Re:Biology is the New Tech: Letter from a conference on CRISPR
by gwern on Sunday July 28, 2019 @10:38AM
(#20547766)
Bear is a good writer, but his genetics are shaky... #2 and #3 are just silly. Doing a little embryo selection or even editing is not going to lead to a race of pod people who all have the same hair style and prefer Earl Gray tea.
Appearance is influenced by millions of variants, which have little or nothing to do on average to do with the variants which influence traits like intelligence (the genetic correlation is ~0 - you don't see genetic correlations between intelligence and well-predicted traits like hair or eye color being reported in the literature).
Even if the same genetic variants which influence something superficial like facial bone structure also influence neurogenesis etc, any editing or selection making people more similar will just be a drop in the pond. Think about siblings: compared to 2 random people, siblings are more similar on what must be tens of millions of genetic variants, and yet, consider how completely different siblings often look, act, and turn out! (Jencks, I think, in _Inequality_ points out that there's almost as much 'inequality' in life outcomes like income within American families as there is across families.)
Further, let's say that all of that turns out to be wrong; nevertheless, parental relatedness to children is desperately desired by parents and this is a major reason why sperm/egg donation are profoundly unpopular and why parents have such stringent demands of donors from banks and why people will deliberately select embryos with Huntington's and other serious disorders if the healthy embryos fail to yield a live birth. It would be entirely possible to simply tilt the selection or editing to preserve relatedness to parents on visible attributes - 'relatedness to parents' is simply yet another metric to optimize for, thrown into the stew with the usual ones, and it is one that parents will put a high value on even at the cost of some potential gains on other metrics.
(#5 is also quite problematic as it assumes that completely novel edits will be done on a mass basis without any kind of testing. In reality, everyone will focus on variants that already are widespread in the population and which are already being tested in the real world. It's the rare variants which are harmful, not the common ones.)
Appearance is influenced by millions of variants, which have little or nothing to do on average to do with the variants which influence traits like intelligence (the genetic correlation is ~0 - you don't see genetic correlations between intelligence and well-predicted traits like hair or eye color being reported in the literature).
Even if the same genetic variants which influence something superficial like facial bone structure also influence neurogenesis etc, any editing or selection making people more similar will just be a drop in the pond. Think about siblings: compared to 2 random people, siblings are more similar on what must be tens of millions of genetic variants, and yet, consider how completely different siblings often look, act, and turn out! (Jencks, I think, in _Inequality_ points out that there's almost as much 'inequality' in life outcomes like income within American families as there is across families.)
Further, let's say that all of that turns out to be wrong; nevertheless, parental relatedness to children is desperately desired by parents and this is a major reason why sperm/egg donation are profoundly unpopular and why parents have such stringent demands of donors from banks and why people will deliberately select embryos with Huntington's and other serious disorders if the healthy embryos fail to yield a live birth. It would be entirely possible to simply tilt the selection or editing to preserve relatedness to parents on visible attributes - 'relatedness to parents' is simply yet another metric to optimize for, thrown into the stew with the usual ones, and it is one that parents will put a high value on even at the cost of some potential gains on other metrics.
(#5 is also quite problematic as it assumes that completely novel edits will be done on a mass basis without any kind of testing. In reality, everyone will focus on variants that already are widespread in the population and which are already being tested in the real world. It's the rare variants which are harmful, not the common ones.)