Sebastian Karcher
LightNews LightNews
banner
adam42smith.bsky.social
Sebastian Karcher
@adam42smith.bsky.social
1.3K followers 790 following 2K posts

Director, Qualitative Data Repository (personal account).
Data, Zotero, Social Science Methods
https://sebastiankarcher.com

Videos
Political science 25%
Computer science 23%
Posts Replies Media Videos

Reposted by Sebastian Karcher

eddieyang.bsky.social
Eddie Yang @eddieyang.bsky.social · 7h
We've updated the localLLM package on CRAN (cran.r-project.org/package=loca...).

It allows you to run LLMs locally and natively in R. Everything is reproducible. And it's free.

Some functionalities on reproducibility and validation🧵👇
localLLM: Running Local LLMs with 'llama.cpp' Backend
Provides R bindings to the 'llama.cpp' library for running large language models. The package uses a lightweight architecture where the C++ backend library is downloaded at runtime rather than bundled...
CRAN.R-project.org

Reposted by Sebastian Karcher

notalegalrecord.net
Tony Corsentino @notalegalrecord.net · 8h
📚 i’m about to head into Boston for the LIS class i teach. As ever, i am inspired and heartened by my students’ commitment to our profession, despite the grave uncertainties of its future (BIBFRAME)

adam42smith.bsky.social
Sebastian Karcher @adam42smith.bsky.social · 13h
Untested (and obviously use with great care given how things can go wrong) but I've long hoped for exactly this as a Zotero plugin
mmilkowski.bsky.social Marcin Miłkowski @mmilkowski.bsky.social · 15h
Here's another @zotero plugin for you! Search & Replace in Library metadata. I always missed a convenient way to remove Worldcat URLs from printed books. Ugly and unnecessary. Now you can. It's a pre-release, so back up your library because it may wreak havoc. github.com/milekpl/zote...
GitHub - milekpl/zotero-search-replace: Zotero plugin for advanced search & batch replace — regex, preview, and data‑quality patterns
Zotero plugin for advanced search & batch replace — regex, preview, and data‑quality patterns - milekpl/zotero-search-replace
github.com

adam42smith.bsky.social
Sebastian Karcher @adam42smith.bsky.social · 14h
Having Claude Code spin of three simultaneous agents is a bit pricey for mere existential dread, but seems perfectly doable.

adam42smith.bsky.social
Sebastian Karcher @adam42smith.bsky.social · 15h
We see them from all over, including the US, but we have seem some of the biggest spikes from China and Singapore, yes.
And it's every site of the open academic internet that's getting hammered. If you feel like you're seeing captcha's everywhere these days, that's why.

adam42smith.bsky.social
Sebastian Karcher @adam42smith.bsky.social · 15h
There's a lot of complaining about NYT headlines on here, but look at this masterpiece.
ditzkoff.bsky.social Dave Itzkoff @ditzkoff.bsky.social · 17h
Stephen Colbert said that the FCC and CBS - which is in no way a compromised or complicit organization - wouldn’t allow him to broadcast an interview with Democratic State Rep James Talarico. www.nytimes.com/2026/02/17/a...
Colbert Doesn’t Give an FCC About Calling Out CBS
www.nytimes.com

Reposted by Sebastian Karcher

mmilkowski.bsky.social
Marcin Miłkowski @mmilkowski.bsky.social · 15h
Here's another @zotero plugin for you! Search & Replace in Library metadata. I always missed a convenient way to remove Worldcat URLs from printed books. Ugly and unnecessary. Now you can. It's a pre-release, so back up your library because it may wreak havoc. github.com/milekpl/zote...
GitHub - milekpl/zotero-search-replace: Zotero plugin for advanced search & batch replace — regex, preview, and data‑quality patterns
Zotero plugin for advanced search & batch replace — regex, preview, and data‑quality patterns - milekpl/zotero-search-replace
github.com

Reposted by Sebastian Karcher

brendannyhan.bsky.social
Brendan Nyhan @brendannyhan.bsky.social · 16h
"Concomitant with the emergent partisan gap [in trust in science] is a massive perceptual gap among Democrats, who perceive a partisan divide more than double its actual size. Democrats vastly underestimate Republicans’ trust in scientists" academic.oup.com/poq/advance-...
Continuity and Change in Trust in Scientists in the United States: Demographic Stability and Partisan Polarization
Abstract. Americans’ trust in scientists has been stable and high, relative to other political and social institutions, for the last half century (Krause,
academic.oup.com

Reposted by Lisa W. Fazio, Ben H. Ansell, Jason Mittell , and 16 more

ditzkoff.bsky.social
Dave Itzkoff @ditzkoff.bsky.social · 17h
Stephen Colbert said that the FCC and CBS - which is in no way a compromised or complicit organization - wouldn’t allow him to broadcast an interview with Democratic State Rep James Talarico. www.nytimes.com/2026/02/17/a...
Colbert Doesn’t Give an FCC About Calling Out CBS
www.nytimes.com

Reposted by David Lazer, Sebastian Karcher, David Darmofal

cambup-polsci.cambridge.org
Cambridge University Press Political Science & IR @cambup-polsci.cambridge.org · 18h
#OpenAccess from @pspolisci.bsky.social -

Introduction: Teaching Qualitative Methods in Undergraduate Education - https://cup.org/46e2pz3

- @shamiragelbman.bsky.social & @adam42smith.bsky.social

#FirstView
Image featuring the letters 'PS' in bold white on a blue background next to a gauge graphic with a world map. The gauge displays multiple colors. The hashtag 'OpenAccess' is at the bottom.

adam42smith.bsky.social
Sebastian Karcher @adam42smith.bsky.social · 1d
I'm pretty sure skeleton racers would have your head for calling it luge :D

adam42smith.bsky.social
Sebastian Karcher @adam42smith.bsky.social · 1d
I have been thinking about Borges a lot lately. The connections to debates about LLMs are everywhere.

Reposted by Sebastian Karcher

realscientists.de
Licinia @ Real Scientists DE @realscientists.de · 1d
Hallo liebe RealScientists Community! Ich bin Licinia, und darf diese Woche Einblicke in meine Arbeit auf diesem Account teilen. Ich bin Politikwissenschaftlerin und promoviere zu Digitalisierung, Parteien, und öffentlicher Meinung an der Universität Oxford. Ich stelle mich hier vor:
Blauer, sonniger Himmel mit überschwemmter Wiese. Im Hintergrund Umrisse alter, prächtiger Gebäude Oxfords.

Reposted by Sebastian Karcher, Nancy M. Wingfield

ykomska.bsky.social
Yuliya Komska @ykomska.bsky.social · 2d
Chicago UP is running a HUGE sale, and you can get extremely well-researched hard-bound books about all sorts of topics and all sorts of places for a price of a cheap paperback press.uchicago.edu/resource/Sal...
press.uchicago.edu

Reposted by Shannon Vallor, Karen Benjamin Guzzo, Sebastian Karcher , and 2 more

joelhs.bsky.social
Joel S. @joelhs.bsky.social · 2d
This message from the US State Department about "not wanting allies to be shackled by guilt and shame" is a pretty direct call-out to the German far-right argument that Germany has spent too long feeling guilty about the Holocaust, and needs to move on and stop doing remembrance.
Post from the Department of State: "We do not want allies to be shackled by guilt and shame"

adam42smith.bsky.social
Sebastian Karcher @adam42smith.bsky.social · 2d
Not the point, but this list (which as I understand it is for law schools) is so weird: the top schools missing are U Chicago, Berkeley, Michigan and UVA.
Berkeley & Michigan are arguably the most progressive top schools.
UChicago has a brainy/academic rep.
What is going on???
chriso-wiki.bsky.social ChrisO_wiki @chriso-wiki.bsky.social · 3d
The US DOD has just blacklisted military personnel from taking courses at most of America's top universities and specialist tech and medicine schools, including Harvard, Yale, MIT and Johns Hopkins. Somehow I don't see the PLA making this kind of unforced error.

adam42smith.bsky.social
Sebastian Karcher @adam42smith.bsky.social · 3d
The first two are fair -- never really my cup of tea (& I don't think either really has taken off), but M definitely pioneered. But pushed reproducibility? How that?

(I agree also on the marketing, btw.; my Mendeley story would be very different)

Reposted by Iikka Korhonen, Sebastian Karcher, Jason Lyall , and 1 more

chriso-wiki.bsky.social
ChrisO_wiki @chriso-wiki.bsky.social · 3d
The US DOD has just blacklisted military personnel from taking courses at most of America's top universities and specialist tech and medicine schools, including Harvard, Yale, MIT and Johns Hopkins. Somehow I don't see the PLA making this kind of unforced error.

adam42smith.bsky.social
Sebastian Karcher @adam42smith.bsky.social · 4d
And to be clear, I'm not trying to argue here -- you're not the first person I've seen talking about this & if there's a way to make data wrangling less error-prone I want to understand that

adam42smith.bsky.social
Sebastian Karcher @adam42smith.bsky.social · 4d
right, but I, too, lack the imagination to do this (even though I obv. write unit tests in software). Do you (or does Claude?) use simulated data with given properties for unit tests?
With software unit tests, I have a clear sense if X assert Y. But where do I get that from in data wrangling?

Reposted by Dorothy Bishop, Brian A. Nosek, Kai Sassenberg , and 7 more

malte.the100.ci
Malte Elson @malte.the100.ci · 4d
New paper, on a worrying trend in meta-science: the practice of anonymising datasets on, e.g., published articles. We argue that this is at odds with norms established in research synthesis, explore arguments for anonymisation, provide counterpoints, and demonstrate implications and epistemic costs.
psyarxivbot.bsky.social PsyArXivBot @psyarxivbot.bsky.social · 4d
Against Anonymising Meta-Scientific Data: https://osf.io/6eyjf

Reposted by Dorothy Bishop, Sebastian Karcher

psyarxivbot.bsky.social
PsyArXivBot @psyarxivbot.bsky.social · 4d
Against Anonymising Meta-Scientific Data: https://osf.io/6eyjf

adam42smith.bsky.social
Sebastian Karcher @adam42smith.bsky.social · 5d
I'm confused here -- the paper doesn't make that assumption. Its main theoretical model uses 75% true hypotheses, and they show robustness with 90% true hypotheses given prevalent power estimates from the polisci lit.

adam42smith.bsky.social
Sebastian Karcher @adam42smith.bsky.social · 5d
Depends.
There's consensus that you need to report using it as an analysis tool, such as in the Briggs et al. paper on null findings; and imo that's where LLMs are game changers: no way that paper gets written wo LLMs.

No consensus on whether you need to declare using it to write R code

adam42smith.bsky.social
Sebastian Karcher @adam42smith.bsky.social · 5d
It's the Meat Loaf caucus

They would do anything for Trump,
But they won't do that

adam42smith.bsky.social
Sebastian Karcher @adam42smith.bsky.social · 5d
Curious how you are setting this up account wise? Do you just expect students to have a Claude Pro account? Sponsor API credits? Does university help?
Even for workshops this is a logistical headache here, in spite of Claude Enterprise (which covers neither API nor Code) & generally helpful IT

Reposted by Benjamin Braun, Sebastian Karcher, William Myers

adam42smith.bsky.social
Sebastian Karcher @adam42smith.bsky.social · 5d
Doch

adam42smith.bsky.social
Sebastian Karcher @adam42smith.bsky.social · 5d
If you're not hearing about DOJ career staffers quitting and blowing whistles left & right, you're not listening, from this week, e.g.: www.npr.org/2026/02/05/n...
(ICE is indeed rotten to the core; that's where 'abolish ice' -- hardly a new slogan -- comes from)
More frustrated prosecutors at the U.S Attorney's office in Minnesota call it quits
Turmoil continues at the U.S. Attorney's office in Minneapolis. This week another attorney was removed from a special assignment dealing with immigration cases after telling a judge she hates her job.
www.npr.org

adam42smith.bsky.social
Sebastian Karcher @adam42smith.bsky.social · 5d
A lovely post -- I use "socio-technical" in virtually every grant application and either my co-authors or (even more likely) my grant folks always complain that it's jargony. But it's soooo useful to describe what we're doing as repositories & data stewards