3 ms·

As someone who actually does have a bit of knowledge of genetics and molecular biology, I couldn't get through all the fluff. What are the actual news presented

As someone who actually does have a bit of knowledge of genetics and molecular biology, I couldn't get through all the fluff. What are the actual news presented here?

Scaevolus 8y ago
It seems to be "genetics are very complicated, and because many traits aren't Mendelian and polygenic additive models are oversimplified, they are useless." Mostly, it seems to copy common misconceptions about statistics. The people working with polygenic models understand that the models don't perfectly represent reality, but that doesn't make them useless! They give estimates for various traits (like height) that have been tested to correctly explain some fraction of the variance in a population. Models aren't reality, hence the name!
api 8y ago
TL;DR: we don't understand genetics well enough to make definite statements about the degree to which genes determine phenotypes, how this happens, how malleable or environmentally influence-able it is, etc. As a result we should not dogmatize statistical inferences from genomics and heritability studies, especially if such dogmas influence policy or prejudice. I studied genetics quite extensively and none of that is news to me at all. Dogmatic genetic determinism is Dunning-Krueger effect territory, not because it's impossible but because we don't know enough to be certain enough about it to make it useful. Trying to assert otherwise is wishful thinking, usually by people who really wish there were a way to make scientifically credible assertions about the value of human beings based on their genetic lineage (hint hint). The best way I can describe what seems to be happening with genes and development is that it's something loosely analogous to procedural world generation in games. You have this genetically encoded series of very complex chemical structures that when embodied in a given environment and "executed" via energetic processes generate a phenotype. The phenotype is the product of a set of genes embodied in an environment. Beyond protein structure genes do not contain any kind of "blueprint" for anything any more than e.g. a fractal equation actually contains its complete output. The blueprint analogy is even wishy washy in terms of direct protein structures due to things like RNA transformations, epigenetics, proteins influencing how other proteins form, etc.
- Kim_Bruning 8y ago
  I really like the idea of comparing procedural generation to genetics! I'll use that myself next time I need to explain things.

As someone who actually does have a bit of knowledge of genetics and molecular biology, I couldn't get through all the fluff. What are the actual news presented here?

Scaevolus 8y ago

It seems to be "genetics are very complicated, and because many traits aren't Mendelian and polygenic additive models are oversimplified, they are useless." Mostly, it seems to copy common misconceptions about statistics. The people working with polygenic models understand that the models don't perfectly represent reality, but that doesn't make them useless! They give estimates for various traits (like height) that have been tested to correctly explain some fraction of the variance in a population. Models aren't reality, hence the name!

api 8y ago

TL;DR: we don't understand genetics well enough to make definite statements about the degree to which genes determine phenotypes, how this happens, how malleable or environmentally influence-able it is, etc. As a result we should not dogmatize statistical inferences from genomics and heritability studies, especially if such dogmas influence policy or prejudice. I studied genetics quite extensively and none of that is news to me at all. Dogmatic genetic determinism is Dunning-Krueger effect territory, not because it's impossible but because we don't know enough to be certain enough about it to make it useful. Trying to assert otherwise is wishful thinking, usually by people who really wish there were a way to make scientifically credible assertions about the value of human beings based on their genetic lineage (hint hint). The best way I can describe what seems to be happening with genes and development is that it's something loosely analogous to procedural world generation in games. You have this genetically encoded series of very complex chemical structures that when embodied in a given environment and "executed" via energetic processes generate a phenotype. The phenotype is the product of a set of genes embodied in an environment. Beyond protein structure genes do not contain any kind of "blueprint" for anything any more than e.g. a fractal equation actually contains its complete output. The blueprint analogy is even wishy washy in terms of direct protein structures due to things like RNA transformations, epigenetics, proteins influencing how other proteins form, etc.

Kim_Bruning 8y ago
I really like the idea of comparing procedural generation to genetics! I'll use that myself next time I need to explain things.