Restored from Instagram stories ·
is bayesian failure to choose a proper prior in the room with us?
it can also mean that physically attractive people don’t game much. it’s not that those who game aren’t physically attractive.
so what’s the prior?
possibly people who are physically attractive spend a lot of time in outdoor games that they don’t sit inside their home.
also possible that i can pick a prior and form a bayesian analysis that gamers have high iq
the thing with bayesian analysis is that, there’s no correct way to choose a prior.
it’s mostly common sense/decent average iq over 100 to question things.
all prior will have a bias. each subjective to an individual’s eye.
inferring here is a skill which requires choosing a subjective prior beliefs into a mathematically formulated prior to write your paper.
ta-da! you have a paper.
when i find incels here propogating scientific misinformation to sell their own stuff or to spread misinformation in social media and telling us there’s a research paper,
how will i explain this to a largely 77-80 iq’ish average population in india on how most of these research papers are just bullshit.
just poking fun into one. i can dissect any research paper the same way - starting with their sample size.
so whenever i read a paper, after i check sample size, i immediately seek where the author has written their justification on choosing their prior.
by intentionally choosing a wrong or biased prior, you can manipulate the outcome of the analysis to support almost any desired conclusion.
lemme dumb it down a bit. for example, let’s say i fund a pharma startup. i want them to write a paper to sell the drug.
if writing a paper on the effectiveness of a new drug, selecting a prior that strongly favors a positive outcome (e.g., assuming a high probability of success) can skew the results in favor of that outcome, even if the actual data is inconclusive or suggests otherwise.
can we hide that? absolutely if you play well with how you justify choosing that prior.
the results will heavily depend on the incorrect prior rather than the true data.
all i need to complete this manipulation is send this to few news channels.
new research is born.
All rights reserved. Copyright © 2026 Jason Samuel