them? Well they brought in Aella_Girl, a high iq hooker, who makes the host look like a tard
https://x.com/pli_cachete/status/1890111857886990723
I’m sure everyone who stats-trolled Aella over the years will be happy to see that she now seems to have a much better grasp of how science and the growth of knowledge work than the average redpill guy.
— Rota 🐜 (@pli_cachete) February 13, 2025
The hypothetico-deductive method and it’s consequences pic.twitter.com/rPO2CkaPPP
Neighbor you do not need a control arm to find correlation.
You can "control" for different factors, but you don't need a dedicated control arm like you think.
I have a buddy who is very similar to this guy but not an oldcel (NCT, mildly Latinx looking, not very high iq (1400 sat but talks about midwits), believes in the redpill, and sends me Save Europe Reels.
Would probably get mogged by this Hooker too
Jump in the discussion.
No email address required.
It's kinda crazy how she's just saying insanely basic high school stats and everyone on the podcast and comments is amazed that she is describing a basic linear regression. And there is still implicitly a "control" in this (using what I assume this guy's definition of a control is). You are saying is this data more likely to follow this distribution then one which is basically random noise. Kinda sad this means having a "much greater grasp of science".
Jump in the discussion.
No email address required.
Neighbor nobody doing actual experiments is diluting the definition of control to this extent. Nobody goes the control is the randomness of the universe. You actively select a control. By this logic every single action taken has a control. Me taking a piss in the woods is a control because it gives more of a direction to pissing than if it was done at random.
Jump in the discussion.
No email address required.
yeah my point is more you need a model to correlation. "Correlation is saying we ca use x to predict y" is the outcome. What the analysis is actually doing is fitting the data to a model and seeing how well the model can explain the data.
Jump in the discussion.
No email address required.
Are you talking about the dependent and and independent variables? That's a much better terminology to use if you're doing an observational study fitting data to a curve rather than a controlled experiment.
Jump in the discussion.
No email address required.
Aella's problem is she sort of lacks a conceptual framework to map whatever calculations she's having the computer do to some quantity we care about in the real world. Like sure height and weight are correlated, why are you doing this and how would you like us to interpret this?
I'm not going to watch the whole 7 hour thing but what was the context of this? Like she does these surveys and I never understand what she's trying to learn about or test or "do science" about.
Jump in the discussion.
No email address required.
You could
s/Aella/gfor basically anyone else in the self-proclaimed "rationalist" community. They're all like 5th graders in the gifted program trying to reinvent the universe from first principles without reading a book. Did you ever see Christopher Langan and his Cognitive-Theoretical Model? Or Curtis Yarvin and his "AI research"/Urbit? Being bright without any grounding in the work of others is a classic recipe for crackpottery.In her defense, I don't think the height/weight thing was a rehearsed point, I think the guy was being a prick, and she was grasping for a ready example of two obviously correlated things where you could easily observe the correlation without doing a controlled experiment. She was (I think) trying to drive home the point that we can gain scientific knowledge from applying statistics for observational studies, even in the absence of a controlled experiment.
Jump in the discussion.
No email address required.
That's fair. Next time she should use my favorite example, college education and earnings, because it's simple, people will have decent intuitions about it, you can actually relate that to which situations you would want a predictive versus causal relationship, what a hypothetical RCT would look like to get a causal relationship, what observational correlations do and do not tell us, and how you might approximate an RCT with the right observational data.
Jump in the discussion.
No email address required.
More options
Context
More options
Context
More options
Context
My point was more you are seeing how well your data fits the assumed model and then seeing the probability that any random data could have just randomly fit this model which should be under 0.05 or 5%. Null hypothesis would be the right term here since your null is that the data is randomly distributed with 0 correlation.
Jump in the discussion.
No email address required.
More options
Context
More options
Context
Yes but none of that is a control. Good talk by the way. Sorry if I came off as aggressive.
Jump in the discussion.
No email address required.
More options
Context
More options
Context
There's "control" like the group in an RCT that doesn't get treated, and there's "control" like you would have in a linear regression with observational data that is conceptually supposed to make the independent variable you care about's assignment "more random" in some sense. You do this because when you do science you don't really care about PREDICTION (what a correlation is), you care about understanding some structural, usually CAUSAL, parameter about the real world and the way you usually have to learn about causation in a lot of these contexts is through random (or quasi-random) assignement of treatment. The dude sort of intuits this tbh and Aella seems to not understand any of this at all, although it's true that she successfully uses more statistics terms correctly than the guy does.
Jump in the discussion.
No email address required.
What you are talking about is referring to a method of randomizing the variable selection as control. Idk if actual researchers refer to that as control but that is still nowhere close to what the guy was talking about. He was calling any variable on a correlation graph by default an automatic control. He didn't intuit anything on the level you think he did lmao.
!biofoids look at this guy reading into an r-slured moid being r-slurred in an attempt to dunk on a foid, caught on vid and misunderstanding it as "intuitively" a genius.
Jump in the discussion.
No email address required.
If you run an RCT often you call the group that gets treated the "treatment" group and the group they doesn't get treated the "control" group. Maybe it's a field-specific thing but I've never heard anyone call that something else. Again though I feel like the moid sort of has the right intuitions for what causal inference is, and Aella is caught up in the mechanics of how you fit a regression line. Both r-slurred imo but for different reasons.
Jump in the discussion.
No email address required.
You first statement is correct.
Causal inference isn't control though.
No, Aella is trying to give an example of a science research tool that is not a control because the moid is saying everything in science is based on a control. I would however agree, that she should have come up with a better example, but to be fair to her, she was being put on the spot and isn't a professional teacher.
Jump in the discussion.
No email address required.
More options
Context
More options
Context
More options
Context
More options
Context
More options
Context
Its amazing to them because none of those people passed middle school math never mind high school.
We're looking at products of public schools here.
Jump in the discussion.
No email address required.
More options
Context
I watched this and also Aella is kind of wrong about a lot of stuff tbh.
Jump in the discussion.
No email address required.
It's like she understands what these things do not what they are. Like yeah a correlation tells us if Y is dependent on X but it's not just "math" that tells us it. Its a science experiment where you see if your data fits an a priori model better then you would expect just random uncorrelated data to do so. The only closed equation is the actual linear regression since that's just a minimization of sum of squares error for each data point.
Jump in the discussion.
No email address required.
That's a good way to put it
Jump in the discussion.
No email address required.
More options
Context
More options
Context
i think instead of turtles the world rests on the back of giant r-slurs
Jump in the discussion.
No email address required.
More options
Context
A midwit for sure, but in the land of the blind the one-eyed man is king.
Jump in the discussion.
No email address required.
More options
Context
More options
Context
Nobody knows what linear regression is unless it's somehow relevant to their field, or they're reddit dweebs who spend all day arguing on the internet with a wikipedia tab open pretending to know stuff. Ask any regular person and they don't know and don't care that they don't know.
Jump in the discussion.
No email address required.
More options
Context
More options
Context