
Re
Yeah I can see how this will be valuable. The Hughes review also complained about under-specificity. The main problem with such an approach is that it’s hard to have a very precise production function for a futuristic technology (which GFI also complained about), though Humbird is the best treatment we’re aware of on that topic.
Thanks. This
Makes me think your pessimism might be premature, in the absence of clear and strong evidence?