In case you missed it last week, I wrote about "doing your own research", about the path from research to public knowledge, and about knowing what you don't know. #scicomm mixture.substack.com/p/i-know-not...
If you know the word you'll know exactly what my penultimate wrong guess was. Wordle 1,762 4/6
🟨⬜⬜⬜⬜
⬜⬜🟩🟩🟩
⬜🟩🟩🟩🟩
🟩🟩🟩🟩🟩
It's the time of year where I remind you that I can write science conference summaries (UK). On my website you can find previous ones I did for Cambridge Uni and the Royal Society.
Unlike AI, I understand the talks and can focus on what's important. #scicomm
evaamsen.com/client-projects/
Why do mainly women use the cycle paths though? Do men not cycle at all, or not on the paths?
I'm happy to see this finally published! It's a longread that I wrote a few months ago about why there is no Strep A vaccine. It has quite an interesting history, with ripple effects of a 1960s trial still affecting how much research funding it gets today. 🧪 press.asimov.com/articles/strep
Injecting chemotherapy into people's eyes sounds nightmarish, even delivering drugs into their eyes with particles from pig semen instead is a huge improvement! www.livescience.com/health/medic...
Doesn't look like @evaamsen.com posted her article when it published, that doesn se(e)m(en) right?!🧪
Ha, I forgot! There was a small mistake in the title at first and then I got sidetracked after it was fixed.
While I was away last week, this article I wrote for @boldinsights.bsky.social was published. It's about a study that followed how mentoring can undo some of the socioeconomic effects that determine which academic track children are placed on.
boldscience.org/do-children-...
I thought they used OpenStreetMap? (At least for the walking/cycling routes, they do. Maybe they rely on Google for transit schedules?)
I am currently in the Lake District, but through the magic of scheduling content (and a lot of prep work) I can still release a newsletter today. This is a long one, combining a lot of interesting discussions and articles about publishing, knowledge and #scicomm
open.substack.com/pub/mixture/...
@molsonismydog.bsky.social Do you remember why I have this? Or rather "had".
Screenshot from Google email. "Your Google Account is being deleted. rolluptherim@gmail.com"
I haven't lived in Canada for over 15 years, so who knows when and why I created the email address rolluptherim@gmail.com but Google is letting me know that it's deleting it. I'm sure it's just an abandoned inbox full of spam.
(This year the people earning over £50k have to start, so that does not include most science journalists!)
I had a brief moment of panic that I forgot to look into an MTD tool (thing for UK self-employment tax stuff) but then I remember that a) I don't have to do it until next year and b) it's built into my business bank account so I already *have* the MTD tool.
In case you missed it when I posted it a few weeks ago: I updated the "From Science to SciComm" guide for 2026. It's aimed at PhD students/postdocs who want to transition to #scicomm payhip.com/b/zSQ0
Great! That means I don't need to change my settings in Buffer. It still knows that it's me :)
Just testing if I can still schedule posts from Buffer, because I haven't used it since I updated my username here. If you see this, it worked!
These are the people who are also attacking universities in the US. They just hate universities in general.
"The Analytical Engine has no pretensions whatever to originate any thing. It can do whatever we know how to order it to perform. It [cannot] anticipat[e] any analytical relations or truths. Its province is to assist us in making available what we are already acquainted with."
— Ada Lovelace, 1843
I just can't believe I walked right into that. Usually I see it and avoid it, but I think I thought that it was too niche for the NYT
I usually get a perfect score in Connections, but the trap got me this time.
Connections
Puzzle #1029
🟩🟨🟪🟦
🟩🟩🟩🟩
🟨🟨🟨🟨
🟦🟦🟦🟦
🟪🟪🟪🟪
Is that how it happens? I'm struggling to see how else these fake references go undetected by the people submitting the papers.
I've edited researchers' papers sometimes and they send them to me with references as well, but I (a human) understands not to mess with that bit. An AI won't.
And I *guess* that they are giving AI the text with references already added. The AI doesn't actually understand any of the text. It's just using patterns to fix sentences, and in fixing it it decides to add some more reference-shaped bits. (It doesn't know that references refer to real papers.)
Writing is hard. It's my entire job and even I think it's hard, but I like it. Scientists didn't choose to be professional writers, though, and might struggle with it, especially because they often have to write in another language. So for the writing part of the process they ask AI for help.
What is supposed to be happening: Over the course of months/years you read papers and save them in a reference manager. You build on this knowledge with your own research, and when you write your own paper you refer back to these earlier papers
What I *think* is happening: the writing part is hard
This is about how to detect these fake citations, but I'm trying to wrap my head around *how* fake citations end up in papers. (thread)
www.nature.com/articles/d41...
In case you missed it when I posted it a few weeks ago: I updated the "From Science to SciComm" guide for 2026. It's aimed at PhD students/postdocs who want to transition to #scicomm payhip.com/b/zSQ0
right now the astronauts are calling houston because the computer on the spaceship is running two instances of microsoft outlook and they can't figure out why. nasa is about to remote into the computer
We had this at the American school I attended in Morocco in the '80s. There was a font that made letters look like balloons and 8-year-old me was OBSESSED. Whenever I got computer time I made cards or banners with the balloon font. It was the first thing I ever did on any computer! #throwback
I've recently written a feature article related to this but it isn't out yet. Soon, though!
