iiinova — where your digital self keeps evolving

Ulrich Atz @ulrichatz.org · 29/09/2026

Now this is news I can get behind: Valve boss Gabe Newell admits his Mongolian throat singing phase was "annoying," and he's apparently not very good at it, either: "I am entirely self taught and it shows" www.gamesradar.com/games/valve-...

gamesradar.com

Valve boss Gabe Newell admits his Mongolian throat singing phase was "annoying," and he's apparently not very good at it, either: "I am entirely self taught and it shows"

He even sought out Paul Pena's help

000

Ulrich Atz @ulrichatz.org · 31/08/2026

Incredible measurement “coincidence” … and stellar work by the data colada team again.

000

Ulrich Atz @ulrichatz.org · 13/08/2026

Amazing resource, and freely available with a NC license.

010

Ulrich Atz @ulrichatz.org · 04/08/2026

It is underappreciated how much US pop-capitalism has changed. For example, Democrats favorable to the socialist (DSA) arm are now the most pro-trade group. datawrapper.de/_/29aq1/

datawrapper.de

"Please indicate whether you support or oppose free-trade agreements with other countries"

100

Ulrich Atz @ulrichatz.org · 24/07/2026

You can vote on the new euro money design: surveys.ecb.europa.eu/10b/banknote... 💶💶💶

000

Ulrich Atz @ulrichatz.org · 22/07/2026

Thanks a lot to Nathan Goldman for being a stellar host, and to my co-authors Tim Martens and Kasper Regenburg for putting up with my never-ending asks to do more. Working paper soon, we hope. Stay tuned!

000

Ulrich Atz @ulrichatz.org · 22/07/2026

The paper behind it: we analyze the full text of 19,000+ articles in top accounting, finance, and economics journals to measure how quickly accounting adopts new methods and research designs.

100

Ulrich Atz @ulrichatz.org · 22/07/2026

I'm giving a brownbag seminar on "Accounting Research in the Twenty-First Century" at NC State University today, and I have summarized the talk for everyone else with this handy graphic...

Boardroom suggestion meme, three panels. A boss tells the meeting: "We need to increase the credibility of accounting research." Three researchers suggest: "More fixed effects," "More matching," and "Careful design." In the last panel, the one who said "Careful design" has been thrown out of the office window.

100

Ulrich Atz @ulrichatz.org · 28/06/2026

This game was absolute chaos. You need to know that the goal also meant Algeria would face Spain instead of Switzerland next, and they knew it.

000

Ulrich Atz @ulrichatz.org · 17/06/2026

Congratulations to the R Core Team and the five named laureates; I’m happy for them and everyone who supports the amazing #rstats community. Bravo! 👏

000

Ulrich Atz @ulrichatz.org · 12/06/2026

If you reprice $TSLA and $SPCX at reasonable, even generous, growth-company multiples, you get a net worth around $350B. Is that a reasonable amount for one person to have? No. Does the exact number matter much in the grander scheme of problems? Also no.

000

Ulrich Atz @ulrichatz.org · 08/06/2026

And the new compact bar icons for #Positron, which I much appreciate. 🙌

030

Ulrich Atz @ulrichatz.org · 04/06/2026

Did you recently update your #rstats to 4.6? Is your matrix algebra slow? Not #BLAS ting? `sessionInfo()` shows which BLAS library R uses. Switching from reference BLAS to Apple Accelerate is easily a 50x speedup. Let me know if this worked for you. 🚀 uatz.org/blog/2026-06-04-r-blas-macos/ #rstats

uatz.org

The Lurking Dangers of the R Version Update – ulrich atz

After you update on macOS, switch R to Apple’s fast BLAS for a 50x speedup.

071

Ulrich Atz @ulrichatz.org · 04/06/2026

Very nice! This one next? 😅 bsky.app/profile/ulri...

020

Ulrich Atz @ulrichatz.org · 03/06/2026

I thought the original design was best until I saw this one. *after all ... why not? meme*

/Pride with Ally Open Source Flag-2048x2048.jpeg

020

Reposted by Ulrich Atz

collapse @rcollapse.bsky.social · 01/06/2026

The article on "collapse: Advanced and Fast Statistical Computing and Data Transformation in R" has now been published (open-access) in the Journal of Statistical Software: www.jstatsoft.org/article/view... #rstats #DataScience

jstatsoft.org

093

Ulrich Atz @ulrichatz.org · 01/06/2026

I would expect ”Green Planet” to show up under “plants”?

000

Ulrich Atz @ulrichatz.org · 29/05/2026

The 📦 `wrds` dev version displays the official product names for a more human-readable version of the table names. The package now comes with a #rstats #pkgdown site: statzhero.github.io/wrds/news/in... All that's missing is a hex sticker? Send ideas.

wrds
0.1.1.9000

Readme
Reference
News

Changelog
Source: NEWS.md
wrds (development version)

describe_table() now shows human-readable column labels and table descriptions from WRDS metadata.
list_subscriptions() now returns a tibble with schema and product columns instead of a character vector. Product names are mapped from the wrds_products dataset.
list_tables() now returns a tibble with table and description columns instead of a character vector. Descriptions come from WRDS metadata.

000

Ulrich Atz @ulrichatz.org · 28/05/2026

Nice idea! What are the actual criteria that make a problem interesting for the podcast?

100

Reposted by Ulrich Atz

Pierre-Simon Laplace @learnbayesstats.bsky.social · 27/05/2026

🚨 @rmcelreath.bsky.social @statmodeling.bsky.social & @avehtari.bsky.social are coming on the show mid-June to discuss their new book, Bayesian Workflow! ONE listener gets to bring a real Bayesian problem onto the recording and have the three of them work through it live. Here's how to enter 🧵

45114

Ulrich Atz @ulrichatz.org · 20/05/2026

New version of the #typst template, now with more epigraph. typst.app/universe/pac...

econ-working-paper 0.4.0 preview screenshot

000

Ulrich Atz @ulrichatz.org · 16/05/2026

What’s going on in this chart? Accounting and finance sit in a strange upper-right quadrant: jobs are available, but fields are highly exposed to AI. Are firms just slow to adapt, or do entry roles fulfill a different purpose vs. IT? Gift link below: www.economist.com/finance-and-... #AI #jobs

Chart: United States, recent university graduates in full-time employment

000

Ulrich Atz @ulrichatz.org · 15/05/2026

Also dann: lassen sich Kernkraftwerke für den Fall einer Katastrophe versichern?

010

Ulrich Atz @ulrichatz.org · 15/05/2026

An interesting corollary, in my view, is figuring out when this approach does NOT work easily. For instance, I imagine deploying a survey instrument, running a structural model for days, adding personal/niche anecdotes, or deviating from the template are examples where the factory would struggle.

010

Ulrich Atz @ulrichatz.org · 15/05/2026

Also, minor: the introductory paragraph was overloaded with citations in a way that I found inelegant and also not appropriate.

110

Ulrich Atz @ulrichatz.org · 15/05/2026

Aha, artisianal! 😅 My diagnosis: is, one, that there was not enough data cleaning and harmonization beyond the primary data source and two, given the limited descriptive nature, the factory got lost in details. (Like the classic “I am helpful assistant and _will_ find many ‘interesting‘ things.”)

110

Ulrich Atz @ulrichatz.org · 14/05/2026

Nevertheless, it is fascinating! Correct me if I'm wrong but I'm sort of missing a description of why the paper factory looks exactly the way it does? E.g., which parts were "bootstrapped" and how much was a motivated design of the orchestrator?

110

Ulrich Atz @ulrichatz.org · 14/05/2026

FWIW I have tried it on a descriptive question as a test, and it underwhelmed on the data cleaning part. @natewilmers.bsky.social are you planning to release the SLURM version as well?

110

Ulrich Atz @ulrichatz.org · 12/05/2026

oof what happened to ebay? If you pay $100 for something shipped from New York, the seller might only keep around two thirds, so $66, after sales tax, shipping, and fees. What a haircut, feels like #ebay is cooked

000

Reposted by Ulrich Atz

Spookysaurus @cheekyian.bsky.social · 08/05/2026

Happy 100, big man

A picture of David Attenborough, in a slight B Boy pose, doctored so he is wearing a cap and carrying a can, in a slightly. The title is David Atten-bruh, and it's captioned "Planet's fucked, fam'

51899476

Reposted by Ulrich Atz

Natural History Museum, London @nhm-london.bsky.social · 08/05/2026

Join us in wishing a massive happy 100th birthday to the man who has inspired millions of people around the world to care about nature and look after our planet - the incredible Sir David Attenborough! 🎉

David Attenborough standing outside the Museum wearing a blue jacket and blue and white patterned tie. He is laughing in the image, which features text saying "Happy 100th birthday Sir David Attenborough!"

1730229

Reposted by Ulrich Atz

Ben Zipperer @benzipperer.org · 05/05/2026

for the targets #rstats fans: highly recommend some targets-specific keybindings for Positron/VSCode

${ "key": "Ctrl+T M", "command": "workbench.action.executeCode.console", "args": { "langId": "r", "code": "targets::tar_make()", "focus": false } }, { "key": "Ctrl+T T", "command": "workbench.action.executeCode.console", "args": { "langId": "r", "code": "tflow::rs_make_target_at_cursor()", "focus": false } }, { "key": "Ctrl+T R", "command": "workbench.action.executeCode.console", "args": { "langId": "r", "code": "targets::rstudio_addin_tar_read()", "focus": false } }, { "key": "Ctrl+T L", "command": "workbench.action.executeCode.console", "args": { "langId": "r", "code": "targets::rstudio_addin_tar_load()", "focus": false } }$

1144

Ulrich Atz @ulrichatz.org · 30/04/2026

Here are three agent skills for Typst typesetting. If you haven't tried Typst yet: it's the modern LaTeX replacement that compiles in milliseconds. #typst #claudecode #codex github.com/statzhero/typst-skill github.com/statzhero/typst-cetz-skill github.com/statzhero/typst-touying-skill

github.com

GitHub - statzhero/typst-skill: An LLM skill for modern Typst typesetting (0.14+)

An LLM skill for modern Typst typesetting (0.14+). Contribute to statzhero/typst-skill development by creating an account on GitHub.

020

Ulrich Atz @ulrichatz.org · 24/04/2026

tidy-r now covers the dplyr 1.1.0 change where if_else() and case_when() accept plain NA. Personally the `NA_character_`, `NA_real_`, etc. always bother me somewhat...

000

Ulrich Atz @ulrichatz.org · 20/04/2026

One takeaway from the paper is that disclosure is necessary but not sufficient: Franchise Disclosure Documents make these restraints visible, but they do not undo switching costs,hold-up, or post-term noncompetes. The FTC's Franchise Rule emphasizes disclosure over substantive regulation.

031

Ulrich Atz @ulrichatz.org · 17/04/2026

Very nice design with the “+” operator. I did my homebrew with function arguments. Does prefab allow you to customize? eg via templates?

110

Ulrich Atz @ulrichatz.org · 17/04/2026

@paulgp.com where do the 70-90% identification rates from the dashboard website come from? I can't find them in the paper (or I'm reading it wrong). paulgp.com/econlit-pipe...

Identification Strategies

This chart shows the frequency of papers matching the following search query:

"causal identification" OR "causally identified" OR "identification strategy" OR "identification approach" OR "identification assumption" OR "identifying assumption" OR "identifying variation" OR "empirical identification" OR "over identified" OR "over-identified" OR "underidentified" OR "under-identified" OR "identification properties" OR "identification test" OR "identification problem" OR "identification issue" OR "problem with identification" OR "problems with identification" OR "issue with identification" OR "issues with identification" OR "problem identifying" OR "problems identifying" OR "issue identifying" OR "issues identifying" OR "threat to identification" OR "threats to identification" OR "threat for identification" OR "threats for identification" OR "over identifying" OR "over-identifying" OR "under identifying" OR "underidentifying" OR "partial identification" OR "partially identified" OR "non-parametric identification" OR "nonparametric identification" OR "non parametric identification" OR "non-parametrically identified" OR "nonparametrically identified" OR "non parametrically identified" OR "identification condition" OR "identifying condition" OR "condition for identification" OR "conditions for identification" OR "condition for identifying" OR "conditions for identifying" OR "point identification" OR "pointidentification" OR "point identified" OR "point-identified" OR "point identifying" OR "point-identifying" OR "set identification" OR "set-identification" OR "set identified" OR "setidentified" OR "set identifying" OR "set-identifying" OR "identification analysis" OR "weak identification" OR "identification result" OR "identification argument" OR "identification framework" OR "identification scheme"

100