Lorenzo Frattarolo
frattalol.bsky.social
Lorenzo Frattarolo
@frattalol.bsky.social
Advocate of the random generation, dad of twins and econometrician @ univr
Reposted by Lorenzo Frattarolo
Academia may not give you job security, flexibility, or wealth, but it will let you unexpectedly connect to eduroam in foreign cities
August 20, 2025 at 4:58 PM
Reposted by Lorenzo Frattarolo
New paper: Rocco Caprio and Adam Johansen have had their work analysing the EM algorithm by using tools from functional analysis accepted by Biometrika: academic.oup.com/bio...
August 19, 2025 at 12:05 PM
Reposted by Lorenzo Frattarolo
Great post on the risk of Google Scholar disappearing!
Obligatory self promotion, here’s my own from last year on possible alternatives blogs.lse.ac.uk/impactofsoci...
August 13, 2025 at 8:39 AM
Reposted by Lorenzo Frattarolo
People have been saying things about Google Scholar, so here are some of my thoughts, I'll follow up with tools and alternatives when I have time. I work with data about research outputs and their impact so I've spent a lot of time considering systems which capture citations. (1/7)
September 2, 2024 at 5:06 PM
Reposted by Lorenzo Frattarolo
"If you’re still not convinced, it’s probably because you think AI is a feature that adds value. That’s all wrong. It’s not a feature; it’s more of an idea."
www.mcsweeneys.net/articles/our...
Our Customers Demand Terrible AI Systems
We’ve been banging our heads against the wall, trying to think of the new “it” thing our customers want. At one point, somebody suggested improving...
www.mcsweeneys.net
February 9, 2025 at 11:15 AM
Reposted by Lorenzo Frattarolo
I have told anyone who asked for years that the most impactful open problem in causal inference would be a reverse mode AD for the Von Mises Semiparametric functional calculus.
This replaces an enormous amount of tricky manual derivation of optimal estimators with a simple script.
Here it's done!
New paper!

arxiv.org/abs/2405.08675

tldr: automatic differentiation can be used to derive efficient influence functions and construct efficient estimators.
June 25, 2024 at 5:17 PM
Reposted by Lorenzo Frattarolo
*As a career sci-fi writer with a fondness for props and gadgets, it's been pretty common for me to meet tech people who say, "Hey remember that thing you wrote about in that story, well I just built one of those"

*However, it's never the thing I had in mind, and always the thing he had in his mind
December 29, 2024 at 8:03 AM
Reposted by Lorenzo Frattarolo
Particle physicists rely heavily on statistics in our research; we deal with large amounts of data. The Combine tool in CMS is the go-to statistical tool for CMS. This highlight of 2024 #CMSPaper describes Combine using the statistics of the Higgs boson discovery 👉 https://buff.ly/3ywtCPz
December 28, 2024 at 6:22 PM
Reposted by Lorenzo Frattarolo
In places most prone to wildfires and hurricanes, state “insurers of last resort” are absorbing trillions of dollars in risk.
There’s a New Financial Crisis Brewing in Uninsurable US Homes
In places most prone to wildfires and hurricanes, state “insurers of last resort” are absorbing trillions of dollars in risk.
www.bloomberg.com
December 28, 2024 at 5:59 PM
Reposted by Lorenzo Frattarolo
Too Much Fun finally tells the C64’s legacy
Too Much Fun finally tells the C64’s legacy
Ryan Rigney on Jesper Juul’s new book about the forgotten history of early home computer games
buff.ly
December 13, 2024 at 7:20 PM
Reposted by Lorenzo Frattarolo
I Datacenter sono un po' come le/i sex worker, non esistono. Ma esistono e fanno girare un sacco di soldi.
La lentezza dei governi ad adattarsi ai tempi che corrono è assurda. www.ilpost.it/2024/1...
Per la legge italiana i data center non esistono - Il Post
È un grosso problema, che sta rallentando e complicando molto investimenti miliardari nel paese
www.ilpost.it
December 2, 2024 at 9:34 AM
Reposted by Lorenzo Frattarolo
When your appendix has 68 (underpowered) robustness tests
November 26, 2024 at 2:47 PM
Reposted by Lorenzo Frattarolo
Spotting and fixing data issues: how we help improve data quality on and off our publication
November 26, 2024 at 5:31 PM
A big thanks to all the #packers. #starterpack s are even more valuable than the old bird's #FF.
November 24, 2024 at 6:40 PM
Reposted by Lorenzo Frattarolo
November 24, 2024 at 6:19 PM
Reposted by Lorenzo Frattarolo
this does raise some very interesting questions
November 24, 2024 at 2:04 AM
Reposted by Lorenzo Frattarolo
Hiya. Search subject matter plus starter eg "musicians + starter" and you get packs like this

bsky.app/starter-pack...
November 24, 2024 at 1:10 PM
Reposted by Lorenzo Frattarolo
In a commentary on Dave Donoho's "Data Science at the Singularity," I summarize this history.

hdsr.mitpress.mit.edu/pub/8dqgwqiu...
The Mechanics of Frictionless Reproducibility
hdsr.mitpress.mit.edu
November 22, 2024 at 2:43 PM
Reposted by Lorenzo Frattarolo
Quick econ tip: here's a list of over 100 economics conferences: docs.google.com/spreadsheets.... Curated by @anne-m-burton.bsky.social and @bartonwillage.com.
Economics Conferences (Compiled by Anne M. Burton and Barton Willage)
docs.google.com
November 18, 2024 at 3:23 PM
Reposted by Lorenzo Frattarolo
🚨 ALERT 🚨
Are you looking for (recent) longitudinal social media data?

We recently released a huge #bluesky anonymized #dataset of social interactions and contents covering 80% of the service users for 13 months.

Available on Zenodo! Read more: arxiv.org/abs/2404.18984
#academicsky #NetSci
"I'm in the Bluesky Tonight": Insights from a Year...
Pollution of online social spaces caused by rampaging d/misinformation is a growing societal concern. However, recent decisions to reduce access to social media APIs are causing a shortage of...
arxiv.org
May 14, 2024 at 8:23 AM
Reposted by Lorenzo Frattarolo
If you have the feeling that @bsky.app is different, maybe you are right.

The current migration is breaking the typical rich-get-richer effect wrt other platforms #scaling

Data:

Twitter: 10.9M users
BlueSky (Mar 2024): 4.1M users, 🙏 @andreajpg.bsky.social
Mastodon: 3.8M users, 🙏 @tiago.skewed.de
November 17, 2024 at 7:50 PM
It's finally out! In our new paper in Statistical Science, we show that almost everything called antithetic sampling in the past is sampling on well-chosen segments in the unit hypercube. Check it out! #math #stat #MonteCarlo
Living on the Edge: An Unified Approach to Antithetic Sampling
We identify recurrent ingredients in the antithetic sampling literature leading to a unified sampling framework. We introduce a new class of antithetic schemes that includes the most used antithetic p...
projecteuclid.org
February 22, 2024 at 9:26 AM