Author | Lightnews

Reposted by Daniel Gerdesmann

Claus Wilke @clauswilke.com · 17h

We need to have a conversation about random seeds. Don't use 42.
blog.genesmindsmachines.com/p/if-your-ra...

If your random seed is 42 I will come to your office and set your computer on fire🔥

Figuratively. More likely you'll get a stern talking to.

blog.genesmindsmachines.com

14 24 63

Reposted by Daniel Gerdesmann

Achim Zeileis @zeileis.org · 19h

🚨 New CRAN Task View: Anomaly Detection
#rstats #anomalydetection

By Priyanga Dilini Talagala, Rob J. Hyndman @robjhyndman.com, Gaetano Romano

URL: cran.r-project.org/view=Anomaly...

Screenshot of the CRAN Task View "Anomaly Detection" at https://CRAN.R-project.org/view=AnomalyDetection

1 8 15

Reposted by Daniel Gerdesmann

Niels G. Mede @nielsmede.bsky.social · 1d

Published today: One of the biggest #science #communication studies to date. We asked 71,922 people in 68 countries how they #engage with information about #science and combined the data with several country-level factors: journals.sagepub.com/doi/10.1177/... #OpenAccess

4 66 130

Reposted by Daniel Gerdesmann

Jay Van Bavel, PhD @jayvanbavel.bsky.social · 3d

Is science getting too competitive?

Grant applications are skyrocketing and funding rates are decreasing.

Yet the article in Nature fails to mention the potential role that AI/LLMs in driving up grant applications.

It's becoming a lottery.
www.nature.com/articles/d41...).

4 13 35

Reposted by Daniel Gerdesmann

Ellie Bastos @elliebastos.bsky.social · 4d

Migrated my book to another domain elliedegb.quarto.pub/psychometrics/ . It's a free resource on psychometric validation of instruments in R. Please share!

Advances in Psychometric Theory and Measurement for Psychological Sciences

This is an educational resource, with content at the cutting edge of scientific knowledge. It aims to improve future research, by improving researchers knowledge in the statistical inferences made, pl...

elliedegb.quarto.pub

2 6 13

Reposted by Daniel Gerdesmann

Melanie Richter-Montpetit @mellimo.bsky.social · 4d

German study shows how far right sets agenda thanks to mainstream parties reshaping their communications to respond to what are initially fringe issues, boosting the spread of these ideas and signalling to voters that these ideas and stances were legitimate

shorturl.at/OFfig

German far right setting agenda as opponents amplify its ideas, study finds

Normalisation of far-right stances likely to affect success of such parties at ballot boxes across Europe, say researchers

www.theguardian.com

9 220 490

Reposted by Daniel Gerdesmann

Tuta @tuta.com · 5d

🚨⏳ Act now! You have until 𝐍𝐨𝐯𝐞𝐦𝐛𝐞𝐫 𝟑, 𝟐𝟎𝟐𝟓 to prevent Microsoft from using your LinkedIn data to train AI.

You're opted in automatically, unless you take action and turn it off.

7 210 220

Reposted by Daniel Gerdesmann

Quantitude the Podcast @quantitude.bsky.social · 9d

Coming tomorrow. It’s almost that time of year…

3 6 14

Reposted by Daniel Gerdesmann

KOJAMF🤘🖤🤘 @kojamf.bsky.social · 17d

Dr. Jane Goodall filmed an interview with Netflix in March 2025 that she understood would only be released after her death.

810 17K 38K

Reposted by Daniel Gerdesmann

Teruni Ahamat @teruniahamat.bsky.social · 20d

Excited to share my first PhD article, published in Scientific Data 👉 rdcu.be/eIRqD

We present the Shared Book Reading Corpus: 44 caregiver-infant dyads, 3 camera views (headcams + overview), speech transcriptions & developmental measures.

Access on Databrary for discovery & reuse! ✨

#OpenScience

1 9 21

Reposted by Daniel Gerdesmann

Ludo Waltman @ludowaltman.bsky.social · 21d

Important observation by Aaron Tay: Articles published by Elsevier and Springer Nature may be increasingly difficult to find. For many of these articles abstracts are not made openly available, limiting article discoverability, in particular using AI tools.

@barcelonadori.bsky.social @oa2020.org

Aaron Tay @aarontay.bsky.social · 21d

[blogged] The Petrol Tank for AI Discovery Might be Running Dry as Publishers close access to scholarly content such as abstracts due to AI incentives aarontay.substack.com/p/the-petrol... - an analysis of OpenAlex abstract coverage changes for Elsevier & Springernature

The Petrol Tank for AI Discovery Might be Running Dry as Publishers close access to scholarly content such as abstracts due to AI incentives

Elicit.com, Consensus, and Undermind.ai are among the new leading comprehensive cross-disciplinary “AI-powered academic search engines” today.

aarontay.substack.com

1 4 6

Reposted by Daniel Gerdesmann

wholeheart @robrad.bsky.social · 25d

Making the european cardiology guideline content accessible by LLMs with full control of the knowledge base. Add or remove recommendations any time: Our cardiology RAG Paper is out!
#cardiology #echo #medsky #ai #llm

doi.org/10.1093/ehjd...

A Multi-Query, Multimodal, Receiver-Augmented Solution to Extract Contemporary Cardiology Guideline Information using Large Language Models

Robert M Radke, MD, Gerhard-Paul Diller, MD, PhD, MSc, FESC, Rohan G Reddy, BS, Pushpa Shivaram, MD, David A Danford, MD, MPH, Shelby Kutty, MD, PhD, MHCM,

doi.org

2 4

Reposted by Daniel Gerdesmann

Jeff Greene @jeffgreene.bsky.social · 26d

⁉️ What's that? You do research on artificial intelligence and how it can be used to promote positive K-12 outcomes?

🗣️Well, then you should apply to this tenure-track assistant/associate professor position and come work with me at @uncschoolofed.bsky.social!!! 🎓

#PsychSciSky #AcademicSky #EduSky

Faculty Position in Artificial Intelligence in K-12 Education Open Rank – Tenure Track/Tenure Eligible

The School of Education at the University of North Carolina at Chapel Hill invites applications for a position at the rank of Assistant or Associate Professor in artificial intelligence in K-12 educat...

unc.peopleadmin.com

6 4

Reposted by Daniel Gerdesmann

Candice Morey @candicemorey.bsky.social · 26d

Call for collaborators! 🧵

The TL;DR: we seek collaborators on a #ManyLabs #RegisteredReport about what causes rapid forgetting.

In-principle accepted Stage 1: osf.io/ahjn5

Expressions of interest: cardiffunipsych.eu.qualtrics.com/jfe/form/SV_...

Further details in the 🧵:

1 16 7

Reposted by Daniel Gerdesmann

🎃 public HELLth guy 🎃 @publichealthguy1.bsky.social · 27d

from worst to best versions (off the top of my head)

Hierarchy of Scientific Evidence
Strongest to weakest top to bottom
Meta-analyses & systematic reviews
Randomized controlled trials
Cohort studies
Case-control studies
Cross sectional studies
Animal trials & in vitro studies
Case reports, opinion papers, and letters

same pyramid but at the top is another pyramid that says 25+ scientists signed an opinion paper

same pyramid but at the top is mom’s advice

a shitpost pyramid with “the other shit” on the bottom and “Thoughtful, well-conducted studies of any design” at the top. probably the best version i’ve seen

5 16 67

Reposted by Daniel Gerdesmann

Richard D. Morey @richarddmorey.bsky.social · 27d

Simonsohn has now posted a blog response to our recent paper about the poor statistical properties of the P curve. @clintin.bsky.social and I are finishing up a less-technical paper that will serve as a response. But I wanted to address a meta-issue *around* this that may clarify some things. 1/x

Data Colada @datacolada.bsky.social · 28d

Would p-curve work if you dropped a piano on it?
datacolada.org/129

PIano being dropped on car in car testing facility

2 29 72

Reposted by Daniel Gerdesmann

Daniel Lakens @lakens.bsky.social · 29d

New blog post from Data Colada, responding to the recent criticisms on p-curve analysis. It is a *very* good response. As in, it addresses exactly the points I would have expected in a reply, and it explains why I will still teach p-curve analysis. datacolada.org/129

8 22

Reposted by Daniel Gerdesmann

David Dunning @daviddunning6.bsky.social · Sep 21

I'm not making this up.

LLMs always will. Says the math.

Prof Mike Yearworth @mikeyearworth.bsky.social · Sep 21

"In a landmark study, OpenAI researchers reveal that large language models will always produce plausible but false outputs, even with perfect data, due to fundamental statistical and computational limits."

www.computerworld.com/article/4059...

OpenAI admits AI hallucinations are mathematically inevitable, not just engineering flaws

In a landmark study, OpenAI researchers reveal that large language models will always produce plausible but false outputs, even with perfect data, due to fundamental statistical and computational limi...

www.computerworld.com

2 7

Reposted by Daniel Gerdesmann

Andrei Cimpian @andreicimpian.bsky.social · Sep 22

💖This paper has been ~11 years in the making - and probably my favorite project of all time. Thrilled to see it in @pnas.org! I'm so lucky that Zach decided to do a second PhD and join my lab @psychillinois.bsky.social back in 2014 - a fabulous scientist & human being! www.pnas.org/doi/10.1073/...

Historical and experimental evidence that inherent properties are overweighted in early scientific explanation

4 9 40

Reposted by Daniel Gerdesmann

Jonas Haslbeck @jmbh.bsky.social · Sep 22

Two new preprints on multilevel HMMs! Time series data is now pervasive in psychology and new methods are needed to model the dynamics in such data. Hidden Markov Models (HHMs) are powerful models for dynamics in which a system is switching between a number of discrete states.

2 21 66

Reposted by Daniel Gerdesmann

Josef Fruehwald @jofrhwld.bsky.social · Sep 22

I wrote up a blog post about how I made those animated sound waves

Animating Sound Wave – Væl Space

jofrhwld.github.io

2 4 16

Reposted by Daniel Gerdesmann

Andrew J. Steinmetz @ajsteinmetz.com · Sep 20

Hmm. 🤔 Enrollment in bachelor's degrees in #Physics seems to have fallen off a cliff in German universities since the pandemic. Anybody know if this trend has reproduced itself in your country or in another STEM field? #EduSky #AcademicSky #Science 🧪⚛️

Source: www.dpg-physik.de/veroeffentli...

5 19 40

Reposted by Daniel Gerdesmann

Mark Rieke @markjrieke.bsky.social · Sep 17

It's very gratifying to write with a focus on statistical rigor and earnestly attempting to get the methodology right while being able to open-source the codebase. If you dig this sort of thing, you should subscribe to Elliott's substack --- multiple publications per week!

www.gelliottmorris.com

2 5

Reposted by Daniel Gerdesmann

Yuriy Gorodnichenko @ygorodnichenko.bsky.social · Sep 16

A Russian drone hits a #university building in #Kharkiv

Senseless and barbaric daily terror

Imagine your university gets a hit like that

#Ukraine

Момент удару по освітньому закладу в Слобідському районі

YouTube video by Харківська обласна прокуратура

www.youtube.com

5 12

Reposted by Daniel Gerdesmann

Jeff Greene @jeffgreene.bsky.social · Sep 16

🗣️ When is using #GenAI a helpful shortcut, and when is using it really just cutting yourself short?

Find out in this (short!) article in Psychology Today by yours truly.

#PsychSciSky #AcademicSky #EduSky #Expertise

Learning With GenAI: Shortcuts Versus Cutting Yourself Short

When should you use generative AI to learn? Some say you should use generative AI all the time; others say never. Your answer should depend on your goals.

www.psychologytoday.com

7 7