Milo Imbeni

@pyramidalmilo.bsky.social

PhD in Maastricht, microglia and microscopes

Shallow-water (but not deep-water) octopuses have mutations producing high-fidelity ribosomes that make a lot fewer protein aggregates of the type involved in Parkinson's, Alzheimer's, ALS & Huntington's Disease. It *may not* be a coincidence that this evolved in large-brained cephalopods.

Science Magazine@science.org · 4w ago

Certain shallow-water species of octopuses carry a mutation that reduces translation errors and prevents toxic protein buildup. Learn more: https://scim.ag/44TVJ86

A molecular quirk unique to octopuses makes them better at building proteins

I'll summarize this: Mamdani asked NYers to keep air conditioning at 78 degrees. A wave criticism hit online. A MSNBC broadcast pointed out the US Department of Energy recommends that 78 degree figure. Immediately after, the US Department of Energy deleted 6,000 web pages related to energy saving.

1/7 🧠 My journey into development begins with this work and question: how does the brain's spatial navigation system develop? We found that the neural networks for spatial navigation (tori and rings) are preconfigured and only later anchor gradually to the world with experience! 🧵

Edvard I Moser@edvardmoser.bsky.social · 5mo ago

Is spatial navigation innate 🧠? Using #NeuroPixels we show that the #torus 🍩 underlying the #GridCell map exists already on day 10 in rats — before pups open eyes and ears and before they start upright walking. 🧵1:4 👇 www.biorxiv.org/content/10.6...

Here is yet one more example. Miller's silhouette is similar to the first Palmolive bottle, whereas he would benefit from having more padding and extension, which would create the second Palmolive bottle.

Two Palmolive dish soap bottles. One is very rounded at the bottom and narrow at the top. The second has a bit more width at the top.

«The biases do not disappear; they get encoded into the model and then laundered into outputs that feel objective because they are numerical. Replacing one prestige filter with another that is harder to see or challenge is not progress toward equity in science; it’s the opposite. »

Ran Blekhman@blekhman.bsky.social · last mo.

I wanted to like QED's new 1% preprint ranking. I really did. But the more I looked at the data, the more uncomfortable it made me. Here is my full peer review. open.substack.com/pub/blekhman...

One understressed item about various AI tools is what providers are getting out of your submissions- is your text being used for something? Who’s to say? Is it going into some model? Is a person leafing through it??? I see no reason to trust anyone on this stuff

Don't use slop. Someone's trying to make money out of you and your colleagues and your friends. Do yourself a favor and read 100% human-crated content in @prelights.bsky.social Upload preprints. Read preprints. Cite preprints. Fuck AI 🧪

Homepage - preLights

Welcome to preLights, the preprint highlights service run by the biological community and supported by The Company of Biologists. Here, a team of scientists regularly review, highlight and comment on ...

prelights.biologists.com

As a scientist, there are so many issues with this. Besides the simple technical things (e.g. no baseline or comparative models), the critical one is that there are so many dimensions of impact that any single metric is bound to be horrible and biased. This is clear already in their analysis.

Rewild the reflecting pool. Make it a wetland bird sanctuary. Fill it with endangered salamanders. Foster the fermentation of whatever it is that makes swamp gas. Put up little signs welcoming fairies. Let stray senators wander in never to be seen again. Elect a bog witch. Make America Moss Again

He kept coming back, repeatedly (in the way AI often loops) to this idea of AI (specifically Claude) representing "a position without an address", and it became increasingly clear to me that there was some odd mystification happening. Of course LLM products have an address. It's a corporate address.