A very straight line on a loglog plot. So straight that even Aaron Clauset would call this a power-law
Super excited about our new paper on mobility that's out in Nature Human Behavior www.nature.com/articles/s41...
I love this paper for many reasons, but one is that we find beautiful 1/x power-law that spans 6 orders of magnitude hidden within the "ugly" distribution raw mobility data.
🔊 More from our recent @nathumbehav.nature.com article from the Technical University of Denmark: Our study shows that behind the apparent complexity of human mobility lies a simple rule shaped by geography and distance.
🔗 www.dtu.dk/english/news...
DOI: 10.1038/s41562-025-02282-7
In this article, Boucherie et al. apply physics-based models to the arrangement of locations to study how geography shapes human movement. They find an underlying pattern in how people choose to move, independent of geographical layout.
Amazing collaboration with brilliant co-authors Benjamin F. Maier (also the artist behind the figure below) & @sunelehmann.com
Looking forward to your thoughts, and let me know if you’d like to collaborate on follow-up work on mobility and the pair distribution function!
Mobility within a city is governed by one power law (which tells us that distance matters less), while another, steeper power law describes moves between cities, reflecting the stronger deterrent of greater distances.
When considering mobility centered around a single city — rather than mobility of an entire country — a more nuanced picture emerges. Instead of a simple global power law, we observed a universal piecewise behavior.
Finally, we link these findings to the well-studied gravity model by extending it to a continuous setting that does not rely on arbitrary administrative units. However, the story does not end here.
The fractal shape of cities explains the meso-scale of the pairwise distance distribution; and at the large scale, we show that city positions are indistinguishable from uniformly random.
Using the pair distribution function, we treat addresses as particles and develop a statistical physics of locations, from the micro-scale of buildings to the macro-scale of cities. By modeling buildings as an ideal gas in a potential, we reproduce local urban densities.
By using a simple tool from physics—the pair distribution function—we separate the influence of geography (coastlines, rivers, road networks) from people’s mobility choices. Once you factor out the map, the remaining behaviour follows a striking, universal power law across five orders of magnitude.
* Decoupling geographical constraints from human mobility *
🌍🚶 How much of our movement is about human choice and how much is constrained by geography and the spatial layout of locations?
Our paper (out in Nature Human Behavior) gives you a practical way to tell:
www.nature.com/articles/s41...
Sketch showing the street network simplification process.
Finally, the preprint+Python package "neatnet" is out: https://arxiv.org/abs/2504.16198
https://github.com/uscuni/neatnet
If you are working with street/planar/spatial networks, this will solve *so* many problems! Many of my projects had this bottleneck - […]
[Original post on datasci.social]
Join us for a Data Discussion on April 25! 📅
In this session, Louis Boucherie will discuss the colours of fashion in a fascinating presentation based on large-scale computational analysis, exploring the dynamics of fashion trends over time 🔎 👔
Link🔗: sodas.ku.dk/events/sodas...
Butterflies on a blue sky
THREAD
The numbers are in. @bsky.app research sharing volumes vs X Formerly Twitter
In March 2024, on most days, Bluesky hosts more posts linked to research published in 2025 than X.
By quite a lot.
Release the Kraken...
#AcademicSky #HigherEd #Altmetrics
1/11
This was a fun experiment we conducted while developing The AI Scientist-v2. With the permission of ICLR, we submitted an AI-generated paper to an ICLR workshop that passed the peer-review process.
We documented the entire process and what we have learned in a blog post: sakana.ai/ai-scientist...
Map showing a highway section in red and social ties in space crossing the highway. Wherever a tie crosses the highway, there is a cross. There are 94 crosses.
🎉 New paper in PNAS: Urban highways are barriers to social ties
https://www.pnas.org/doi/10.1073/pnas.2408937122
Highways are barriers that cut opportunities for social ties. We quantify this effect by overlaying the US highway network with millions of social ties from Twitter.
🌊 Today in @nature.com: Is the AMOC on the brink of collapse?
Unlikely before 2100—but the risks are real 🚨
We find Southern Ocean winds keep this vital ocean "heat engine" running, even under extreme #climatechange. But the Pacific holds a surprise…
tinyurl.com/yt6u4e7d
Let’s explore 🧪👇
And we are rolling! Submissions are open and will be accepted on a rolling basis!
Invited speakers include @tiago.skewed.de and Renaud Lambiotte! All info on our website! 💪🏼 Spread the word 😁
The satellite will be a half day satellite on Tuesday afternoon!
signet-friends.github.io
Can simple closed-form mathematical models predict human mobility as well as deep learning? In a new paper in
@naturecomms.bsky.social we show that the answer is YES
Human mobility is well described by closed-form gravity-like models learned automatically from data www.nature.com/articles/s41...
DeepSeek-AI, Daya Guo, Dejian Yang, Haowei Zhang, Junxiao Song, Ruoyu Zhang, Runxin Xu, Qihao Zhu, Shirong Ma, Peiyi Wang, Xiao Bi, Xiaokang Zhang, Xingkai Yu, Yu Wu, ...
DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning
https://arxiv.org/abs/2501.12948
More than 40% of postdoctoral researchers leave academia, according to a study of some 45,500 researchers’ careers
https://go.nature.com/3Ej9RxA
We're hiring a postdoc in the &-Lab at Northeastern's Network Science Institute!
Looking for a curious, collaborative scholar to work on computational social science questions, at the intersection of data justice + network science.
northeastern.wd1.myworkdayjobs.com/careers/job/...
Have been using Cursor for writing LaTeX documents for a while. The idea was simple: Cursor is helpful for coding, and TeX is code in a way, so I gave it a try, and it worked perfectly.
It does require some configuration, so I'm sharing mine here: github.com/yang3kc/curs....
1/2
Poster for a webinar
Join the talk to hear more about transformers for social science: www.soc.cuhk.edu.hk/event/nov-14...
