Simon Munzert
@simonsaysnothin.bsky.social
390 followers 420 following 22 posts
Professor of Data Science and Public Policy | Hertie School Data Science Lab | Elections, Public Opinion, Data
Posts Media Videos Starter Packs
simonsaysnothin.bsky.social
... in which I accidentally imitate Otto Wels, while saying things far less momentous than he did back then.
jbgruber.bsky.social
@simonsaysnothin.bsky.social at #MEDem Conf: we need to integrate our efforts instead of researchers all building their own datasets and infrastructure. Couldn't agree more!


Integrate

○ Researchers get lost in building fragmented data
infrastructures (case in point: me).
○ specialized collection is fine, but fragmented
dissemination hinders use.
○ Lack of integration blocks comparative research.

Whot I'd like MEDem to build towards

○ Incentivization of shared infrastructures
○ Toolkits to standardize data collection (measurement and interoperability)
simonsaysnothin.bsky.social
Day 1 at #MEDem Conference @gesis.org in Cologne, where I had the chance to share a personal perspective on data infrastructure projects for democracy research. Slides here: simonmunzert.com/medem25/
medem.bsky.social
🚀 Kicking off day 1 of the 3rd #MEDemConference at @gesis.org! From @sldelange.bsky.social discussing radical-right normalization 🗳️ to calls to consolidate democracy data by @simonsaysnothin.bsky.social 📊 — keynotes spotlight why #MEDem matters for democracy.
Sarah de Lange presents at the 3rd MEDem Conference 2025. Behind her, a slide shows the book “The Normalization of the Radical Right” and a quote on how radical-right behavior grows when previously hidden views become socially acceptable. Simon Munzert speaks at the 3rd MEDem Conference 2025 on the past, present, and future of democracy research.
simonsaysnothin.bsky.social
Congrats Andreu, this is so cool!! 🎉
Reposted by Simon Munzert
bschlipphak.bsky.social
Deadline for the position as Full Professor at our dept is coming closer: Great position, great colleagues & very positive working atmosphere. Please contact @danbischof.bsky.social if you have any questions.

And thanks for asking - yes, the pic shows today's view from the position's office.
Reposted by Simon Munzert
argohdes.bsky.social
I have studied the targeting of journalists for many years, but the scale and public display of these killings is truly shocking: “Israel admits deliberate attack on the journalist, known for frontline coverage, in a strike on a tent outside al-Shifa hospital“

www.theguardian.com/world/2025/a...
Anas al-Sharif, prominent Al Jazeera correspondent, among five journalists killed in Israeli airstrike on Gaza
Israel admits deliberate attack on the journalist, known for frontline coverage, in a strike on a tent outside al-Shifa hospital
www.theguardian.com
simonsaysnothin.bsky.social
Amazing to see what those students pulled together! And I'm delighted that our @hertiedatascience.bsky.social‬ Data4Good festival got the ball rolling... stay tuned for the next edition in 2026! www.hertie-school.org/en/data4good...
simonsaysnothin.bsky.social
If you're a researcher (or, less likely, a policymaker) reading this, you'll definitely want to read on. Featuring a dataset that may include you - and findings that make a solid case for hanging out in this space and sharing your work.
seramirezruiz.bsky.social
🤔 How much do politicians engage with academic researchers online?

In my latest paper, I find that politicians from 12 countries rarely engage with researchers on social media, but this can change when expertise gains salience

Preprint: osf.io/preprints/osf/wqbe4_v1

🧵👇
Title page of the preprint
Reposted by Simon Munzert
rebekahtromble.bsky.social
The worst happened. We were DOGE’d. Our NSF funding is gone.

So now there’s nothing stopping me from sharing Expert Voices Together, a crisis response system for US-based researchers and journalists facing harassment.

It's a true passion project. 🧵 1/

expertvoicestogether.org
simonsaysnothin.bsky.social
A shout-out to the people who did all the hard work for this study - in particular @dawiet.bsky.social and Amin Oueslati, who paved the way in his master's thesis he wrote @hertieschool.bsky.social - super proud of him!
simonsaysnothin.bsky.social
Finally, black-box audits are not ideal. Researchers need better access to the models behind those services. More implications, and more findings, in the paper!
simonsaysnothin.bsky.social
Multiply that by the number of calls made to those APIs every day, and the decisions they inform. (We don't know that number, but it's probably no less than ~500k calls over the minute you've engaged with this thread.)
simonsaysnothin.bsky.social
And what is more, those decisions probably affect those most who are not perpetrators but members of attacked groups.
simonsaysnothin.bsky.social
But our results indicate that if your moderation pipeline largely builds on automated decisions informed by those services, you're going to produce A LOT of questionable decisions.
simonsaysnothin.bsky.social
Hate speech moderation is hard, and we wouldn't expect any model to do a perfect job. Also, it's not straightforward to agree on what constitutes hate speech in the first place, which is why our benchmark datasets are not beyond any doubt.
simonsaysnothin.bsky.social
Main finding #2: Group markers drive over-moderation. Words like "muslim", "gay", or "jews" make mis-classifying non-hate speech as hate speech more likely.
Table providing results from a SHAP Values analysis, showing which speech tokens contribute to misclassification decisions.
simonsaysnothin.bsky.social
Some APIs seem overly sensitive (in particular Google's NL API) while others tend to under-moderate (Perspective and Microsoft).
simonsaysnothin.bsky.social
Main finding #1: Performance varies wildly across moderation services, datasets, and metrics, but some of the failure rates are astonishing (FPR & FNR > 75% on balanced samples for some implicit and explicit speech!).
Table reporting some key findings from the paper - hate speech classification performance statistics across moderation services and datasets.
simonsaysnothin.bsky.social
Our paper investigates how commercial content moderation APIs handle group-targeted hate speech. This is a black-box audit (we have no idea how the models look exactly) of 5 major APIs with five million queries based on four datasets.
Info chart describing the design of the study. Figure title: Our black-box audit framework to evaluate commercial content moderation APIs.
simonsaysnothin.bsky.social
Have you ever used a content moderation API, such as Perspective API or OpenAI's Moderation API, for your research or to inform moderation decisions? Well, they might not have given you what you think they would.
Bill Murray sitting on a bed, somehow lost. Scene from movie Lost in Translation. Title says "Lost in Moderation"
simonsaysnothin.bsky.social
Congrats, Lukas! Very nice paper and visuals!
simonsaysnothin.bsky.social
Just checking in on how the young fellas are doing and... great, apparently! Congrats Elias and Andreas on this wonderful publication! And very timely as we enter a new session of the Bundestag that promises lots of contentious debating.
eliaskoch.bsky.social
Really glad to see this published! It was a true pleasure to collaborate with the wonderful @ankuepfer.bsky.social throughout every stage of this project 🙏 🎉
ankuepfer.bsky.social
🚀 Our paper (with @eliaskoch.bsky.social) on ‘The Politics of Seeking and Avoiding Discourse in Parliament’ is finally out as EarlyView at the EJPR!
simonsaysnothin.bsky.social
Aus, aus, das Spiel ist aus!! Zumindest für unser Vorhersagemodell von @zweitstimme.bsky.social für die #btw25. Hier unsere vorläufig finalen Prognosen, vor ca. 70 Zeugen an der @hertieschool.bsky.social vorgestellt. zweitstimme.org/presentation
Redirecting...
zweitstimme.org