Listed here are a number of hyperlinks I must get off my chest as they’re burning a gap in my browser tabs. A number of love in right now’s posts, in addition to AI, however I don’t suppose as a lot love in the direction of AI itself, although perhaps. Thanks once more in your help of the substack! For less than $5/month, you can also get a gajillion posts per week about love and AI!
Love is available in numbers
The 777 rule: each 7 days you’ve got a date. Each 7 weeks you do a weekend away. And each 7 months you go someplace kind of for a giant getaway, perhaps for 7 days? I feel that was it. Anyway, I might get behind that sort of rule.
And listed below are 8 methods to carry heat again into your relationship. Considered one of them I used to be glad to see — take accountability in your personal unhappiness. I heard somebody say although that nobody may cause our happiness however they will trigger our unhappiness, which can be why the eighth factor in that checklist is to just accept when it might be time to exit.
Six issues {couples} can do to domesticate a thriving relationship. Quantity six is cultivating friendships outdoors of the connection — which is one thing I see always in these lists. As an illustration it exhibits up on this checklist of 5 issues too.
Individuals who love individuals study to just accept 4 issues about them.
White chocolate highlights.
Michael Pollan has a brand new ebook popping out about consciousness. Right here’s a NYT interview with him. There’s additionally one thing in Nature masking it too.
Attention-grabbing sounding memoir by the previous CEO of Sony (now chairman of Snap). He recounts the worst mistake of his profession which was to inexperienced gentle the Seth Rogen and James Franco film the place they hatch a scheme to assassinate the chief of North Korea. That led to a serious hack the place practically all of Sonys servers have been destroyed, scripts, emails, films, well being information, social safety numbers — principally every part they may seize — was launched on-line. The ebook sounds prefer it’s of him processing it and his entire life.
Peter Hull contributing to IO.
What lecturers can study from comics and performers.
Right here’s a information for freshmen wanting to enhance their espresso bean shopping for recreation.
Stephanie Eckman on the usage of artificial knowledge in social science (a deck).
Tabaré Capitán, an economist, on his expertise writing his first python bundle.
Claude code starter kits for researchers
Right here’s a few model new Claude code starter kits for researchers, each by coauthors and mates. First, Jared Black, has written a extremely nice piece on workflow for AI-Assisted analysis that a few of you could discover helpful.
And Christopher Cornwell on the College of Georgia has a guide for empiricists (economists, knowledge scientists, everybody) on Claude Code that additionally dives into the relative usefulness of Code over Cowork, which is the primary starter pack I’ve seen anyplace that’s diving into that.
The finest cordless stick vacuums from Wirecutter.
Again in BBS days from the late 80s, ASCII was king. Here’s a deep dive into ASCII graphics.
Join Claude or ChatGPT on to Indias official authorities statistics.
Harvard physicists use a chatbot as a coauthor on a paper bc of a serious discovery made collectively. New instances.
The historical past of the Kaplan-Meier survival curve.
Evaluating Codex and Claude Code, and when one gave the impression to be stronger at what sorts of duties and why.
A few of you could have been following my video sequence the place I’ve been sending 300,000 Congressional speeches and presidential communications from 1880 to now. This wasn’t precisely a replication — extra like I wished to indicate use Claude Code to submit texts to OpenAI through batch requests (to get the 50% low cost), get it again inside hours and spending solely $11, and have all of it analyzed. I simply was attempting for instance one other use case for Claude Code in different phrases. However what occurred was that I used to be genuinely shocked to seek out that gpt-4o-mini had reclassified round 100,000 of the 300,000 speeches that the unique authors had performed with their very own RoBERTA method (which used round 7 college students who annotated 7500 speeches after which predicted the remaining), and but the reclassification apart yielded the identical outcomes. After which I spent much more substacks attempting to nail down as to why that was, and so the sequence kind of branched off.
Nicely, take a look at this new paper by Asirvatham, Moski and Shleifer. They seem to supply extra proof for this method of utilizing the one-shot GPT classification strategies. Now they’re utilizing some sort of GPT software program known as GABRIEL, so perhaps there’s some good points utilizing GABRIEL over gpt-4o-mini. However put that apart — it’s extra proof that GPT can be utilized to categorise texts for educational analysis and since I did it for under $11 and in round 2 hours on 300,000 speeches, I feel it speaks volumes to how AI goes to democratize analysis potentialities to under-resources researchers globally. However not solely that — I guess it value the Card, et al. staff upwards of $10,000+ for 7 college students to label 7500 speeches. And for what? To assign a -1, 0 or +1 to a speech. Nicely, simply wait till you see what I’ve been engaged on with a serious mission I’m nearly able to launch. You then’ll see we badly want the marginal prices of classification to fall so that really significant classifications can occur. LLMs are velocity readers with excellent reminiscence and impeccable analytical expertise. And as David Autor has mentioned, they will extract the tacit data in human communication — issues which can be fairly onerous for anybody to tug out or clarify. So that is simply one other public service announcement that Claude Code will help you stage as much as the place the place you’re processing these texts in scale, and I feel we’re going to most likely see that coming very quickly, if it’s not already right here now. Assume it’s right here now really.
New York Instances opinion piece reviews on Claude Code saying it and brokers prefer it was the AI disruption we’ve been advised was coming. This essay on Twitter went viral and I believe the errors are correlated with each popping out on the identical time.
One paragraph from it.
That is weirdly instantly gripping. It’s an animated historical past of the Tower of London going again to 50 AD when there was no tower there finest I might inform.
I’ve determined to consider that this thread occurred not by means of an LLM, however good quaint detective work geared toward completely pointless and but useful duties.
Dario Amodei, CEO of Anthropic, has an extended and severe essay from January 2026 on “highly effective AI” which can be practically right here and which has an honest quantity of unhealthy stuff in a dense stack of chance on the tail.
Right here’s Jim Heckman at Texas Tech this week. Wreck ‘em! Pew pew! (HT LinkedIn)
Undecided what this YouTube ought to be about besides that it’s a Bloomberg podcast about Claude Code from a month in the past. So even journalists are discovering Claude code and being blow away.
Ethan Mollick had I feel both Claude code or ChatGPT make a sequence of PDFs containing all of the weights for GPT-1, that are offered fantastically right here.
Ethan additionally says perhaps suppose twice earlier than utilizing another person’s /talent.
Boris Cherny, creator of Claude Code, thinks laptop engineers as a job title and a human position is just not going to be right here after this yr. Observe it isn’t that the people are gone however quite the duties they did will now be automated. And that’s coding.
Right here’s a giant cushion for sitting on the ground.
College of Western Australia is internet hosting its annual Labour Econometrics. Keynote audio system are Muriel Niederle and Orley Ashenfelter. The Key dates (AWST) are:
-
Paper submissions open, Registrations open: Tuesday, 17 February 2026
-
Submission deadline: Friday, 10 April 2026
-
Notification of outcomes: Friday, 24 April 2026
-
Registrations shut: Thursday, 9 July 2026
-
Workshop dates: Thursday, 16 July – Friday, 17 July 2026
There’s three issues up to date AI can not try this people can do and that’s within the space of long run planning, consistency and study persistently. (Some people anyway).
Marriage and the intergenerational mobility of girls.
AI relationship apps might be a matchmaker, however they received’t use swiping. I’ll be curious to see what this subsequent stage of relationship markets turns into.
Success will depend on creating expertise of tolerating uncertainty and the unknown.
Microsoft AI CEO says most white collar jobs might be automated inside 18 months.
Taylor and Travis are getting married on the Ocean Home, a wide ranging resort on the coast of Rhode Island, and there’s a penthouse there that may be yours for under $20m! I proceed to dream of dwelling on the coast of Rhode Island.
The nice Tom Hagen, aka Robert Duvall, has handed away at 95. One of many biggest actors in Hollywood historical past.
In the event you’re ever in Waco texas, take a look at Kitok‘s, a 50yo establishment within the metropolis with wonderful burgers and fries. However first examine if it’s open.
Individuals who lived into their 80s adopted these 8 issues. (I like these numbered articles).
Gen Z prefers a great evening sleep to intercourse.
Is gettyimages the very best picture web site ever?
Spotify builders are not coding — they haven’t written new code in a month. They’ve AI brokers doing it as a substitute. Relatively they’ve moved to supervisory roles solely. That’s btw a partial equilibrium not a basic equilibrium.
The essential parts of that means and objective are to do with staying true to what issues most to you. And the second you cease doing that, that’s when every part turns into hole. Have Claude learn this, after which ask him to interview you one query at a time till he can articulate what your factor is. Mine was my curiosity.
John Lennon’s piano at Christie’s for round half one million.
AI video of Brad Pitt combating Tom Cruise.
Through the federal shut down, squatters and unlawful BASE jumpers confirmed up at Yosemite Nationwide Park. Noticed a few of that on Untamed final evening, which I extremely suggest on Netflix. Right here’s a story of a man who overstayed his welcome on the park.
The story of the discovering knowledge fabrication by an outdoor forensic audit of a excessive profile publication predates the Gino case and Information Colada. A decade in the past, it was the LaCour (2014) examine which was found to have been fabricated knowledge, found by then grad college students Brookman and Kalla, with Peter Aronow, revealed as a working paper titled “Irregularities in LaCour (2014)”. That was the primary time I had ever seen R Markdown. I’ve the scholars replicating this forensic train, the unique LaCour and Inexperienced Science, in addition to the observe up RCT by Brookman and Kalla (2016) in Science. These are my slides.
I first realized about Peter Aronow by means of that “irregularities” paper and now use his ebook with Miller, Foundations of Agnostic Statistics, for my Gov 2001 class. Thursday I taught all three papers, and college students have a homework project to duplicate all of them for Gov 51. You will discover the papers, slides and so forth right here.
Talking of foundational statistics. I used to be reminded of this ebook the opposite day whereas prepping my chance class. Undecided what popped it into my thoughts. It’s by Leonard Savage and it was a break by means of in trendy statistics. The Foundations of Statistics. The ebook was thought-about one of the essential within the Bayesian statistics mission ever.
I’m considering of getting a small banquet right here on the apt serving brisket tacos, white queso and chips, salsa, perhaps guacamole and margaritas. Right here’s a 32 oz mason jar for when you’ve got mates over and wish one thing totally different. However perhaps these 16 oz coloured ones are extra your velocity? Or maybe these extra conventional heavy base glasses are higher. That is the Dutch Oven I acquired for braising the brisket. I made a decision to not get the highest of the road Le Creuset one as a result of I figured why try this after I’ll hardly ever use this once more, however that’s the one we had. I’m debating having door prizes (I may need gone overboard). Visitors could depart with Texas Blue Bonnet seeds, some De la Rosa Mazapan peanut sweet, and perhaps these magnets of Texas. I wished my new mates to have a little bit little bit of my world.
I’m going to make grilled cheese sandwiches one evening and determined to get this unnecessarily heavy 3.5 pound chrome steel burger smasher to do it. It’s like a huge slab of steel and I like it.
Don’t overlook in regards to the coming deadline for papers on the Journal of Financial Psychology on Human-AI interactions. You continue to have time! Observe they’re searching for empirical papers, experimental and quasi-experimental. It’s a slim subject, however you should still have the precise stuff.
OpenAI additionally has a chunk on leverage Codex higher on this AI-driven analysis world that I feel is beneficial.
Extra on aphanasia, and what’s lacking and the way perhaps it really works. Ask Claude to interview you one query at a time utilizing actual survey questions after which announce on the finish you probably have aphanasia and in that case what does it imply? Simply be sure to say one query at a time.
And that’s about it! Have a fantastic weekend. Keep hydrated and don’t get sick!













