@imdef.bsky.social
Computational scientist, swedish, don't like to make it too easy for social media companies to collect data. Some statistics/decision theory thoughts at https://intemittdefault.wordpress.com/
Does social media harm everyone?
No. But it harms *most* adolescents.
However, not all platforms are harmful.
An analysis of 44,211 diaries from 479 adolescents over 100 days finds that 60% of adolescents experienced small, negative effects of social media
link.springer.com/article/10.1...
Katten är ute ur säcken - det här skrev jag om här för några veckor sen.
Bra att det nu lyfts i media, för det är så otroligt allvarligt.
(Jag sitter med i den rekryteringsgrupp vid Uppsala universitet som har detta ärende)
www.sydsvenskan.se/lund/uppsala...
It would be interesting to also vary the consequences/costs of the event occuring/not occuring in the poll. For example, if you judge that there's a 1% probability that your plane is gonna crash, you don't call it "virtually certain" that it won't.
03.01.2026 17:44 — 👍 1 🔁 0 💬 1 📌 0
Will you incorporate LLMs and AI prompting into the course in the future? No. Why won’t you incorporate LLMs and AI prompting into the course? These tools are useful for coding (see this for my personal take on this). However, they’re only useful if you know what you’re doing first. If you skip the learning-the-process-of-writing-code step and just copy/paste output from ChatGPT, you will not learn. You cannot learn. You cannot improve. You will not understand the code.
In that post, it warns that you cannot use it as a beginner: …to use Databot effectively and safely, you still need the skills of a data scientist: background and domain knowledge, data analysis expertise, and coding ability. There is no LLM-based shortcut to those skills. You cannot LLM your way into domain knowledge, data analysis expertise, or coding ability. The only way to gain domain knowledge, data analysis expertise, and coding ability is to struggle. To get errors. To google those errors. To look over the documentation. To copy/paste your own code and adapt it for different purposes. To explore messy datasets. To struggle to clean those datasets. To spend an hour looking for a missing comma. This isn’t a form of programming hazing, like “I had to walk to school uphill both ways in the snow and now you must too.” It’s the actual process of learning and growing and developing and improving. You’ve gotta struggle.
This Tumblr post puts it well (it’s about art specifically, but it applies to coding and data analysis too): Contrary to popular belief the biggest beginner’s roadblock to art isn’t even technical skill it’s frustration tolerance, especially in the age of social media. It hurts and the frustration is endless but you must build the frustration tolerance equivalent to a roach’s capacity to survive a nuclear explosion. That’s how you build on the technical skill. Throw that “won’t even start because I’m afraid it won’t be perfect” shit out the window. Just do it. Just start. Good luck. (The original post has disappeared, but here’s a reblog.) It’s hard, but struggling is the only way to learn anything.
You might not enjoy code as much as Williams does (or I do), but there’s still value in maintaining codings skills as you improve and learn more. You don’t want your skills to atrophy. As I discuss here, when I do use LLMs for coding-related tasks, I purposely throw as much friction into the process as possible: To avoid falling into over-reliance on LLM-assisted code help, I add as much friction into my workflow as possible. I only use GitHub Copilot and Claude in the browser, not through the chat sidebar in Positron or Visual Studio Code. I treat the code it generates like random answers from StackOverflow or blog posts and generally rewrite it completely. I disable the inline LLM-based auto complete in text editors. For routine tasks like generating {roxygen2} documentation scaffolding for functions, I use the {chores} package, which requires a bunch of pointing and clicking to use. Even though I use Positron, I purposely do not use either Positron Assistant or Databot. I have them disabled. So in the end, for pedagogical reasons, I don’t foresee me incorporating LLMs into this class. I’m pedagogically opposed to it. I’m facing all sorts of external pressure to do it, but I’m resisting. You’ve got to learn first.
Some closing thoughts for my students this semester on LLMs and learning #rstats datavizf25.classes.andrewheiss.com/news/2025-12...
09.12.2025 20:17 — 👍 330 🔁 100 💬 13 📌 31Fun!
08.12.2025 11:12 — 👍 0 🔁 0 💬 0 📌 0
Diederik Stapel was a massive watershed moment in psychology.
However, he was -- and let's be slightly glib here -- some guy from The Netherlands who wrote social psychology papers.
The full accounting of the Eysenck case is approx, at minimum, TWO STAPELS.
retractionwatch.com/2025/12/03/n...
Det här var bra. www.sverigesradio.se/avsnitt/sa-b...
06.11.2025 18:41 — 👍 7 🔁 1 💬 0 📌 0
Riktig Dammsugare i grönt
bakverket dammsugare
Dammsugare av märket Hugin Top vac.
Utseendet på dessa äldre dammsugare har gett namn på bakverket som också kallas punschrulle.
Interesting paper that I'll have to spend slow mathy times with later, but had a real record-scratch-freeze-frame moment while skimming...
21.08.2025 08:12 — 👍 32 🔁 2 💬 2 📌 0
😯 I recently came across a sobering paper on here: among 16,649 hypothesis tests reported in political science, the median test has only about 10% power to detect the consensus effect size reported in the literature. Fewer than one in ten tests reach the conventional 80% power threshold.
13.08.2025 11:56 — 👍 60 🔁 14 💬 3 📌 2A logic, with syntactic rules for proofs, corresponding to Pearl's causal inference. Fun stuff.
philsci-archive.pitt.edu/26213/
Thoughtful post by @daisychristo.bsky.social on learning to write in an era of AI: substack.nomoremarking.com/p/what-is-th...
18.07.2025 10:33 — 👍 36 🔁 12 💬 1 📌 1With arithmetic, people understand why. Same with learning the letters the alphabet and then learning to spell out words. With AI, many struggle to understand, because they don't expect ability to do the task AI does for them to be a required ingredient in future learning.
18.07.2025 13:18 — 👍 1 🔁 0 💬 0 📌 0I think the major point is that many things you learn are supposed to be building blocks in future learning. E.g., the first years in school, students do endless exercises in basic arithmetic... 5+7=?, 13-6=?, ... Why not just let them use a calculator instead? >
18.07.2025 13:15 — 👍 1 🔁 0 💬 1 📌 0Becker et al. "Measuring the Impact of Early-2025 AI on
Experienced Open-Source Developer Productivity"
metr.org/Early_2025_A...
A glimpse of the AI-assisted future
arxiv.org/abs/2506.08872
I'm not so optimistic, because it requires a great deal of linguistic proficiency and discipline to check that the thoughts were actually expressed correctly. I think what will happen is that even the technical literature becomes filled with AI slop.
17.05.2025 07:54 — 👍 0 🔁 0 💬 1 📌 0“Lots of good statistical methods make use of two models…
Perhaps, rather than trying to systematize all statistical learning into a single inferential framework, we would be better off embracing our twoishness” statmodeling.stat.columbia.edu/2011/06/20/t... via @seabbs.bsky.social
Another one, a bit less concrete, in the same genre
arxiv.org/abs/1708.09032
The most concrete proposal I've ever seen for reasoning about logical uncertainty
arxiv.org/abs/1609.03543
1. LLM-generated code tries to run code from online software packages. Which is normal but
2. The packages don’t exist. Which would normally cause an error but
3. Nefarious people have made malware under the package names that LLMs make up most often. So
4. Now the LLM code points to malware.
today we will all read imbens 2021 on statistical significance and p values, which is a strong contender for having the best opening paragraph of any stats paper
pubs.aeaweb.org/doi/pdf/10.1...
OTOH, the Blue Bus Company will be under-penalized if they are never sanctioned when a pedastrian is hit and can't remember the color of the bus.
IMO, legal ideas like, e.g., the presumption of innocence are a poor fit for probabilistic reasoning. They are closer to some kind of default logic.
This is a fundamental misunderstanding of how to use such services. To expect them to correctly cite sources is to use them incorrectly. The output is of varying quality and accuracy. If one is not prepared detect and correct factual errors, then simply don't use AI services.
15.03.2025 10:06 — 👍 0 🔁 0 💬 0 📌 0
Fun list, wonderfully simply presented
www.youtube.com/watch?v=1vJs...
These researchers designed a neat macroscopic object to mimic the spinorial behavior of a spin-1/2 fermion, e.g. an electron. The object is a football made up of pentagons and hexagons, the color of which are determined by the spin up and down components, respectively.
pubs.aip.org/aapt/ajp/art...
Top notch introduction to a scientific paper!
14.01.2025 14:38 — 👍 0 🔁 0 💬 0 📌 0
