Pinned Tweet
nature.com/articles/s41586-0…
Are we past Replication Crisis?
I co-authored this study where Economics and Political Science, against original predictions, got convincing scores in terms of credibility of their scientific results.
I have a theory for this.
Giulio Giacomo Cantone retweeted
I have always liked the Plutarchian tradition of "companion pieces", reading 2 works together aids criticism, the "how can this be interpreted under the light of that other thing".
Some books from my weird education that "go together", emphasis on the classics + medieval:
Giulio Giacomo Cantone retweeted
The surplus generation (trailer for an upcoming movie), youtube.com/watch?v=bEjrEJ1H…
Giulio Giacomo Cantone retweeted
New working paper: "Do Modern Difference-in-Differences Estimators Change Published Findings in Business Research?"
Peter Hull's recent meta-analysis found that in economics, two-way fixed effects and heterogeneity-robust DiD estimators rarely disagree. Using Claude, I ran his protocol on seven business journals, screening 1,829 papers, and re-estimated every eligible design in Management Science and Marketing Science, the two with public replication archives. That gave 355 published staggered-adoption specifications from 40 papers, re-estimated with Callaway-Sant'Anna (CS) and Borusyak-Jaravel-Spiess (BJS). On the comparisons that pass Hull's leverage screen, from 28 (CS) and 29 (BJS) papers:
- Correlations between TWFE and modern estimates are 0.61 and 0.70, against 0.88 and 0.93 in economics.
- Estimates change sign in 25 and 18 percent of comparisons, and 10 and 7 percent show a statistically significant difference.
- Modern standard errors are 1.6 and 1.3 times those of TWFE.
- Disagreement concentrates in unbalanced panels and in designs without never-treated units.
Every table and figure rebuilds from the public repository; re-estimating a specification from scratch requires the journal's replication package.
Paper: lnkd.in/gXmtvfBC
Code, protocol, and estimates: lnkd.in/gZijx_dy
Giulio Giacomo Cantone retweeted
I made a little visualisation to try and help explain p-values for students. Some people might enjoy: psy104-p-values.netlify.app/
Giulio Giacomo Cantone retweeted
“Causal Effects Versus Causal Mechanisms: Two Traditions With Different Requirements and Contributions Towards Causal Understanding”
statmodeling.stat.columbia.e…
Giulio Giacomo Cantone retweeted
PRIN 2026, settore SH1. 7 progetti finanziati su 9 (!) hanno un PI di Bocconi o Bologna. I soliti 2 atenei monopolizzano quasi tutta la graduatoria. Meritocrazia? O è il caso di chiedersi quanto conti la cooptazione e il potere nelle reti accademiche? (Io non ho applicato in SH1)
Giulio Giacomo Cantone retweeted
Anthropic found vaults in bacteriophages, bacteria and archaea. This has to make vaults a foundational feature of life, on par with ribosomes. There aren’t many things that are shared so deeply. We still have no idea what vaults do. Preprint has cryoEM, site directed mutagenesis. Will probably be a Nature paper. Full dive into wet biology, buckle down guys.
Italian chef prefers to lose the job and get deported instead of serving ananas on a pizza.
Uno chef italiano, Davide Tamburella, è stato licenziato a Brisbane, in Australia, dopo essersi rifiutato più volte di servire pizza con l'ananas, considerandola un "insulto" alla propria cultura. Avrebbe mantenuto la sua posizione anche durante un evento da circa 10.000 euro. La Fair Work Commission, il tribunale australiano competente in materia di lavoro, ha respinto il suo ricorso, confermando la decisione del ristoratore Elliot Ingram anche alla luce di altri problemi comportamentali, tra cui insulti e scatti d'ira sul lavoro. La vicenda ha inoltre compromesso la procedura per il visto 186, che consente di risiedere stabilmente in Australia grazie alla sponsorizzazione di un datore di lavoro: Tamburella dovrà ricominciare l'iter
I am seriously worried about the negative impact that Bocconi is going to have on the development of Economic science in Italy. A single university got indirectly too much public funding. The trickledown model works when working on the Manhattan project, not on economic research.
Giulio Giacomo Cantone retweeted
Democrats won enough trifectas last night to theoretically obtain an absolute majority of the electoral college for the National Popular Vote Interstate Compact, and they certainly have Congressional majorities large enough to authorize such a compact.
The only question is whether it would survive a court challenge. If they win the Presidency in 2028 and stack the Supreme Court as many Dems have threatened to do the moment they have the votes, there’s a chance that we will never again have Presidential elections in America decided by the Electoral College.
I do not think people fully understand yet that Democrats won enough elections last night to permanently change this country in a way that may make things impossible to ever undo.
Political parties operate off incentives. Their goal is to win elections, not permanently pander to a specific demographic constituency. While they can spend years or even decades in the wilderness, eventually, someone will look at the median voter and determine that he too can win an election by pandering to everyone to the Left or Right of that individual.
So what happens when one side of the equation permanently changes the rules so that the median voter shifts from someone living in Ohio to say someone living in Virginia?
The Beautiful Losers may not immediately return to power tomorrow, but make no mistake they are quietly waiting in the wings, hoping that Democrats change the rules of the American electoral process in order to permanently enshrine them as the natural controlled opposition.
Swing voters I personally know (ok in Italy but it's the same) are like "I really really really miss the times when the Socialist Party ruled in Italy but I cannot stand the smell of kebab"
(To get it: when the Socialist Party ruled in Italy, Berlusconi was a member of it)
Giulio Giacomo Cantone retweeted
How do you know a research theme in 2015 is the "same" theme in 2020?
Most thematic evolution maps answer by comparing keyword lists. Our new paper in the Journal of Informetrics argues that is not enough. 🧵 #ScienceMapping
Lately I am driving much more so I am listening mainstream music on the radio; so I discovered the singing style of Roan Chappelle and she sings exactly as I expected her to sing (very uvular).



























