Sign in

Malcolm 朝精 Barrett

@malcolmbarrett.malco.io
5.3K followers 834 following 772 posts

Ph.D., epidemiology. research software engineer @ Stanford Health Policy. living in Ann Arbor. open-source data science. causal inference. doing poems on aircrafts. approximately Bayesian. formerly Posit, Apple, AmeriCorps. 心を燃やせ。sic semper tyrannis.

PostsRepliesMedia
Malcolm 朝精 Barrett @malcolmbarrett.malco.io · 6h
It’s true that determinism’s time has come
110
Reposted by Malcolm 朝精 Barrett
Libby Heeren @libbyheeren.bsky.social · 02/10/2026
Nic ( @niccrane.bsky.social) taught us all about FREAKING PARQUET FILES on the #DSLab, which are wayyyyy cooler than I realized. Parquet files: Preserve data types Are fast & small Are column-oriented so you can read just 1 col if wanted! & Positron displays'em out-the-box! ▶️ youtu.be/2mQ4kc9J9uQ
youtu.be
What is a Parquet file anyway? Why is Parquet good? | Nic Crane | Data Science Lab
YouTube video by Posit PBC
56110
Reposted by Malcolm 朝精 Barrett
Dax Kellie @daxkellie.bsky.social · 02/10/2026
This thread is so great! 🤩 To avoid annoying resizing, figured I'd mention an alternative to {ggview}: {showtext} can resize all text to a specific dpi Text is the part that I find I have to fiddle with most when saving #ggplot2 plots, so this usually works for me 😀 #rstats #dataviz
library(showtext)
showtext_opts(dpi = 300)
ggsave(
  plot = my_plot,
  file = "file_name.png",
  width = 6,
  height = 4,
  dpi = 300
)
2316
Malcolm 朝精 Barrett @malcolmbarrett.malco.io · 15h
And so is stochasticity!
130
Reposted by Malcolm 朝精 Barrett
Jeremy Berg @jeremymberg.bsky.social · 01/10/2026
From a discussion of updating a department website...
A photo with baby Yoda and adult Yoda with the caption "Could it be time for a new photo?"
934670
Reposted by Malcolm 朝精 Barrett
Ariel M. (she/her) @byarielm.fyi · 30/09/2026
If you’ve seen me trying not to reveal what I’ve been working on the last few months, well, here it is 🎉 Our first big offering - aside from Atmosphere account hosting - is personal websites for researchers! Drop your must-haves, preferences, ideas, or questions about personal/prof sites below 🧪🧬💻
810424
Malcolm 朝精 Barrett @malcolmbarrett.malco.io · 30/09/2026
I’ve seen some decent looking typst templates for them
000
Reposted by Malcolm 朝精 Barrett
Per Engzell @pengzell.bsky.social · 30/09/2026
Two people at a cocktail party. One says: "I'm a fiction writer in the causal inference genre."
04413
Reposted by Malcolm 朝精 Barrett
Andreas Brandmaier @brandmaier.bsky.social · 28/09/2026
Ever looked at your graph and thought: this needs a more dramatic exit strategy? Make your R base plots self-destruct with style. If you must, download here: github.com/brandmaier/d... #RStats #DataVisualization #RProgramming #DataScience #OpenSource #AcademicHumor #noAI
47622
Malcolm 朝精 Barrett @malcolmbarrett.malco.io · 29/09/2026
meanwhile Phase 1 Phase 2 Phase 3 LLMs ? AGI
062
Malcolm 朝精 Barrett @malcolmbarrett.malco.io · 27/09/2026
Now make them orange juice and milk
110
Reposted by Malcolm 朝精 Barrett
Jacob Kresovich @jkresovich.bsky.social · 24/09/2026
I get that AI in academia/research is a little controversial, but personally, it has greatly improved my ability to produce low-impact science.
051
Reposted by Malcolm 朝精 Barrett
weeder @weeder.bsky.social · 24/09/2026
God, grant me the courage to change the things I can and incredible new powers to change the things I can't.
63106932532
Reposted by Malcolm 朝精 Barrett
Emil Hvitfeldt @emilhvitfeldt.bsky.social · 23/09/2026
I think I have cracked one of the final things that kept people in keynote over quarto revealjs slides. Introducing ✨️ quarto-revealjs-magic-move ✨️ emilhvitfeldt.com/post/quarto-... #quarto #slidecrafting
25510
Reposted by Malcolm 朝精 Barrett
Brady West @bradytwest.bsky.social · 22/09/2026
The University of Michigan National Survey Panel (M-Panel) is now openly accepting questionnaire content from researchers. Would you like to ask questions to a national probability-based panel of adults in the U.S., with an oversample from Michigan? Tap below! mpanel.isr.umich.edu/for-research...
mpanel.isr.umich.edu
For Researchers | M-Panel: National Survey PanelFor Researchers – M-Panel: National Survey Panel
12118
Reposted by Malcolm 朝精 Barrett
Mikhail Popov @mpopov.com · 21/09/2026
My team (Product Analytics) is hiring a Sr. Data Scientist to help guide Wikipedia’s anti-abuse and security strategy
job-boards.greenhouse.io
Senior Data Scientist, Safety & Security
Remote
161
Reposted by Malcolm 朝精 Barrett
Catherine Rampell @crampell.bsky.social · 20/09/2026
It really is striking how many authoritarian states have ended up in this exact same place, where the leader seizes control of scientific inquiry. Lysenkoism in the Soviet Union, Aryan Physics in the Third Reich, Great Leap Forward (also related to Lysenko) in China.
8527281074
Malcolm 朝精 Barrett @malcolmbarrett.malco.io · 20/09/2026
The perfect notebook would fix me
040
Malcolm 朝精 Barrett @malcolmbarrett.malco.io · 20/09/2026
Seems like he hates you, too tbh
011
Reposted by Malcolm 朝精 Barrett
The Onion @theonion.com · 18/09/2026
Trans Teen Hatches Nefarious Plot To Undergo Years Of Medical Treatments And Counseling To Win At Swimming theonion.com/trans-teen-hatches-nef…
Trans Teen Hatches Nefarious Plot To Undergo Years Of Medical Treatments And Counseling To Win At Swimming
69129182821
Malcolm 朝精 Barrett @malcolmbarrett.malco.io · 17/09/2026
They got better plots
140
Malcolm 朝精 Barrett @malcolmbarrett.malco.io · 17/09/2026
This is giving me a talk idea bout constraining agentic LLMs to improve correctness
040
Reposted by Malcolm 朝精 Barrett
Nicolas Hénin @nicolashenin.net · 17/09/2026
Nice campaign by the Norwegian Consumer Council about the declining lifespan of products. Enjoy!
180140516471
Malcolm 朝精 Barrett @malcolmbarrett.malco.io · 15/09/2026
dowistrepla lol
010
Malcolm 朝精 Barrett @malcolmbarrett.malco.io · 15/09/2026
Yes To All
010
Reposted by Malcolm 朝精 Barrett
caroline cash @cashbrowns.bsky.social · 15/09/2026
Nancy 9/8/26
Nancy’s at school…. Her teacher asks her a math problem: if Nancy had three cupcakes, and Sluggo asked her for one, how many would she have left? Nancy’s reply is “three.”
4379591424
Reposted by Malcolm 朝精 Barrett
Malcolm 朝精 Barrett @malcolmbarrett.malco.io · 14/09/2026
Once again 🕯️summoning the Logistic Regression Is Not Fucked from the Netherworld 🕯️ web.archive.org/web/20190806...
web.archive.org
Logistic regression is not fucked
Summary: Arguments against the use of logistic regression due to problems with “unobserved heterogeneity” proceed from two distinct sets of premises. The first argument points out that …
2171
Malcolm 朝精 Barrett @malcolmbarrett.malco.io · 14/09/2026
Once again 🕯️summoning the Logistic Regression Is Not Fucked from the Netherworld 🕯️ web.archive.org/web/20190806...
web.archive.org
Logistic regression is not fucked
Summary: Arguments against the use of logistic regression due to problems with “unobserved heterogeneity” proceed from two distinct sets of premises. The first argument points out that …
2171
Reposted by Malcolm 朝精 Barrett
Gabor Csardi @gaborcsardi.org · 14/09/2026
New beta of #Rstats rig: LOTS of EXCITING stuff: - new: web site: rig.r-lib.org - new: user-mode R installation: rig.r-lib.org/admin-vs-use... - new: manage R packages (CRAN + git): rig.r-lib.org/pkg-guide.html - new: manage R projects and virtual environments: rig.r-lib.org/proj-guide.h...
rig.r-lib.org
rig – the R Installation Manager – rig
34919
Reposted by Malcolm 朝精 Barrett
Mike McQuaid @mikemcquaid.com · 13/09/2026
Today, I’m proud to announce Homebrew 7.0.0. The most significant changes since 6.0.0 are faster installations, stronger sandboxing, native macOS app, vulnerability checks, advisory database, end of macOS 10.15 support and Intel Macs to Tier 3.
brew.sh
7.0.0
Today, I’m proud to announce Homebrew 7.0.0. The most significant changes since 6.0.0 are faster installations and upgrades, stronger sandboxing, a native macOS app, built-in vulnerability checks and an advisory database, the end of macOS 10.15 support and Intel Macs moving to Tier 3.
523226
Malcolm 朝精 Barrett @malcolmbarrett.malco.io · 12/09/2026
It’s not my fault! (Sneak me a sticker and meet me in Indiana)
120
Malcolm 朝精 Barrett @malcolmbarrett.malco.io · 12/09/2026
First one I miss since 2017 has WHAT
121
Reposted by Malcolm 朝精 Barrett
Andrew Heiss @andrew.heiss.phd · 11/09/2026
I like this a lot
1305
Reposted by Malcolm 朝精 Barrett
Carlos Scheidegger @cscheid.net · 10/09/2026
quarto-hub is really coming together, y'all! I'm so excited to share some of this with you at conf next week. Check out our upcoming collaborative wysiwyg editor working on a fully-styled Quarto 2 page:
56514
Reposted by Malcolm 朝精 Barrett
Julia M. Rohrer @dingdingpeng.the100.ci · 03/09/2026
Very few people know that this is how it all began @celestialsem.bsky.social
Sewall Wright's guinea pig path diagram, but the guinea pigs are wearing space helmets and there's a space background
27414
Reposted by Malcolm 朝精 Barrett
Josiah @josiah.rs · 10/09/2026
I'm starting go all in on S7. Just waiting for new development version to be released 🫣 {s7x} has been an absolute cheat code for my package development lately. s7x.josiah.rs #rstats
s7x.josiah.rs
Extensions to the S7 Object System
Provides utilities that extend the S7 object system: enumerated types with a fixed set of variants, scalar-constrained properties, range-bound numeric properties, and property-level unions and interse...
2203
Malcolm 朝精 Barrett @malcolmbarrett.malco.io · 10/09/2026
Listening to this Mingus classic on the importance of version control
Album artwork for Charles Mingus’s Mingus Ah Um, featuring an abstract geometric painting in muted red, blue, yellow, pink, and cream. Below the cover, the track “Better Git It in Your Soul” and artist Charles Mingus are displayed.
0315
Reposted by Malcolm 朝精 Barrett
public health guy @publichealthguy1.bsky.social · 10/09/2026
📈📈📈
Gas Station Price Sign Using Scientific Notation

the ONION
173291609
Reposted by Malcolm 朝精 Barrett
Emily Riederer @emilyriederer.bsky.social · 10/09/2026
We could save years of education across data science and data engineering by just printing t-shirts that say "know what makes a row"
1408
Malcolm 朝精 Barrett @malcolmbarrett.malco.io · 10/09/2026
we have this little problem where we are always selecting on the data that we *have*
0205
Reposted by Malcolm 朝精 Barrett
Mike X. Nichols @mikexnichols.bsky.social · 08/09/2026
Thinking about this exchange today
A YouTube video for the Water Temple in The Legend of Zelda: Ocarina of Time. The comments section has an exchange as detailed below. “@TheMrdangles16 • 11y ago: 
i hate the water temple... too confusing :(“ “@KeithHalsell • 2y ago: 
it's been 9 years, i hope you are doing great in life now.” “@TheMrdangles16 • 2y ago: it couldn't be going worse Keith.”
34181034489
Malcolm 朝精 Barrett @malcolmbarrett.malco.io · 08/09/2026
One reason I know LLMs won't take over academia is that the simply cannot feel so annoyed by not knowing something, the true essence of an academic
011
Malcolm 朝精 Barrett @malcolmbarrett.malco.io · 08/09/2026
BUT I DON'T KNOW, AND IT IS ANNOYING
100
Malcolm 朝精 Barrett @malcolmbarrett.malco.io · 08/09/2026
*Today*, I think there is a lot you can do to make these things generate output that is more rigorous and reproducible than a lot of human-generated stuff out there. *Tomorrow*, I have no idea if we will need to do anything at all. I *think* it's an 80% problem with a lonnnngggg 20% tail.
100
Malcolm 朝精 Barrett @malcolmbarrett.malco.io · 08/09/2026
I *think* it will remain true for some time that having foundational knowledge in statistics and coding will allow you to use these tools for those purposes more effectively and quickly. But I don't know!
100
Malcolm 朝精 Barrett @malcolmbarrett.malco.io · 08/09/2026
What's really annoying about all this is that, RIGHT NOW, I think I have a good grasp of what tools and skills scientists should have to use these well, and I have no idea how long that will be true.
100
Malcolm 朝精 Barrett @malcolmbarrett.malco.io · 08/09/2026
I don't know enough about current models to have an answer to that, but I DO think it underlies how useful having a principled tool to constrain LLMs is. In this case, that's Lean. But in experimenting a lot lately, it seems that's true for using the current generation of models in general
071
Reposted by Malcolm 朝精 Barrett
Adam Austin @ataustin.bsky.social · 07/09/2026
It me
Ben Affleck smoking meme but instead of a cigarette it's .Last.vale
053
Reposted by Malcolm 朝精 Barrett
Bob Wilson @rwilson4.bsky.social · 07/09/2026
New Paper Alert! I have posted a pre-print to the arXiv: Randomization Inference for Matched Pairs with Binary Outcomes (link in comments). If you're doing propensity score matching, optimal matching, etc, and you have binary outcomes, give it a look!
121
Reposted by Malcolm 朝精 Barrett
Lisa DeBruine @debruine.bsky.social · 07/09/2026
I love the concrete examples of how to avoid anthropomorphism in this article: artificial intelligence → probabilistic automation image recognition → image labeling speech recognition → automatic transcription shows bias → reflects bias hallucination → undesirable output goal → success condition
39237