, 11 tweets, 3 min read Read on Twitter
Re: @jpjjr1961's piece on Quillette and race science. Objects of scientific study can be (1) hard edged—there are three generations of quarks; (2) soft edged—there are two sexes, with intermediate cases; (3) property clusters or groupings.
Objects of type three are often confused with objects of type two. This is an error, and @jpjjr1961's history of science gives a nice example of how this happened. Clustering algorithms take parameters; most obviously, N, the number of clusters you wish to find.
(You may use model selection to choose N, but—outside of the generative models that often define the fitting process—this has all the usual arbitrariness of AIC vs BIC vs Bayesian Evidence, etc.)
Topic modeling in NLP provides a nice example. This is a tool that identifies (quasi) semantic units by inferring underlying co-occurring word patterns ("topics"). Topics *do* track meaning. But a topic unit is not a unit of meaning.
You can vary the number of topics from 2 to 200 (and beyond). Because language, like the genotype-phenotype relation, is complex, and the same objects (words, genes) do multiple things, re-running the model often produces entirely different units.
You can still do science with topic modeling! And we do. Topics at some scale do pick out causal features of the world. We even use them in empirical studies of cultural evolution. But, perhaps because "topic" is not as loaded as "race", we use them correctly.
For example, we often want to make sure that our causal claims are robust to (1) re-running the model fits, and (2) reasonable variations in N.
We would never do this with a concept such as biological sex. Though we would certainly do it to track the rhetoric gender self-identification on Twitter! Freebie project for someone there.
But it is absolutely necessary when we look at the weak and complex selection that often shows up when we want to talk about (biological or cultural) evolution. Selection for lactose or alcohol digestion operates on a type two (or even type one) object.
The things that people want to do with race, of course, are type three. Leaving aside the second failure of (e.g.) a quantity like IQ that is used as a tracer. (@rjhaier tells me that an equivocation between IQ and intelligence is a fallacy. It returns like a vampire.)
Type three objects are much harder to deal with. At the risk of stating the obvious, race does exist—you can watch its causal effects in the 2016 election. It is a socially-constructed type two object.
Missing some Tweet in this thread?
You can try to force a refresh.

Like this thread? Get email updates or save it to PDF!

Subscribe to Simon DeDeo
Profile picture

Get real-time email alerts when new unrolls are available from this author!

This content may be removed anytime!

Twitter may remove this content at anytime, convert it as a PDF, save and print for later use!

Try unrolling a thread yourself!

how to unroll video

1) Follow Thread Reader App on Twitter so you can easily mention us!

2) Go to a Twitter thread (series of Tweets by the same owner) and mention us with a keyword "unroll" @threadreaderapp unroll

You can practice here first or read more on our help page!

Follow Us on Twitter!

Did Thread Reader help you today?

Support us! We are indie developers!

This site is made by just three indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3.00/month or $30.00/year) and get exclusive features!

Become Premium

Too expensive? Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal Become our Patreon

Thank you for your support!