Sign in

Jan Lorenz

@janlo.de
886 followers 910 following 82 posts

Computational social scientist, agent-based models, opinion dynamics and collective decision. Sometimes english, manchmal deutsch datasci.social/@janlo

PostsRepliesMedia
Jan Lorenz @janlo.de · 26/09/2026
Es knallte laut in London, wenn Überschalljets landen.
020
Reposted by Jan Lorenz
Pedro Magalhães @pcmagalhaes.bsky.social · 24/09/2026
The Atlas of Social Attitudes is now live: www.atlasofsocialattitudes.eu What is it, what can you do with it, and why did we build it? A short🧵. (1)
atlasofsocialattitudes.eu
Atlas of Social Attitudes
Explore comprehensive public opinion data from the European Social Survey and the European Values Survey. Discover insights across topics, countries, an...
33325
Reposted by Jan Lorenz
Artjoms Šeļa @artjomshl.bsky.social · 23/09/2026
a PhD position in LLMs, annotations, language and benchmarking @ University of Tartu. please share! digits.ut.ee/join-the-team/
01911
Reposted by Jan Lorenz
Alexander Wuttke @kunkakom.bsky.social · 23/09/2026
We are hiring at LMU Munich: two PhD positions and one postdoc on democracy, political psychology and computational social science. Start: January 2027 or as soon as possible. We review applications on a rolling basis until October 20. Details below. Reposts very welcome! 1/7
25879
Jan Lorenz @janlo.de · 23/09/2026
Was ist denn hier los? Das stimmt ja doppelt nicht. Bei RRG in Berlin stünden RRG Regierungen im Bundesrat bei 22 Stimmen weit von 35 (Mehrheit) entfernt. Außerdem ist Berlin (wie SH) schon jetzt ohne RRG für den Antrag auf AfD Verbot (laut PRÜF).
020
Reposted by Jan Lorenz
Dino Carpentras @justanormaldino.bsky.social · 22/09/2026
1/ People who feel optimistic about AI tend to agree with one another. But people who worry about AI? Not so much. A belief-network analysis of six questions about AI and technology reveals a striking asymmetry. 🧵
162
Jan Lorenz @janlo.de · 20/09/2026
Die Diskussion unter diesem Post in Richtung Wahlbetrug von AfD Leuten in Buchholz befindet sich auf dem ähnlich niedrigen Niveau wie man es Wahlbetrugsvorwürfen von AfD Supporten zu recht vorwirft.
000
Reposted by Jan Lorenz
Fariba Karimi @frbkrm.bsky.social · 20/09/2026
I’m excited to share our new preprint about network fairness and its ethical challenges. We spent many years thinking about this topic and shaping our arguments. /w @lespin.bsky.social @eliassi.bsky.social Ricardo Baeza-Yates , Pak Wong , Erich Prem , Meike Zehlike ,Suresh Venkatasubramanian
0116
Jan Lorenz @janlo.de · 20/09/2026
010
Reposted by Jan Lorenz
Sebastian Ziaja @sziaja.bsky.social · 17/09/2026
Geo-coded electoral data offers immense potential for understanding democratic processes. @dielea.bsky.social, Stefan Jünger, and I will build a research data infrastructure to support the global research community using such data with our DFG-funded project "Electoral Spaces". We are hiring!
31813
Reposted by Jan Lorenz
UVA School of Data Science @uvadatascience.bsky.social · 17/09/2026
The School of Data Science at UVA is committed to educating the next generation of data science leaders. The Ph.D. in Data Science is designed to impart the skills and knowledge necessary to enable research and discovery in data science methods. Apply today: bit.ly/3q9IEqs
Four people are standing together in a modern building with stairs in the background. They appear to be part of a promotional image for a Ph.D. in Data Science Program. A bright "APPLY NOW" button is at the bottom.
012
Jan Lorenz @janlo.de · 13/09/2026
Als Follow-up zum selber basteln: www.secrethitler.com (und um von Werwolf weg zu kommen).
secrethitler.com
Secret Hitler
A social deduction game for 5-10 people about finding and stopping the Secret Hitler.
002
Jan Lorenz @janlo.de · 04/09/2026
I was referring to the original post's idea of less implications through AI GDP growth + share of labor shrinking. Sure, climate cost is relevant and likely supersedes this question soon, but that's a different story, and the post was not claiming that this GDP growth will really happen.
100
Jan Lorenz @janlo.de · 04/09/2026
I have no clear assessment of whether the claim as such is realistic. For me, the interesting point was that, even if we believe in the growth potential, the effect need not be that drastic.
100
Reposted by Jan Lorenz
Jacob Edenhofer @jacobedenhofer.bsky.social · 04/09/2026
marginalrevolution.com/marginalrevo...
marginalrevolution.com
How Much Redistribution Will AI Require? - Marginal REVOLUTION
How much redistribution will AI require? A common scenario is that AI raises output enormously, but labor’s share of income collapses. GDP per capita goes up but workers get poorer, and making workers...
131
Jan Lorenz @janlo.de · 26/08/2026
Thanks for this! Also "All models are wrong, but some are useful." has value. It doesn't give the reason explicitly (truth is bad category for models). However, it provides an answer to the (irrelevant) truth/correctness question. You can agree on "wrong" and move to the purpose discussion.
110
Reposted by Jan Lorenz
Julia Haaf @juliaha.bsky.social · 26/08/2026
I still feel like people should quote Box (1978) instead of the other one. As I understand it, he is basically saying that “Is the model true?” is the wrong question. Therefore, the statement "all models are wrong" is the answer to a *question* that isn't useful.
2318
Reposted by Jan Lorenz
Yihui Xie @yihui.org · 21/08/2026
I just came back from summer vacation, and would like to announce a tiny new R package I developed recently: {lt} yihui.org/en/2026/07/l...
yihui.org
The lt Package: Lightweight HTML Tables for R (and Beyond) - Yihui Xie | 谢益辉
Recently I have released a new package on CRAN: lt, short for lightweight tables (or less than). Its origin story is a small moment of shock. About two years ago a project required gt, so I ran &helli...
48023
Reposted by Jan Lorenz
Wilko Zicht @zicht.de · 13/08/2026
Grüne und AfD sind halt die Parteien, deren Umfragewerte seriöse Institute am stärksten nach oben bzw. unten manipulieren MÜSSEN, um realistische Projektionen zu erhalten. Siehe Forschungsgruppe Wahlen vom 17. Juli: Nach Gewichtung AfD und Grüne gleichauf bei 20%, in Projektion AfD 27%, Grüne 14%.
211
Reposted by Jan Lorenz
Miranda Lubbers @mirandalubbers.bsky.social · 13/08/2026
Again: Science policy worldwide/regionally should support collective investment in university-led infrastructure for journals,surveys,open science platforms,knowledge transfer, indexing services, so that academia no longer depends on highly profitable private corporations & controls quality & access
111
Reposted by Jan Lorenz
Julio Iturra-Sanhueza @jciturras.bsky.social · 06/08/2026
📢 Call for Abstracts: "Comparative Study of Social Cohesion and Economic Inequality" XXI ISA World Congress of Sociology 📍 Gwangju, Korea | July 2027 🗓️ Deadline: Oct 14 🔗 l1nq.com/bkx6xb7 #Sociology #ISA2027
023
Reposted by Jan Lorenz
Nils Steiner @nilssteiner.bsky.social · 06/08/2026
We are hiring a postdoc in Comparative Politics at JGU Mainz! Join our friendly and productive team, led by Armin Schäfer (@arminschaefer.bsky.social). • Full-time position • Five-year contract (!) • Start: from 1 October Apply by 22 August: stellenboerse.uni-mainz.de/jgu/job/56328 Please share!
05362
Reposted by Jan Lorenz
Daniel Mullis @daenumullis.bsky.social · 05/08/2026
Zur gegenwärtigen Relevanz von Linksextremismus ist der Sachsen-Anhalt-Monitor 25 aufschlussreich: 33 % der Menschen wünschen sich „Revolution“ gegen den Staat, der abgeschafft gehöre. Geteilt werden die Einstellungen vor allem Rechtsaußen, während sie links weniger ausgeprägt sind als in der Mitte.
2025
Die Lebensbedingungen werden durch Reformen nicht besser - wir brauchen eine Revolution. 33 Prozent Zustimmung

Der Staat ist ein Instrument der Unterdrückung, das abgeschafft werden muss. 22 Prozent Zustimmung
Zustimmung zur Aussagen ist vor allem Rechtsaußen sehr hoch.
3328
Reposted by Jan Lorenz
Darren Dahly @statsepi.bsky.social · 04/08/2026
Stop trying to explain what a p-value is in a single sentence (ICYMI) statsepi.substack.com/p/no-you-can...
statsepi.substack.com
No, you can't explain what a p-value is with one sentence (Parts I, II)
But I'm going to try and do it in as few as possible. #fml
106822
Reposted by Jan Lorenz
Pat Savage @patrickesavage.bsky.social · 26/07/2026
I completely re-wrote the AI section of my guide to academic reading/writing to be much stricter and more negative (now titled "AI-generated fabrications as academic misconduct"). Please use/share with your students, and let me know if you have any more changes to suggest! doi.org/10.31234/osf...
32317
Reposted by Jan Lorenz
Taha Yasseri @tahayasseri.bsky.social · 23/07/2026
Chapter 18: @janlo.de on Exploring Theory with Agent-Based Modeling and Simulation. Showing how agent-based models can move CSS beyond prediction toward explanation, revealing how simple behavioural rules can generate complex social phenomena such as segregation, polarization, and filter bubbles.
261
Reposted by Jan Lorenz
Philipp Lorenz-Spreen @lorenzspreen.eurosky.social · 16/07/2026
Ich freue mich schon sehr auf diese besondere Zusammenarbeit und auf die Möglichkeit, alternative News Feed Algorithmen außerhalb des Labors zu testen – in der Praxis und unter realistischen Bedingungen!
041
Reposted by Jan Lorenz
Simon Munzert @simonsaysnothin.bsky.social · 01/07/2026
Stop! 🛑 Scroll no further - this is the coolest thing you'll see today. Presenting The Data Dispatch: data journalism made by my students at the Hertie School 👇 data-journalism-26.github.io
Screenshot of The Data Dispatch website
1256
Reposted by Jan Lorenz
Michael Friendly @datavisfriendly.bsky.social · 27/06/2026
Who wears lipstick? An April 1944 ISOTYPE-like graph by Jean-Andre Sainte-Laguë(1882-1950) in Science et Vie. See: en.wikipedia.org/wiki/Andr%C3...
083
Jan Lorenz @janlo.de · 27/06/2026
Here a fitting solution!
Displays the creative process from "Work begins" to "Deadline" with phases "Fuck off" (long), followed by "Panic" (short), and "All the work while crying" (super short).
000
Jan Lorenz @janlo.de · 26/06/2026
How can we see the fraction of blue-haired students in the graph???!???!!
080
Reposted by Jan Lorenz
Nicolas Lambert @neocarto.bsky.social · 25/06/2026
The R package 'geovizr' is now available on CRAN 🥳 It allows users to create vector-based, interactive, and zoomable thematic maps directly from R. #rstats #rspatial #maps #cartography #dataviz #webmapping Explanations, examples, and documentation are available here: riatelab.github.io/geovizr
1298
Reposted by Jan Lorenz
Daniel Lakens @lakens.bsky.social · 21/06/2026
New preprint: There is only one correct analysis. osf.io/preprints/ps... with @sajedehra.bsky.social and @mntunc.bsky.social We take a critical look at recent proposals for multiverse or many analyst procedures, and strongly argue that epistemic uncertainty should be reduced, not “embraced”. >
osf.io
OSF
55823
Reposted by Jan Lorenz
Per Engzell @pengzell.bsky.social · 19/06/2026
What people keep forgetting is that even when LLMs can do 80% of the work, the remaining 20% still takes >80% of the time
1324
Reposted by Jan Lorenz
Jeffw @jeffw.bsky.social · 25/04/2026
If this time travel experiment works, I should be able to go from here and now and reappea—
11692235
Reposted by Jan Lorenz
Per Engzell @pengzell.bsky.social · 12/06/2026
What doesn’t kill you becomes a limitation section
212017
Reposted by Jan Lorenz
Andreas Jungherr @ajungherr.bsky.social · 10/06/2026
Yesterday Anthropic released its new Claude model Fable. As widely and controversially discussed, Fable comes with strong guardrails, some visible to users, others less visible in their operation. This gives us a powerful preview of where the alignment of public AI models is headed.
242
Reposted by Jan Lorenz
Katrin Weller @kwelle.bsky.social · 10/06/2026
several current open positions with my colleagues at @gesis.org
003
Reposted by Jan Lorenz
Martin Gross @martingross.bsky.social · 02/06/2026
Bewerbungsfrist endet in einer Woche! Apply, apply, apply! 😊
0109
Jan Lorenz @janlo.de · 28/05/2026
Aperiodic monotile copied from https://kottke.org/23/03/a-potential-major-discovery-an-aperiodic-monotile
020
Reposted by Jan Lorenz
Zach Weinersmith @zachweinersmith.bsky.social · 27/05/2026
Bonus panel here: www.smbc-comics.com/comic/ai-17 Also last day to get my new book(s) on kickstarter! www.kickstarter.com/projects/wei... #smbc #comics #ai
327839
Reposted by Jan Lorenz
Julia Partheymüller @schnizzl.bsky.social · 24/05/2026
Can LLMs Emulate Human Belief Dynamics? 🧪 The answer is a clear no. [The findings] raise a sharp warning against deploying LLMs as human proxies in social simulations. arxiv.org/pdf/2605.18781 #SyntheticData #EsWirdImmerAbsurder
Screenshot of the abstract of the preprint.

Can LLMs simulate how humans form and change beliefs in social net-
works? We put this to the test by replicating an established study on belief
dynamics, evaluating 12 LLMs across multiple model families and param-
eter sizes. The answer is a clear no, and in systematic ways. LLMs fail
to capture initial human belief distributions and tend to be overall more
conformist than humans, shifting their responses to align with those around
them. They also take a nuanced approach to emulating human homophilic
tendencies within networks. Our findings carry a double payoff: they
highlight fundamental properties of LLM behavior, and they raise a sharp
warning against deploying LLMs as human proxies in social simulations.
03915
Reposted by Jan Lorenz
Hadley Wickham @hadley.nz · 18/05/2026
I'm bringing my substack back to life to talk about AI and data science: open.substack.com/pub/tidydesi.... I have conflicted feelings about both AI and writing about AI but I want to try and work through them in the open, and I hope you'll join me on this journey.
open.substack.com
Returning to life!
Welcome back to the tidy design principles substack.
27416
Jan Lorenz @janlo.de · 15/05/2026
I guess it was always somehow like this, but on changing metrics. I think this perception is built by seniors emphasizing with their brilliant young colleagues and those brilliant juniors envying and admiring tenure. Both ignore survivorship bias.
010
Jan Lorenz @janlo.de · 07/05/2026
The circular order of dimension must have meaning is what I meant.
010
Jan Lorenz @janlo.de · 07/05/2026
Sure, but the critique does not apply so well to my case. Radar charts only make sense when there is a clear theoretical (and at best empirically plausible) meaning. Also, one must not search meaning in the area (when there is not fill color). This is complicated to interpret and rarely useful.
110
Jan Lorenz @janlo.de · 07/05/2026
Yes that is how it is used. I don't get the small multiples idea.
010
Jan Lorenz @janlo.de · 07/05/2026
Opposing dimensions rarely go together, although it is possible. The closer dimensions are the more likely is values are similar. So typically social Democrats have a bubble left. Capitalist liberals top right and anti migrant conservatives bottom right. Without radar structure this value vanishes.
000
Jan Lorenz @janlo.de · 07/05/2026
I see no chart which is substantially better here. Convince me on that :-) The value of the radar chart is something else. 2/3
convince.me
110
Jan Lorenz @janlo.de · 07/05/2026
I am not in Switzerland anymore, but I remember these were used for party, candidate, and voter profiles (summarizing input into voting advice application smartvote.ch). A typical use case is to put 2 or 3 profiles in one chart, e.g. me and two candidates I consider to elect. 1/3
smartvote.ch
Willkommen bei smartvote!
110