Philipp Schlegel @uni_matrix@qoto.org

The Rezavall lab is recruting postdocs, PhD students and research assistants:

"Our lab investigates how the brain makes decisions "
https://www.rezavallab.org/

#neuroscience #PhDPosition #PhDJobs

Oxford relying on ‘Deliveroo-style’ contracts with most tutorials not taught by full-time staff

"Data obtained through freedom of information requests by the University and College Union and shared exclusively with the Observer reveals that about 61% of core tutorial teaching is done by academics on fixed-term contracts or in hourly-paid roles. The union says that pay is based mainly on contact hours with students, but once preparation, marking and supporting students are taken into account, this often falls below the minimum wage in real terms."

Proud of my peers Oxford UCU for brining this to light.

https://www.theguardian.com/education/2024/nov/16/oxford-deliveroo-contracts-tutorials-full-time-staff-gig-economy

Mit der Reform des #Schwangerschaftskonfliktgesetzes gilt nun: Bis zu 5.000 Euro Bußgeld drohen bei Belästigung von #Schwangeren vor Beratungsstellen oder Arztpraxen. Gleiches gilt, wenn Ärztinnen und Ärzte, die #Schwangerschaftsabbrüche vornehmen, behindert werden.

The RNA technology that was used to develop most of the covid vaccines is driving an absolute revolution in medicine right now. A vaccine for malaria, the disease that has killed more people than anything else in history. Personalized vaccines for cancers — the list goes on and we're just getting started. Much of this development is locked up in commercial pharma companies — there are papers, of course but replicating them in practice, let alone continuing development, is very hard without active access to the team involved. It's also very expensive. A lot of that development money comes from the US government. Now, yes it's a problem that that funding is going to private companies that are ransoming patients for their life savings in exchange for life-saving medicine, but this is as much a regulatory problem as it is a structural one. As long as those teams exist, the possibilities of these tools exist also.

Having an American government that wants to actively dismantle vaccine-related research and eliminate its funding is an existential threat to humanity:

https://mas.to/@Fitnessfoundry/113484127007301813

Facebook memories reminded me of the one truly viral thing I ever made, this graph, circa 2011.

A cleaned up version appeared in the Guardian and it's now common to find versions of this graph for different countries.

To the very best of my knowledge, my janky copy and paste job was the first.

A whopping (~90%) reduction of progression to Type 2 diabetes with tirzepatide (GLP-1 drug, dual receptor) vs placebo in a randomized trial of >2,500 participants with obesity, absolute reduction of 10/100 treated
https://www.nejm.org/doi/full/10.1056/NEJMoa2410819

I stopped using Twitter a year ago, but kept the account active for reasons. Today, I decided I didn't care about those reasons any more.

https://whatever.scalzi.com/2024/11/11/twitter-the-final-exit/

SO GOOD. European alternatives for nearly all Amazon, Google, Microsoft products along with alts to Github, Slack, etc. now we need to create a North American version of this. thank you, @vortex! https://european-alternatives.eu/alternatives-to

My Fediverse Advice:

* Follow more people. No, even more people than that. Basically, if you find a real person and their posts are good follow 'em
* If you get a good reply to a post boost it. If you make a good reply to a post boost it. Replies are not visible in the feed unless you do this. As long as the post is an OK start to a conversation or interesting boost it.
* Write thoughtful replies. And if you put effort into a reply boost it or probably only the people tagged will see it.

When an article says "some scientists think" then remember this: I, a scientist, once thought I could fit a whole orange in my mouth. I could, it turns out, get it in there, but I hadn't given sufficient thought to the reverse operation. 🧵

More evidence -- this time with audio tapes from the original experiment -- showing that the Stanford Prison Experiment was rigged to encourage terrible behavior by the guards

In Medium: https://gen.medium.com/the-lifespan-of-a-lie-d869212b1f62

Unpaywalled: https://archive.is/K7gMN

Talking about this with my son, I was struck by how many pieces of major science/theory/culture in the mid-20th-century insisted that they "proved" humanity was inherently, natively selfish -- but turned out to be much less true

1/x

#statistics #pvalues #math #analytics

@eric_normandeau @jonny @LudoWaltman @eLife

It's all too clear, isn't it. #WebOfScience would like the status quo to not change at all – they are profiting enormously from it. Whereas scientists would like the status quo to change – our grants are being drained dry from publishing fees, our libraries are also drained dry from subscription fees, and academic administrators spend countless moneys on bibliometrics supplied by Clarivate to "evaluate" academics for their "throughput" – in quotes because using these words for what they actually do (counting papers and citations) is a perversion.

By the way #Clarivate owns #EndNote, #Publons and #ScholarOne – we really ought to not use *any* of these. It's in our power.

#ScientificPublishing #academia

#programming #names Aargh

Throw the programmers in the sea

The NeuronBridge website, for matching #ElectronMicroscopy and #LightMicroscopy data for #Drosophila #Neuroscience research, now includes EM data from FlyWire. Getting it working was a great effort by @neomorphic, Cristian Goina, Hideo Otsuna, Robert Svirskas, and @konrad_rokicki. Here is a screen capture of looking up a neuron on FlyWire Codex, finding its match in NeuronBridge, then viewing the match in 3D with volume rendering in the browser.

https://neuronbridge.janelia.org

This is a mistake

https://ec.europa.eu/commission/presscorner/detail/en/mex_24_5324#:~:text=Commission%20concludes%20that%20online%20social%20networking%20service%20of%20X%20should%20not%20be%20designated%20under%20the%20Digital%20Markets%20Act

Interesting feature of the Apple LLM reasoning paper. I always tell my students that exams include no irrelevant information, which gives you a clue as to the answer. LLM's have learnt this too well, and can't ignore irrelevant info (perf drops 17-70%).

https://arxiv.org/abs/2410.05229

In other words: exam-style questions have massive leakage that many students don't pick up on but that LLM's find impossible to ignore. I suspect this tells us more about the way we write exam questions than anything else. They're not a good measure of LLM performance, nor human!