Take a look what's under the hood intheweights.com

This week's post: Why governments should not be in hock to the bond market mainlymacro.blogspot.com/2026/06/why-... Policy based on fears about the bond market, or worse about what journalists are writing about that market, will be bad policy. This has become one of Andy Burnham's first tests.

This week's post: The Consequences of Tolerating the Far Right mainlymacro.blogspot.com/2026/06/the-... It is so easy for the media to give the impression that violent crime is associated with immigrants or asylum seekers by publicising selected cases. Media that does this should be called out.

In _The Ideas Letter_, Leif Weatherby, Tyler Shoemaker, and I wrote about the creation of reality through pseudoscience, religion, and psychosis. Of course, the piece is mostly about AI.

I’ve officially resigned as Associate Editor for Frontiers in Systems Neuroscience. It used to be a reputable journal, but became a case study in how forced automation destroys academic integrity. 👇

I've apparently been erased by AI slop. A paper cites: "Agarwala & Coyle (2024), Challenges of wealth-based sustainability metrics: A critical appraisal" The actual paper is by McLaughlin, Ducoing & Hanley. Same title, journal, and year, but the AI hallucinated the entire authorship.

Backstory: we originally wrote this paper in response to 2013–2016 Ebola outbreak, but by the time it worked its way through peer review that outbreak had largely subsided and the paper was rejected as no longer timely. We shelved it until the next relevant outbreak, which happened to be mpox.

1. How common is LLM use in scientific publishing, and how does it vary across field, publisher, journal prestige, author demographics etc.? @kylesiler.bsky.social has new paper in PNAS that addresses this question on a massive scale: 7.3 million papers from Elsevier, PLOS, MDPI, and Frontiers.

1/ Can AI help researchers check whether published social science results actually reproduce? In our new PNAS paper, we tested this directly in the AI Replication Games: 288 researchers, 103 teams, and real replication packages from quantitative social science.

Deus ex machina: Pope Leo says AI must be 'disarmed' in first major teaching www.bbc.co.uk/news/article...

IMHO possibly the most interesting question in finance and economics right now is the extent to which future AI dominance can retain any monopoly characteristics. People assume because it's "Silicon Valley" that those explosive network effects will necessarily pertain. They really might not... 1/

"Taking macroeconomic advice from bond market traders is like taking advice on how to deal with pandemics from undertakers."... Great quote from @sjwrenlewis.bsky.social #economics #endausterity

Lydia Ashton (University of Wisconsin-Madison): Vibe Econometrics and the Analysis Contract https://arxiv.org/abs/2605.08071 https://arxiv.org/pdf/2605.08071 https://arxiv.org/html/2605.08071

#RomanSiteSaturday — The Roman lighthouse at Patara in Lycia (Türkiye), built under Nero beside the ancient harbour. Probably destroyed by an earthquake in the medieval period, it has now been reconstructed using its original blocks, a rare Roman lighthouse brought back to life in situ.

Maybe teaching #rstats was a mistake? When I first advocated for teaching R in under/grad psych, it felt like the right move: SPSS was rigid and limited only to what can be selected from a stat-analysis decision tree; R on the other hand... 1/

A different perspective. Always helpful.

This is the clearest and most accessible introduction to DML I've ever read: arxiv.org/abs/2504.08324 Congrats to the authors @aahrens.bsky.social, @markeschaffer.bsky.social, et al on this amazing paper and great accompanying R package! I'm now DML-pilled.

Some 3,300 years ago, an ancient Egyptian artist painted this comic illustration of a cat herding geese! 😻 They used a flake of limestone as a sketchpad for this fun scene. From Deir el-Medina, Thebes. 📷 Egyptian Museum, Cairo. #Archaeology

so i guess as a pulse check here's where I'm at on AI: 1) i don't think the AI corps are going to successfully take over the world. local models are too accessible. not enough moat.

When does using AI to produce research slip into fraud? And why does so much AI-for-science 'optimism' actually rest on a deeply pessimistic view of scientific agency? 🤔 New post: statmodeling.stat.columbia.edu/2026/04/22/f...

Swiss court has ruled that vegan milk substitutes are neither allowed to be called “milk” nor (literally) “not milk” which I guess technically leaves the set of admissible names for vegan milk substitutes empty. www.srf.ch/news/schweiz...

But I would have gotten the haircut right.

🧪 #FossilFriday Mummified baboon, Namibian cave, prob lost beyond the roosting chamber. Re: taphonomy of Homo #Naledi, this baboon was +70m in, via small tunnels, and despite humid cave it formed in arid ledge microclimate. as did a 2nd one, in <5 yrs. digitalcommons.usf.edu/cgi/viewcont...

Happy 75th anniversary to the Fender Telecaster! Checking, it turns out I’ve had a few over the years.

People have long predicted that IT would greatly reduce search frictions, including in labor markets. Surprisingly, it hasn’t. One reason is cognitive limits. But another is adversarial adaptation, including signal jamming. In many markets, an arms race develops, blunting any efficiency gains.

New post: Idiot Wind mainlymacro.blogspot.com/2026/03/idio... What Trump’s pointless war against Iran tells us about the US government, the UK political right and the mainstream media in both countries.

I don't think we will have a Journopoclypse open.substack.com/pub/joshuaga...

To elaborate on this a little in a health economics context: I do some work on the effectiveness of algorithmic prediction on pt outcomes. Identifying both mechanisms and outcomes is an *incredibly* complicated process, not least because patients are always being intervened upon. > #statsky #episky

Where have all the comments gone? For decades, the American Economic Review regularly published formal comments — papers that replicate, reassess, or challenge earlier AER articles. In our latest blog post, we show: they’ve nearly disappeared.

Is this bad

New post: Budgets are about much more than fiscal sustainability mainlymacro.blogspot.com/2026/02/budg... The impression that the OBR forecast sets fiscal policy is because politicians and the media have ignored or downplayed other key aspects of what fiscal policy should be about.

This is stenography, not news reporting. The BBC ought to provide some context, such as putting the scale of the changes which Reform claim are possible into context. This omits the major policy of revoking ILR and deporting legal migrants at unprecedented scale. www.bbc.co.uk/news/article...

New post: How Labour makes fighting right wing populism harder mainlymacro.blogspot.com/2026/02/how-... 'Tilting left' won't help. Labour has lost much of its base not so much because it hasn’t been left wing enough in economic terms, but more because it has sounded and acted in an illiberal way.

Interviewers were paid per interview, not per hour, so they nudged respondents to skip those questions. Make the question less burdensome, train or monitor interviewers, or (radical idea) pay them properly and missing friends reappear. 3/

Psychologists following a third-variable control strategy that may be best described as "sequential."

New post out: We have a guest post today from the excellent @dsquareddigest.bsky.social. "Build the Rail! Save the snails!" Or how we don't need to sacrifice the environment to speed up our mad planning processes. (Free to read) open.substack.com/pub/samf/p/b...

I've written about the UKRI funding situation and the 'buckets' explainer, looking in detail at QR, STFC and Quantum. www.ersatzben.com/p/the-bucket...

Weekend plug for my new personal post: Why does regulation often feel so toothless? Sometimes regulators are just useless, but more often the problem is that governments don’t give them the tools or the powers to deter bad behaviour

Today we’re looking back at Episode 165: ‘Dark Star’ by the Grateful Dead. Cocking in at 4.5 hours long, Andrew delves into the career of the American rock band - but also the historical and cultural context. Listen on Spotify, Apple Podcasts and via the website: 500songs.com/podcast/epis...

A reminder to the news media: “conflicting accounts” is what you say BEFORE the incontrovertible video evidence appears. After that, your job is to ask why one side is lying, not to repeat the lie and pretend no one knows the truth.

Brutal and excellent www.ft.com/content/6f84...

The BBC coverage is absolutely terrible. It leads on "sharply contested narratives" It has a dramatic skew to the US government It has posted the video but has failed to report on what it shows: it shows the US govt account is untrue

What a Long Strange Trump It’s Been Uncle Don’s hair has nary a touch of grey, but he and Epstein shared the women, shared the wine. And his “new” right-wing boogie? It’s a hand-me-down. by Maureen Dowd

So this is sort of fascinating because on the one hand obviously the Roman imperial senate was a shadow of what, say, the Senate of the third and second centuries BC had been, but at the same time it was a pretty crucial component of imperial governance into the third century.

You can see some of these stereotypes - the simple, morally pure countryside vs. the morally compromised, inauthentic city - play out in Greek and Roman literature. So this is a very old idea that recurs regularly.

Tell all the truth but tell it slant— Success in Circuit lies Too bright for our infirm Delight The Truth's superb surprise This paper finds poetry is a universal single shot jailbreak for LLMs. Systems built to stop prosaic attacks fail when the request is phrased in verse arxiv.org/abs/2511.15304

Incredible headline.

When asked, most people (72.4%) living near lynx in Norway say they LIKE living with lynx. Just 8.3% say they dislike lynx. In Scotland, where we're currently denied the opportunity to live near lynx, the majority of people WANT to see lynx reintroduced.

The automated problem factory!