iiinova — where your digital self keeps evolving

Reposted by Eric Pedersen

David Ho @davidho.bsky.social · 03/10/2026

The internet is so weird sometimes.

Eric Pedersen @ericjpedersen.bsky.social · 15/09/2026

Canadian negotiators will be pushing hard for a moratorium on Europeans calling anything "poutine" until they have been properly trained in how to make a proper cheese curd

Banner hanging in Leipzig train station advertising "Kabadische Poutine" for 7.20€, with a picture of fries covered in what looks like salsa and salad dressing

070

Reposted by Eric Pedersen

Vincent Arel-Bundock @vincentab.bsky.social · 04/09/2026

Big news! 🎉 𝚖𝚊𝚛𝚐𝚒𝚗𝚊𝚕𝚎𝚏𝚏𝚎𝚌𝚝𝚜 1.0.0 for #Rstats is out. It’s a big number and it feels like a big step. I wrote a blog on the challenges of interpreting statistical models, SPEED, cool new features, the future, and a 5 year package development and writing odyssey. arelbundock.com/posts/margin...

1642299

Eric Pedersen @ericjpedersen.bsky.social · 23/08/2026

"Hark a Vagrant" might be the best possible introduction to Canadian humour

191

Eric Pedersen @ericjpedersen.bsky.social · 23/08/2026

:) this made my night (the joke, not the website)

010

Eric Pedersen @ericjpedersen.bsky.social · 23/08/2026

We have all of Murder She Wrote on DVD here :). Although be warned: Farthing is maybe better described as an uncosy mystery... It's framed like one and starts out like a typical take on Agatha Christie, but you definitely won't finish the book feeling cosy about the world after reading it

010

Eric Pedersen @ericjpedersen.bsky.social · 23/08/2026

Like cosy murder mysteries? Like alt history? Like dark examinations of how the path to fascism is easily hidden beneath middle class gentile politeness? Read Farthing by @bluejo.eurosky.social www.jowaltonbooks.com/books/the-sm...

jowaltonbooks.com

Farthing | Jo Walton -- Science Fiction and Fantasy Author

2186

Reposted by Eric Pedersen

Michael Samonek @samonek.bsky.social · 22/08/2026

A huge part of my interest in "indie" Animation comes from being exposed to weird-ass projects funded by the National Film Board.

2583

Eric Pedersen @ericjpedersen.bsky.social · 23/08/2026

I hold a grudge against downtown Ottawa after trying to find a single restaurant open for dinner at 6 pm

111

Eric Pedersen @ericjpedersen.bsky.social · 23/08/2026

Broken Social Scene: pretty much everything they have ever done youtu.be/wLaDksDOcE4?... youtu.be/qeMOtslWIws?...

youtu.be

Broken Social Scene - Anthems For A Seventeen Year Old Girl (Official Video)

YouTube video by Broken Social Scene

010

Eric Pedersen @ericjpedersen.bsky.social · 23/08/2026

"One Great City": Best love song for a city from a punk rocker who became a folk poet. I fall in love again with the lyrics all over again every time I hear it youtu.be/YdxzNy-apdY?...

youtu.be

One Great City!

YouTube video by The Weakerthans - Topic

141

Eric Pedersen @ericjpedersen.bsky.social · 23/08/2026

I can see why you'd think that living in Ottawa. We need to get you to Montreal 😁

110

Reposted by Eric Pedersen

Kent Lind @camasonian.bsky.social · 17/08/2026

We have data on this. The blue is all the non-ag human use:

18865171

Eric Pedersen @ericjpedersen.bsky.social · 06/08/2026

I have (somehow) reached a new low point in the journal spam I'm getting. I just got an invite from the "Online Journal of Conference Proceedings". They're on the brink of releasing a "highly anticipated issue", and just need my submission by August 10th!

130

Eric Pedersen @ericjpedersen.bsky.social · 05/08/2026

In brief: "probability of observing your data" isn't a well-defined statement. For example: the probability of flipping exactly 500 heads out of 1000 from a fair coin is ~2%, but that would not be a useful p-value. Likewise, the probability of observing any specific values from a Normal dist is 0

040

Reposted by Eric Pedersen

Danielle Navarro @djnavarro.net · 18/07/2026

And so after a very brief hiatus, version 0.7.0 of "learning statistics with R" exists. Full rebuild in quarto, stylistic fixes, "epilogues from 2026" to comment on how the world changed since original publication, and as an added bonus, no longer misgenders the author learningstatisticswithr.com

learningstatisticswithr.com

Learning Statistics with R

217146

Eric Pedersen @ericjpedersen.bsky.social · 19/07/2026

"we can't both keep playing 'paper'! I called dibs"

120

Reposted by Eric Pedersen

Ramin Nasibov @nasibov.me · 17/07/2026

I’m so happy for them

364474782

Reposted by Eric Pedersen

lauren @lauren.rotatingsandwiches.com · 08/07/2026

You just deleted 600 unread emails. Finally, inbox zero. Phew! You refresh the app. There are 600 more unread emails. There will always be 600 more emails. You close your laptop in frustration and look at your phone, only to find there are 600 more emails there too. You cannot escape 600 more emails

1947835

Eric Pedersen @ericjpedersen.bsky.social · 08/07/2026

I've heard of this most commonly in math, econ, chem, and engineering, but I am sure it happens in all fields

010

Eric Pedersen @ericjpedersen.bsky.social · 08/07/2026

Ah yes, secret goal #8: validating the instructor's priors about students or vengeance for being made to teach. "You all failing the class proves that I am smarter than my students, and that kids these days are not as talented or hard-working as my generation."

110

Eric Pedersen @ericjpedersen.bsky.social · 08/07/2026

I'd call this "cargo cult grading". I've seen cases in undergrad where the prof didn't really bother teaching, the students didn't really bother studying, but since there were grades at the end that followed a curve that roughly matched other classes, learning must have happened, right?

120

Eric Pedersen @ericjpedersen.bsky.social · 08/07/2026

Apologies for the ranting :) I've been thinking about these issues a lot this last year when thinking about how to teach post-sabbatical, and I'm taking over as our undergraduate program director this year so it's even more front-and-center on my mind

110

Eric Pedersen @ericjpedersen.bsky.social · 08/07/2026

True. There's a secret 7th goal: "I assign grades because that is what I've been told to do, and I won't get in trouble if I do it"

120

Eric Pedersen @ericjpedersen.bsky.social · 08/07/2026

Good point (although this might be an example of 1&3, where the desired skill is "ability to thrive in bureaucracy". Given how universities are structured, this is implicitly one of the core skills we want students to have at the end of it

010

Eric Pedersen @ericjpedersen.bsky.social · 08/07/2026

Yep, because they are at odds with each other. 1&2 are almost mutually exclusive with #3&4; a "strong" student could do very badly in a poorly taught class and a "weak" student can thrive in a well-taught one Ironically, an instructor can get in trouble if *too many* students learn the material

120

Eric Pedersen @ericjpedersen.bsky.social · 08/07/2026

LLMs have shattered this détente because they allow some students to avoid 1-4 + 6 for take-home exams, but they still want the grades to do 3&5. And many students think they are genuinely learning when using LLMs to avoid 6, so they are surprised when they do poorly on in-person evaluations

110

Eric Pedersen @ericjpedersen.bsky.social · 08/07/2026

All six goals have coexisted in an uneasy peace because of varying priorities among stakeholders. Typically: instructors value 1/2/6 employers 3/4/ + bit of 1 school admin 2/3/5 students 3&5 We can't design assignments that do all 6 well simultaneously, but we pretended that wasn't the case

110

Eric Pedersen @ericjpedersen.bsky.social · 08/07/2026

Forgot to add: 6. incentivize students to try challenging problems / learn new ideas

120

Eric Pedersen @ericjpedersen.bsky.social · 08/07/2026

I'm planning on shifting more grade weights to in-person evaluation. I might also switch to giving participation-grades for take-home assignments. I plan to tell students that these are meant as training on how to solve the kinds of problems I expect them to be able to do.

030

Eric Pedersen @ericjpedersen.bsky.social · 08/07/2026

I'm not convinced that it's just a tragedy of the commons; there's always been ways to cheat that were hard to prove, but it was relatively rare to see pre-LLMs I think many people don't see using LLMs as cheating, but justify their use as study tools or as shortcuts to "help" with hard problems

140

Eric Pedersen @ericjpedersen.bsky.social · 08/07/2026

I think it also breaks applied problem solving. Working through novel problems is proven to be one of the best ways (maybe the only real way) to learn a subject (rather than just memorize), but it's hard to do. Grades were the only tool we had to motivate trying problems, but now that's broken

020

Eric Pedersen @ericjpedersen.bsky.social · 08/07/2026

This case highlights a core issue with grades that LLMs have brought to the fore: we are using grades to try to measure/do too many different things: 1. learning+readiness for later material 2. teaching effectiveness 3. Sorting by latent ability 4. Effort put into class 5. as a tool to open doors

481

Eric Pedersen @ericjpedersen.bsky.social · 08/07/2026

This isn't a good comparison; Google can't solve a novel problem if you don't have the skills to work through it. The closest previous analog was posting Q's to Stack Overflow or Chegg to have them answer you. The first was slow and unpredictable, the 2nd led to garbage answers that are easy to spot

120

Reposted by Eric Pedersen

Stephen Wild @stephenjwild.bsky.social · 08/07/2026

Today's ragebait for the professors who follow me. 3 points: 1) if I am a student, a take home exam is implicitly open-book. Using ChatGPT is fine and no different. 2) the professor changed the rules after the fact. The students should be pissed 1/2

11241

Eric Pedersen @ericjpedersen.bsky.social · 08/07/2026

I agree voiding the midterm is extreme and I wouldn't have gone that far, but it does make sense if you want a grade to measure learning. If you were doing a meta-analysis and found out that one study used fake data, isn't it justified in giving that study zero prior weight?

010

Eric Pedersen @ericjpedersen.bsky.social · 08/07/2026

Also, if his university is anything like mine, he probably hasn't received much if any guidance on how to adapt teaching to LLMs. As a prof, I feel a bit like a dinosaur watching the meteor approach; all of our best teaching tools are completely voided by LLMs, and it's only going to get worse

240

Eric Pedersen @ericjpedersen.bsky.social · 08/07/2026

d) He should have realized this was an issue w/ take-home exams, but I have a hard time blaming him, as I only updated my priors on the strength of LLMs when I tried some test exam questions on Gemini during my sabbatical. They now give A+ answers (whereas it was B- on good Q's 3-4 years ago)

120

Eric Pedersen @ericjpedersen.bsky.social · 08/07/2026

c') the prof's response was the best that was available. It is impossible to make a LLM-proof take home assignment now, so we can't use them to evaluate student learning. He didn't lower grades for the midterm, just switched the final eval. Generally this is possible when required and he gave notice

210

Eric Pedersen @ericjpedersen.bsky.social · 08/07/2026

b) I've seen a sea change in norms around cheating. During COVID pre-ChatGPT, with online exams, I only detected < 10% of students cheating with online tools. Now it's harder to detect, but given differences between take-home and in-person evaluation, I would guess it's closer to 25% -50%

130

Eric Pedersen @ericjpedersen.bsky.social · 08/07/2026

A couple counter-points: a) an open book exam isn't equiv to using LLMs; I didn't see a spike in grades when we did online exams during COVID. If you don't grasp the ideas, having your notes doesn't help you answer a novel Q. LLMs are very different; basically having a 2nd person write the test

140

Reposted by Eric Pedersen

Grant McDermott @gmcd.bsky.social · 03/07/2026

𝐭𝐢𝐧𝐲𝐩𝐥𝐨𝐭 v0.7.0 now available on CRAN 🎉 This is big release, with loads of new features. Same tiny footprint, though ;-) Short highlight thread below, borrowing examples from our gallery: grantmcdermott.com/tinyplot/vig...

https://github.com/grantmcdermott/tinyplot/blob/main/vignettes/gallery_figs/bubble-quakes.R

https://github.com/grantmcdermott/tinyplot/blob/main/vignettes/gallery_figs/simpsons-paradox.R

https://github.com/grantmcdermott/tinyplot/blob/main/vignettes/gallery_figs/barplot-meat.R

https://github.com/grantmcdermott/tinyplot/blob/main/vignettes/gallery_figs/spineplot-titanic.R

48726

Eric Pedersen @ericjpedersen.bsky.social · 02/07/2026

Good to know! Is it possible to have a default session with common packages precompiled that can be reopened when needed, without having persistent environment?

100

Eric Pedersen @ericjpedersen.bsky.social · 02/07/2026

The startup isn't huge if you're spending a few hours coding, but I rarely find time to do that. If I want to quickly spin up a regression model to check something, or make a figure for a paper, the extra compile time was frustrating. May be better now; it was 2 or 3 years ago when I last tried

020