iiinova — where your digital self keeps evolving

Johannes B. Gruber @jbgruber.bsky.social · 09/10/2026

If `coffee = productivity` I was extremly productive during my time at QUT in Brisbane.

040

Johannes B. Gruber @jbgruber.bsky.social · 08/10/2026

Dear publishers: just give me the damn pdf, I don't care about your stupid reader

"Reader Environment Loading..." message in a browser window

3200

Johannes B. Gruber @jbgruber.bsky.social · 23/09/2026

Alright, let's give this paper a final read before resubmitting it 🤗

View of a laptop with text with the Brisbane City Botanic Gardens in the background.

040

Johannes B. Gruber @jbgruber.bsky.social · 23/09/2026

FYI, here is the course material, Session 6 is now 25% just summarizing the article 😅 jbgruber.github.io/ess-web-scra...

jbgruber.github.io

Introduction to Web Scraping and Data Management for Social Scientists

010

Johannes B. Gruber @jbgruber.bsky.social · 21/09/2026

I needed to come to Brisbane to see people being good at riding e-scooters. I wasn't even aware such a thing existed. But quite often I see people in groups casually doing a wheelie here. And just now, someone drifted around a corner at full speed on a lime. Pretty wild.

010

Reposted by Johannes B. Gruber

GESIS - Leibniz-Institut für Sozialwissenschaften @gesis.org · 14/09/2026

Many people misjudge how many migrants actually live in their community and how well they are integrated. Berbée, Brinkmann, Gallegos Torres, Lange, Naumann, Sommerfeld, Perceptions of migration and integration in Germany: The role of regional context and migrants’ origin, doi.org/10.1093/pnas...

045

Johannes B. Gruber @jbgruber.bsky.social · 09/09/2026

New #rstats 📦: mediacloud2 (happy about name recommendations). Since the exitsting R packages were stale and I needed data, I wrote a new package. MediaCloud is an amazing resource for getting news media data, especially if you rehydrate the articles (e.g., through paperboy).

jbgruber.github.io

MediaCloud API (v4) Wrapper

A simple wrapper for version 4 of the MediaCloud API, which provides programmatic access to the MediaCloud archive of online news content (<https://www.mediacloud.org>).

0218

Johannes B. Gruber @jbgruber.bsky.social · 09/09/2026

To prepare for this talk, I created an entire R package (with an actual purpose) from scratch in less than 8 hours (would have been less without the meetings in between). Let's see how it does while I'm using it for work today.

jbgruber.github.io

MediaCloud API (v4) Wrapper

A simple wrapper for version 4 of the MediaCloud API, which provides programmatic access to the MediaCloud archive of online news content (<https://www.mediacloud.org>).

011

Reposted by Johannes B. Gruber

MZES Social Science Data Lab @mzes-ssdl.bsky.social · 08/09/2026

▶️ Social Science Data Lab: Fall 2026 Events Four input talks by great researchers (see below ⤵️)! 🗓️ Details & Zoom: socialsciencedatalab.mzes.uni-mannheim.de/page/events/ 👥 Organizers: @rubac.bsky.social, @denis-cohen.bsky.social, Nadia Granato and Alexander Wenz

Upcoming
2026-12-02 | Workshop | Johannes B. Gruber (GESIS – Leibniz-Institut für Sozialwissenschaften)
Vibe, Check: Authoring R Packages for Open Science in 2026 with usethis, devtools, and Coding Agents
more

2026-11-04 | Input talk | Anne-Kahtrin Stroppe (University of Greifswald)
Measuring Infrastructural Deprivation from Big Geodata: On the Inferential Validity of OpenStreetMap Proxies
more

2026-10-14 | Input talk | Valentyn Melnychuk (LMU Munich)
Causal ML for treatment effect estimation
more

2026-09-16 | Input talk | Marc Ratkovic (University of Mannheim)
Harnessing GPT for Enhanced Academic Writing: Evidence from a Field Experiment with Early-Career Researchers in the Social Sciences
more

264

Johannes B. Gruber @jbgruber.bsky.social · 02/09/2026

This with Max+ 395 - 128GB shared memory. It's unfortunately a lot more expensive now. And it is also slower than an NVIDIA GPU, but I can use 105GB of the RAM with the GPU, which let's me play with bigger models 😀

frame.work

Order a Framework Desktop with AMD Ryzen™ AI Max

Framework Desktop: A simple, quiet, mini PC with performance far beyond its size, powered by a highly-integrated processor from AMD.

110

Johannes B. Gruber @jbgruber.bsky.social · 01/09/2026

Right after my last post I see this. Absolutely. I also hate how AI is shoved in my face constantly. There are a few tasks at which it is great, and a million where I don't need or want AI and I especially hate that everything is called Gemini or Copilot now.

061

Johannes B. Gruber @jbgruber.bsky.social · 01/09/2026

What changed recently is that small models are no longer meant to 'know' things, instead became good at *doing* things. For example, the model I used probably wouldn't get it right when I asked it about a niche topic. But it can search the internet and find out what it doesn't know now..

010

Johannes B. Gruber @jbgruber.bsky.social · 01/09/2026

The hardware I used cost me about 2000€. If you spend more, it will also generate the code faster.

210

Johannes B. Gruber @jbgruber.bsky.social · 01/09/2026

Qwen wrote several wrong solutions before it ended up getting it right. In the process, it discovered a bug in the package. Overall, I think if you need speed, Claude is still king. But local AI is winning in a lot of other aspects: no subscription costs, privacy, nobody can take it away.

110

Johannes B. Gruber @jbgruber.bsky.social · 01/09/2026

What this ran is a task defined in a lot of detail and with tests built in. This is what I consider the sweet spot for AI coding because I can verify if things actually work and I have a very clear idea how the code is supposed to look.

github.com

110

Johannes B. Gruber @jbgruber.bsky.social · 01/09/2026

Qwen3.8 27B comes extremely close to Claude in programming tasks. It's been said, but I wanted to share a test I ran in #rstats for {paperboy} 📦. TLDR: Qwen in opencode got it done in 28 mins, Claude Code ( Opus 5) got it done in 3.5. Results differ a little, but both are perfect solutions.

1165

Johannes B. Gruber @jbgruber.bsky.social · 26/08/2026

Absolutely. This makes it easier to parse and validate the answers. It doesn't improve accuracy and does not mean the model 'understands' your categories.

000

Johannes B. Gruber @jbgruber.bsky.social · 24/08/2026

rollama 0.3.1 just hit CRAN! 🥳 Highlight of the new version is that it now supports batch annotation of images with structured outputs. Screenshot from the updated vignette below:

Annotating Several Images with Structured Output

You can combine this setup with the structured output workflow to get a systematic classification of images. First, define the schema for the fields you want back for each image:

```r
animal_schema <- create_schema(
dog = type_boolean(),
cat = type_boolean(),
human = type_boolean(),
bird = type_boolean()
)
animal_schema
#> <rollama structured output schema>
#> ├─object: <NULL> (required)
#> └─properties
#> ├─boolean: <dog> (required)
#> ├─boolean: <cat> (required)
#> ├─boolean: <human> (required)
#> └─boolean: <bird> (required)
``

184

Johannes B. Gruber @jbgruber.bsky.social · 19/08/2026

Had a fantastic warm welcome at @qutdmrc.bsky.social where I'll be spending the next 8 weeks working on CSS and data donation topics with @antmandan.bsky.social and meeting fantastic new people every day so far 🤗

2131

Reposted by Johannes B. Gruber

Aleksandra Urman @aurman21.bsky.social · 12/08/2026

Thanks @adfichter.eurosky.social :) And it was exciting and refreshing to write in a non-academic genre for once! Glad I had the opportunity!

0103

Reposted by Johannes B. Gruber

Katrin Weller @kwelle.bsky.social · 28/07/2026

While we are still waiting for the final funding decision (which will be made by the BMFTR) for RIDLOP, we are delighted to see the Wissenschaftsrat's report that was published yesterday. So keeping our fingers crossed for the last steps in the evaluation process!

2163

Johannes B. Gruber @jbgruber.bsky.social · 27/07/2026

There is some hype in the world of AI this month and it's about open weight models 😮 Today, Kimi-K3 will be published and on paper, it's head to head with the biggest commercial models. You can't run it (unless you happen to have a supercomputer), but still exciting that you can download it!

huggingface.co

moonshotai/Kimi-K3 · Upcoming release · Hugging Face

The latest model from Moonshot AI.

040

Reposted by Johannes B. Gruber

GESIS - Leibniz-Institut für Sozialwissenschaften @gesis.org · 27/07/2026

#RIDLOP #Wissenschaftsrat #ResearchInfrastructure The German Science and Humanities Council (Wissenschaftsrat) supports the implementation of RIDLOP, the Research Infrastructure for Data from Large Online Platforms, in its statement published today.

1187

Johannes B. Gruber @jbgruber.bsky.social · 15/07/2026

I guess I should have read the note before posting 🫠

https://doi.org/10.31235/osf.io/86zqa_v1

DOIs are minted by a third party, and may take up to 24 hours to be registered.

040

Johannes B. Gruber @jbgruber.bsky.social · 15/07/2026

Hmm, I double-checked, that's the one OSF shows me. Maybe they didn't register the DOI properly yet? Sorry, here is the OSF link osf.io/preprints/so...

osf.io

OSF

140

Johannes B. Gruber @jbgruber.bsky.social · 15/07/2026

New preprint: my contribution for @marhai.bsky.social and @emesedomahidi.bsky.social 's The Handbook of Computational Communication Research discusses the transition of the field from bag-of-words to word embeddings to transformer based contextualized embeddings. doi.org/10.31235/osf...

2122

Johannes B. Gruber @jbgruber.bsky.social · 06/07/2026

If you have any questions about this job, let me know! It is a great opportunity and very important work!

043

Reposted by Johannes B. Gruber

Daniel Thiele @thieled.bsky.social · 03/07/2026

Thanks @christinaviehmann.bsky.social, @jbgruber.bsky.social and @gesis.org for the invitation! I had an absolutely fantastic time with many inspiring conversations and new connections with excellent researchers. Thanks for treating me so well, Cologne! #CSS #DSS

0111

Johannes B. Gruber @jbgruber.bsky.social · 26/06/2026

Finally back home again, sleeping between my loved ones: my wife, our dog, and our Midea Portasplit 12000 BTU 3.5kW air conditioner 💗

010

Johannes B. Gruber @jbgruber.bsky.social · 24/06/2026

So so much 🥲 This weather at 18°C is my perfect summer

110

Johannes B. Gruber @jbgruber.bsky.social · 24/06/2026

3 conferences in the last 3 weeks done! Was ist too much? Maybe, I'm definitely exhausted. But it was great to see so many colleagues doing new and interesting research, and I feel updated about what's happening in #commsky, data archiving, and #polsky

140

Johannes B. Gruber @jbgruber.bsky.social · 24/06/2026

I read an article in a German newspaper yesterday that passionately argued for age restricting social media. Just a reminder that this is a very stupid idea. pluralistic.net/2026/06/23/d...

pluralistic.net

Pluralistic: Spying on kids to save kids from spying is very, very stupid (23 Jun 2026) – Pluralistic: Daily links from Cory Doctorow

011

Reposted by Johannes B. Gruber

GESIS - Leibniz-Institut für Sozialwissenschaften @gesis.org · 22/06/2026

#pride26 #pridemonth This Pride Month we want to highlight two very different publications: a practical guide on how to account for gender diversity in evaluation, and a cartoon on what diversity looks like in everyday life. 🔗doi.org/10.31244/zfe... 🎨pushbacklash.eu/wp-c...

0103

Reposted by Johannes B. Gruber

Katrin Weller @kwelle.bsky.social · 19/06/2026

#cessda50 was such a great event - and lots of fun being there with my amazing colleagues. Thanks to @cessda.bsky.social for bringing the community together

181

Reposted by Johannes B. Gruber

EPSS @epssnet.bsky.social · 19/06/2026

About to begin-💡 EPSS Showcase: Using AI in Political Science Research virtual.oxfordabstracts.com/event/75765/... 15:20-17:00 | Lagan (ICC), with @kenbenoit.bsky.social, @mdigiuseppe.bsky.social, @jbgruber.bsky.social, Allison Wun-hui Koh, Simon Luck and Leonce Röth

virtual.oxfordabstracts.com

OA Virtual conferences

041

Reposted by Johannes B. Gruber

Katrin Weller @kwelle.bsky.social · 17/06/2026

Happy Birthday to @cessda.bsky.social ! Looking forward to an exiting conference progam here in Bergen, Norway.

072

Johannes B. Gruber @jbgruber.bsky.social · 13/06/2026

Einen Monat später immer noch nicht gefixt...

000

Reposted by Johannes B. Gruber

Johannes B. Gruber @jbgruber.bsky.social · 05/05/2026

@gruene.de ihr habt da irgendwas komisch eingestellt. Wenn man auf bsky.app/profile/grue... geht, kommt man zu @felixbanaszak.bsky.social . Oder soll das so?

bsky.app

101

Johannes B. Gruber @jbgruber.bsky.social · 08/06/2026

You now also have full control what shows up in link previews and alternative PDS like #eurosky are supported!

030

Johannes B. Gruber @jbgruber.bsky.social · 08/06/2026

Biggest new feature is the ability to stream from the firehose

Documentation for the stream_firehose function

110

Johannes B. Gruber @jbgruber.bsky.social · 08/06/2026

Announcing #rstats {atrrr} 0.2.0 is on CRAN now! 🥳

170

Johannes B. Gruber @jbgruber.bsky.social · 08/06/2026

Posting live from stage at #ica26 where I present our #rstats 🎁 {atrrr} which can interact with 🦋

0123

Reposted by Johannes B. Gruber

Jakob Ohme @dscheykopp.bsky.social · 06/06/2026

Hi #ica26. We have an open 4.5 yr Postdoc position @weizenbauminstitut.bsky.social Berlin in PolComm/Comp Social Scienes on opinion dynamics in climate topics. Hit me up if you want a to talk about it in CT or send to colleagues who might be interested. www.weizenbaum-institut.de/news/detail/...

01916