Me reading the abstract: Well, 'coding error' is a bit strong, isn't it, they just disagreed on some variable coding.
Me getting to p. 3: Oh, OOOHHH, OH NO!!!!
Remember that recent paper claiming to show that "ideological bias" determined the results obtained by immigration policy research teams?
A careful, open-code reanalysis by @kauspurg.bsky.social & Josef Brüderl finds that the result arises from a coding error. —> doi.org/10.31222/osf...
Michael Clemens
Sebastian Karcher
🚨 A few months ago, a paper published in Science reported that “LLM adoption is associated with a large increase in researchers’ scientific output.”
In a comment released today, we show that the reported effects are driven by a methodological issue in the empirical design. arxiv.org/abs/2605.17979
Remember that recent paper claiming to show that "ideological bias" determined the results obtained by immigration policy research teams?
A careful, open-code reanalysis by @kauspurg.bsky.social & Josef Brüderl finds that the result arises from a coding error. —> doi.org/10.31222/osf...
Also surprised it has an H5N1 test.
At the back it says, if the result is positive for H5N1 but not flu A, it's a false positive.
But if it's positive for both, you must immediately go to hospital for confirmation.
Claude Fable's "biosecurity" flags are incredibly restrictive.
This model is going to be almost useless for me.
Every time I visit my parents, they're like "So when's the date for YOUR wedding?"
And I answer "Yesterday, you weren't invited."
Co-founder & editor, Works in Progress. Writer, Scientific Discovery. Podcaster, Hard Drugs. Advisor, Coefficient Giving. // Previously at Our World in Data.
Newsletter: https://scientificdiscovery.dev
Podcast: https://harddrugs.worksinprogress.co
🏳️🌈
