RIP Bonnie Tyler

Also, data scientist here, let me echo another theme: Get your SQL down. Not so much for the language itself but the way it reflects an amazingly useful way to think about structuring & manipulating data. You won't regret the way it rearranges your brain, even if you don't end up writing much of it.

Once a Catholic, always a Catholic (unless the church has sent you a strongly-worded letter), which means I get to really enjoy the bizarre responses to papal authority in certain subsets of the community.

ultra-trad Catholics when the Holy Father says something they don't like: "He's not speaking ex cathedra here so I get to insult and deride the Pope"

Ultra-Traditionalist Catholics: "We are not cafeteria Catholics, we do not pick and choose the doctrines of the Church" Also Ultra-Traditionalist Catholics: "I don't vibe with this Ecumenical Council"

I specifically used this image in a professional talk once and I learnt people haven’t seen then best movie ever.

Follow that thread…here.

I just started this and I need y’all to know that it’s a) fabulous and b) you will be so fucking sad by the end of the first section. I’m reading it slowly so I can savour it. Because it’s unique(Also this makes three RADICALLY different books all of which are my best available example of its type)

Happy Pride to everyone who is constantly typing json.dumps(json.loads(…))

Happy pride y'all music.apple.com/us/album/im-...

good morning

good morning 🌯

Wake up everybody. New BIG paper on priors just dropped.

I mean. The Queen.

I did not see this because Bianca Leigh was doing a solo show. But fucking work.

She brought the TONYYYYYYYYYYY

The untold story of Mrs. Lovett... 🥧 THE BUTCHER'S DAUGHTER by Corinne Leigh Clark & David Demchuk is out in paperback! Discover how a butcher’s daughter from Whitechapel grew up to become the baker who stuffed her pies with Sweeney Todd’s victims. 🩸🪓 https://tinyurl.com/4p7u6esf

Anyway. I think if you’re not acknowledging that aspect of things you should be actively acknowledging that to yourself. And if that’s fine with you that’s cool. Lessening participation is an option. Because supporting high output people who make new contributors never want to again is a choice.

A prize to men who are, at least in part, shorthand for “cunts who make people never want to product open source statistical software ever again? Revolutionary.

Come on, Gland.

Crash (1996, d. David Cronenberg)

This was my twitter header for a long time. And then someone was like “hey you’re on the job makes you should change it”

Also surprise EoE draft was fun, even if I am bad at magic.

My big thought for this evening is “I like that Reeves Gabrels is in The Cure”. If you want to hear peak guitar music, those mid-90s tours where he was Bowie’s live guitarist off Outside etc is it.

Pick a Bond villain from the photos you already have on your phone.

Such a fun pile of absolute twaddle (again, extremely complimentary). Also the supporting cast is just phenomenal

I don't care that an LLM fixed your conjecture or whatever, Don. LaTeX is still a cludge

my gut is that these tools cannot possibly work because the whole point of LLMs is to produce texts that are statistically indistinguishable from their training data

this review of PINK NARCISSUS put me on the floor

If you program, it pays to understand the program that programs programs

A long time ago I even wrote a post about how applied statisticians and people doing statistics in science are programmers and how we should accept that. It also had far and away the most unhinged conclusion I ever wrote (not screenshotted) statmodeling.stat.columbia.edu/2019/11/13/w...

I mean I didn’t suggest either of these insane options, but the first one could be a way to teach applied Bayesian students how to deal with the complexities of MCMC as a tool. (I’d prefer to give them “draws” and teach them to analyse them because it’s more focussed). As for the second question…

This is a subtweet because it’s 2am and honestly no one cares what I think, but a real challenge with teaching a class that introduces a new concept using X is making sure students still know how to do it without package X.

Stan embedded Laplace was a long project and happy that it's finally released. We are not competing with INLA and TMB software, as they are orders of magnitude faster and scale better with data size for models that you can implement with them 1/

What about blokes?

I switched from late thirties to “late” “thirties” recently and girl my life is now begging men to drive their elbows into my shoulder blades. Sex is nice but have you ever thought about being able to sleep?

Sometimes a book is too Toronto.

Decompressing from being aggressively reminded of all these things I am still extremely good at that I just don’t do anymore. And I think I expect that to make me feel some kind of way but really it’s just really fun to be able to dip in for a day or two and then leave.

Every time someone turns 60 I give a talk. They get stranger and stranger the further I get from the material

Andrew’s 60-ish birthday workshop has been nice.

Miles Moran (Oregon State University), Rob Trangucci (Oregon State University), Lisa Madsen (Oregon State University) Bayesian Inference of Mixing and Transmission Heterogeneity in Stratified Disease Surveillance Models https://arxiv.org/abs/2605.29081

But it is certainly possible that higher forms of consciousness are an intricate web of md files.

You make fun of horoscope girlies ⭐️ but then you let a pile of linear algebra tell you what to do….

Tarra Conner Jones, people. Tarra. Conner. Jones.

Also I’ve not read the entire open letter but I’m glad the church is engaging with this from this angle. It’s nice for them to point their intellectual and ethical rigour at the most looming thing for a change.

The New York Times has broken new ground on how to refer to a papal encyclical

Me getting back into the reading academic papers