PaCSS (Politics and Computational Social Science) is back! APSA preconference at BU joint with Political Communication, deadline for submission is May 8. Should be great!
cssi.umass.edu/pacss2026
I think your link got cut off - is this the right one? kylesaunders.substack.com/p/ai-and-pol...
Judge Lin says it's not like Anthropic can secretly update the software without DoD's consent. DOJ says there's an audit underway to better understand what may be in the platform. Essentially no one seems to get how AI works.
This could be a great way to build a reliable classifier for domain specific detection. Use the "can you do X papers in Y time" to find people that are v good at the task, and then show them unlabeled but suspected LLM generated papers.
Show us a leaderboard!
🚨🎉Excited to announce that our paper “Grok in the Wild: Characterizing the Roles and Uses of Large Language Models on Social Media” is accepted at
@icwsm.bsky.social 2026! In this paper, we investigate how, when, and to what effect Grok is used on X.
Yes! We're working on trying to measure sustained vs abandoned sci software based on some rough heuristic categories like paper scripts, lab tools, disciplinary infra etc
Give @yang3kc.bsky.social the best reviewer award now!
If you want to collaborate on anything above get in touch.
Also, please use our models and data. There are lots of open questions about software in science, and we need more rigorous empirical work to answer them ✌️
We're working on building a larger corpus of paper and software repository pairs, linking funding to repos, and a suite of metrics related to software quality, sustainability etc.
Open licenses and publicly available:
- Models: doi.org/10.5281/zeno...
- Code (and most data): doi.org/10.5281/zeno...
- Data (all): doi.org/10.7910/DVN/...
There are many potential confounders for this negative relationship. We hope this finding encourages more work on career trajectories and software contributions.
Our most depressing result ...
We find, paradoxically, that there is a significant *negative* relationship between frequency of code contribution and an individual’s h-index
That is, for an individual scientist the more they contribute to software development the smaller their h-index
There is a small but significant citation advantage for each additional author who contributes code to a paper.
Put another way, papers with more authors contributing to software tend to accumulate more citations
(This is not a causal claim and there are potential confounders)
Some findings we think are interesting:
- >25% of papers have code contributors that are not authors (this is stable across time and field)
- On average papers that produce software have just one code contributor
- 70% of all 1st authors contribute code (only ~29% for corresponding authors)
In short, we wanted to study collaboration and the allocation of credit (at both group and invidual levels) in research that produces software... To do that we have to connect papers to repositories, and authors to their developer accounts (more details in our paper on how we did this at scale)
'Code Contribution and Credit in Science' - new paper led by @evamaxfield.bsky.social w/ @isaacslaughter.bsky.social
1. We build a model that matches scientists (authors) with their GitHub accounts
2. We apply our model to ~140K paper-software repository pairs
direct.mit.edu/qss/article/...
An image of beautiful Boulder, CO, USA, with the iconic flatiron rock formations and red roofed campus.
The International Conference on the Science of Science & Innovation #ICSSI2026 will be in beautiful Boulder, CO, USA!
☀️ June 29 - July 1 ☀️
& Open Data Hackathon, June 28 🧑💻
We welcome submissions on all topics in the science of science and innovation, broadly defined.
Learn more at 👉 icssi.org 👈
also a v good novel
This is good - adopting LLMs to write consumer complaints results in better relief.
I am really excited to co-organize this year's Science of Science Satellite of NetSci 2026 in Boston.
Submissions are very open and can include work in progress, work under review, or accepted publications.
Please consider submitting before March 18th!
Website: netscisci.github.io
Excited for the Networks of Science of Science satellite at NetSCI 2026 this year in Boston!
netscisci.github.io
Be sure to submit by March 4th!
@dwillis.bsky.social - Is there an archive of pol campaign SMS messages? For some reason, I remember you working on this... But I might be conflating this with the 10000 other projects of yours that I've envied :)
You can change the spinner verbs in CC by editing `~/.claude/config.json`
Mine follw the $cash theme
{
"spinnerVerbs": {
"mode": "replace",
"verbs": ["The first day I got me a fuel pump"
,"And the next day I got me an engine and a trunk"]
}
}
www.youtube.com/watch?v=uErK...
multiplex > multitask
too many polecats not enough deacons
"do you have 5 minutes to chat?"
Extremely good visualization of NBA games. I especially like being able to see the +/- of different lineups over the course of a game (and not just because it supports my thesis that Devin Vassell is the engine that makes SA run) hoopflow.limboy.me?tab=games&ga...
Good idea _ Out of curiosity what set is this from?