iiinova — where your digital self keeps evolving

Reposted by Sean Mackinnon

Cobalt60 @co60.bsky.social · 06/10/2026

"Confidence level at 2 sigma is fine!" I shout as the bailiff hauls me out of court. "It's not like we're doing particle physics!"

2502

Sean Mackinnon @seanpmackinnon.bsky.social · 06/10/2026

This is an oldie, but as a Haligonian I do really like this short: www.youtube.com/watch?v=M9RK...

youtube.com

22 Minutes: Heritage Minute - The Donair

YouTube video by CBC Comedy

140

Sean Mackinnon @seanpmackinnon.bsky.social · 06/10/2026

I ended up doing the job in SPSS because that was already pre-installed, and would be faster and easier than trying to convince IT to install R for me lol

020

Sean Mackinnon @seanpmackinnon.bsky.social · 06/10/2026

I will never forget the time that IT services at a hospital let me install Rstudio, but they drew the line at also installing R to go with it

1162

Sean Mackinnon @seanpmackinnon.bsky.social · 05/10/2026

It totally seems like its treason. I feel like CSIS must know what is going on, I wonder if they are waiting for the election to happen to find evidence of foreign interference to get a smoking gun (legally speaking). Foreign influence is dubiously legal, but not actively interfering in votes

020

Sean Mackinnon @seanpmackinnon.bsky.social · 04/10/2026

It feels like journals are only a filter for baseline quality, there really is a niche out there for people to personally curate reading lists for others. There is a cottage industry for this in other content like movies, so it would be cool for an academic to get paid to specialize in this way

5224

Sean Mackinnon @seanpmackinnon.bsky.social · 04/10/2026

Interesting thread under this OP My academic reading time is limited due to the demands of the job, and the primary problem for me is that it is hard to find good papers. The literature is vast and most journal articles are boilerplate I'm dying to find some truly interesting ideas/theory

2185

Sean Mackinnon @seanpmackinnon.bsky.social · 02/10/2026

A student asked me how to change the default colors of the text in script files in RStudio and I realized I never did this before. There are a ton of different themes under Tools -> General Options -> Appearance They were so pumped about being able to customize their themes

2101

Sean Mackinnon @seanpmackinnon.bsky.social · 30/09/2026

I was very confused trying to figure out what beardo was a slur for and its just .... guys with beards? America is a strange place

110

Sean Mackinnon @seanpmackinnon.bsky.social · 30/09/2026

I backed up an author in a journal review as reviewer 1 by calling out a bad reccomendation by the other reviewer who was saying it was out of scope and other weak criticisms. As a reward for my good karma, I was able to log into the journal website 1st try without resetting my password, incredibly

0120

Reposted by Sean Mackinnon

Stephen Wild @stephenjwild.bsky.social · 29/09/2026

Guy who thinks linear regression is progressive because the outcome is on the left

3252

Sean Mackinnon @seanpmackinnon.bsky.social · 29/09/2026

Sigh, Canada basically did the same thing. Made a huge fund to attract fleeing US researchers (who won't teach) while still letting the actual institutions themselves slowly die away from frozen funding It's better than nothing I guess, but very badly allocated money to the HE sector

020

Sean Mackinnon @seanpmackinnon.bsky.social · 28/09/2026

But ... but ... numbers go up! Aren't numbers always supposed to go up?

000

Sean Mackinnon @seanpmackinnon.bsky.social · 27/09/2026

East Coast of Canada we use it like the US as the arranged board (we are like 50/50 US vs. British on a lot of stuff)

010

Sean Mackinnon @seanpmackinnon.bsky.social · 23/09/2026

Feel like maybe there is something up with those stats or the methods, if 15% of bi men won't date other bi men

000

Sean Mackinnon @seanpmackinnon.bsky.social · 22/09/2026

I think that most online sources are gating access with identity verification because it is the only viable solution to handle the bot traffic. Anything that is online, unsupervised, anonymous, and provides compensation is bot food now.

110

Sean Mackinnon @seanpmackinnon.bsky.social · 22/09/2026

All options are bad options for online data, but the second definitely doesn't push to in-person data collection. A lot of folks do identity verification first, then allow people do to the online study. Most ppl don't use AI for surveys, but those that do impersonate 100s if fully anonymized

110

Sean Mackinnon @seanpmackinnon.bsky.social · 22/09/2026

Has anyone out there tried using visual illusions as CAPTCHA to fool AI models? Stuff that would look incorrect to biological visual perception, but not to computers? You'd have to use illusions not already annotated with text explaining what they are supposed to see, but in theory maybe possible?

000

Sean Mackinnon @seanpmackinnon.bsky.social · 22/09/2026

Soon enough online studies will be dead on arrival. Best defense at the moment is probably to compensate less money than the token cost. A lot of researchers I know also require a meeting on video call with ID prior to letting them participate so one person can't imitate 1000

120

Sean Mackinnon @seanpmackinnon.bsky.social · 18/09/2026

Yet another example of "Huh, these results don't look at all like what I expected. I wonder what ... ah yes, my code is rubbish."

030

Sean Mackinnon @seanpmackinnon.bsky.social · 18/09/2026

Baking cookies is an underrated teaching strategy, tbh. I've done it a few times!

110

Reposted by Sean Mackinnon

Joshua Raclaw @joshuaraclaw.com · 18/09/2026

Still seeing faculty talk about how they had Chat write a thing and then had students critique what was good or bad about it as if this is a pedagogical innovation for the ages

2258

Sean Mackinnon @seanpmackinnon.bsky.social · 18/09/2026

Honestly, it really is a good analogy in a lot of ways if interpreted like the book. A lot of people are like, "But I will use the power for good!" and yet... AI tools are even massive environment-destroying technologies, which is also basically consistent with the core message of the books

150

Sean Mackinnon @seanpmackinnon.bsky.social · 18/09/2026

This post was in truth, in part for you: I knew you would be as outraged as I was hahaha

260

Sean Mackinnon @seanpmackinnon.bsky.social · 18/09/2026

Omg the thread there is incredible. Check out this analogy a guy uses to defend their use of AI The implication being that they are Frodo because they are able to use the one ring wothout being corrupted. I uh, got news for this guy about what Lord of the Rings was about

1261

Sean Mackinnon @seanpmackinnon.bsky.social · 18/09/2026

Assuming the details in the reddit post are true, there is a lot more wrong than inconsistency. The PI is showing a complete lack of engagement in the process, such that they probably didn't even read the thesis themself (or else they would understand the issues the student raiseed)

050

Sean Mackinnon @seanpmackinnon.bsky.social · 16/09/2026

I'm happy if people name the papers, scholars, or specific theoretical frameworks! Just not new for the sake of being new

020

Sean Mackinnon @seanpmackinnon.bsky.social · 16/09/2026

The asymmetry is that that if there really isn't much in an area, it is almost impossible to prove the negative that newer or more varied references don't exist. Conversely, it's easy to prove that an author is missing something by providing examples, so I feel the onus is on the reviewer, imo

130

Sean Mackinnon @seanpmackinnon.bsky.social · 16/09/2026

I think that (when I do get this) it is frustrating when people are like "This looks thin, add more references" but then don't provide any examples of what they want to see, because it feels like a guessing game. Maybe it's right, but it doesn't feel helpful to me, personally.

110

Sean Mackinnon @seanpmackinnon.bsky.social · 16/09/2026

Notably, every single time I have gotten this comment nobody ever provides an example of a study that I am missing (e.g., cite this person or this paper), just that they want to see a larger number of citations with a year larger than X

130

Sean Mackinnon @seanpmackinnon.bsky.social · 16/09/2026

One of my least favorite comments when getting a journal review back is some variation: "I looked at your literature review, and only X references have publication dates in the past 5 years. Update the literature review to include more recent references" Just bean counting by date is annoying

3110

Sean Mackinnon @seanpmackinnon.bsky.social · 15/09/2026

My least favorite variation on this is labels that are identical except for differing numbers of trailing spaces...

010

Sean Mackinnon @seanpmackinnon.bsky.social · 15/09/2026

I took MANOVA off the curriculum for obvious reasons, but there is one small part of me that still likes the elegant beauty of ANOVA on matrices, even if it's not that useful. The math of it kind of has this "This is neat" energy.

static.klipy.com

Marge Simpson: I Just Think They're Neat!

ALT: Marge Simpson: I Just Think They're Neat!

040

Sean Mackinnon @seanpmackinnon.bsky.social · 14/09/2026

Seems like bait to get dunked on tbh

010

Sean Mackinnon @seanpmackinnon.bsky.social · 14/09/2026

Woof, things could always be worse huh? That's awful. It's a dark time for the university sector

010

Sean Mackinnon @seanpmackinnon.bsky.social · 14/09/2026

If only there were some way a bridge could pay for itself and its own maintenance, a kind of tax on bridge crossing that wasn't too onerous on citizens and already longstanding and expected.

020

Sean Mackinnon @seanpmackinnon.bsky.social · 14/09/2026

Her I know, since she's in my department; plus, we've co-published a bit. I'm a stats person, so I've gotten to work with a lot of varied people over the years. The silos are real though; when we were locked out and on strike last year, the upside was that I got meet ppl I'd never otherwise meet!

010

Sean Mackinnon @seanpmackinnon.bsky.social · 14/09/2026

They stopped for me, but now they are replaced with "Today is the first day of classes and I haven't prepped anything, and am teaching English or something not in my discipline"

010

Sean Mackinnon @seanpmackinnon.bsky.social · 14/09/2026

No, I don't know them (but I don't know that many from social work, truth be told; different faculties!)

110

Sean Mackinnon @seanpmackinnon.bsky.social · 14/09/2026

This ... can't be psychometrically valid right? It's basically calculating interaction terms and then treating them like some sort of numerical variable for analysis like a conventionally averaged scale. Am I missing something here, or is this as weird as I'm thinking it is?

140

Sean Mackinnon @seanpmackinnon.bsky.social · 14/09/2026

I have encountered a truly bizarre coding scheme for a survey. People are asked two 6pt scale items for each of 9 stems (concern & expectancy). The total score is calculated by *multiplying* these two together, then averaging those 9 multiplied terms together www.researchgate.net/publication/...

260

Sean Mackinnon @seanpmackinnon.bsky.social · 14/09/2026

I get it, though at the university I work at there is a formal hiring freeze. A lot of consultants are proposing admin doing this instead of firing people, so cuts happen slowly with retirements and no replacement. Older folks not retiring has historically been a problem, but the game had changed

210

Sean Mackinnon @seanpmackinnon.bsky.social · 13/09/2026

The hand gesture in panel 4 is what sells this comic lmao

010

Sean Mackinnon @seanpmackinnon.bsky.social · 12/09/2026

Yeah, my bet is she is just bought and paid for by American tech here. She's already borderline treasonous already with the whole referendum thing, tbh, so this would be sort of consistent in terms of aligning herself with the forces trying to destabilize Canadian government

160

Sean Mackinnon @seanpmackinnon.bsky.social · 12/09/2026

Good to know that the slope of .00 is not significant, not sure the reader would get that without the *

020

Sean Mackinnon @seanpmackinnon.bsky.social · 11/09/2026

I like my linux mint machine at home, but for a student who is not at all tech savvy it is still tricky

110