Beavers, like all species, are convenient fiction. Thinking of beavers as true category in nature is pre-Darwinian.
Beavers, instead, are a number of animals that cluster together quite closely in an n-dimensional space on a large number of features.
Outrageous? Bear with me.🧵
Beavers, instead, are a number of animals that cluster together quite closely in an n-dimensional space on a large number of features.
Outrageous? Bear with me.🧵
What are these features? They include things like length, hairiness, intelligence, number of limbs, distance of ears to claws, and so on.
These features cluster together because they are causally related (often in complex ways).
These features cluster together because they are causally related (often in complex ways).
Imagine this 2-feature plot, except with 3.7 billion features.
You can see that many elephants and many beavers clusters on 2 features. You can also see an outlier elephant (lots of hair) and an outlier beaver (exceptionally heavy).
You can see that many elephants and many beavers clusters on 2 features. You can also see an outlier elephant (lots of hair) and an outlier beaver (exceptionally heavy).
You would never mistake a beaver for an elephant, even if it were an n-dimensional super-outlier beaver. But our "beaver" may be thought of as another species, if the features were in the middle of two clusters.
This beaver could possibly breed with that other species.
This beaver could possibly breed with that other species.
This is, I believe, how major depression or schizophrenia work. Boyd calls these groupings "homeostatic property clusters".
What I am trying to figure out now is how to determine prototypical beaverness. Do you know work on the topic? Thanks.
What I am trying to figure out now is how to determine prototypical beaverness. Do you know work on the topic? Thanks.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
