iiinova — where your digital self keeps evolving

I’m also curious how partial arcs would change this picture. Your simulations show that PF arrangement strongly affects whether the ring is trapped or can escape. An open arc might behave differently, potentially making partial assemblies better suited to dynamic coupling than a fully closed ring.

100

That seems especially interesting given the in vivo cryo-ET data, where partial rings predominate in metaphase. It would be fascinating to know how the force-coupling/escape behavior changes for, say, a 10–12-mer arc versus the closed 16-mer ring.

100

Your simulations actually seem to motivate the possibility that partial arcs may be the physiologically relevant state. Oligomer mobility scales smoothly with subunit number, and the “fuzzy” CTT-mediated interface seems inherently compatible with an open multivalent assembly.

100

Really nice preprint and lovely visuals. One outstanding question: why does the Dam1c oligomer need to close at all? Why assume the complete ring is the physiological coupler?

Max Haase @maxhaase.bsky.social · 28/09/2026

can one be confident these do anything?

Needhi Bhalla 💅🏽 @needhibhalla.bsky.social · 25/09/2026

Reposted by Max Haase

LMAO gotta love peer review: ”But outside experts said that the Anthropic researchers have yet to do the research required to confirm that they’ve found something new.” 🧪 www.nytimes.com/2026/09/24/s...

nytimes.com

Anthropic’s A.I. Is Teaching Itself Biology. Now It’s Made Its First Discovery. (Gift Article)

Scientists at the company announced their first finding: a group of enzymes whose function is unknown.

99519

Max Haase @maxhaase.bsky.social · 24/09/2026

If you look in any phage or bacterium genome you are also guaranteed to find a RT that’s associated with a tandem dna repeat.

020

yeah... I agree doesnt look great

okay maybe I am too critical, the new paper does cite the older one. well I dont know what to think...

120

Given this lack of integrity, I am now with you that the nikon videa is likely AI slop.

210

Interesting, I think I am giving up my good faith, I looked at two of his papers: from 2024 www.nature.com/articles/s41... and from 2026 www.nature.com/articles/s41... basically figures 3 and 4 from 2024 are exactly the same as figure 3 from 2026.

Probably that should be clarified on the webpage

100

Max Haase @maxhaase.bsky.social · 17/09/2026

I don't think this is AI, I think it is a pseudocolored reconstructed video from label free diffractive super-resolution microscopy, given the author's publications. So it is probably not a simple video composed of frame 1, 2, 3... rather is heavily processed, which could explain the oddities

210

I mean do we really need ai to look at supplemental ?

André Marques @amarques.bsky.social · 16/09/2026

Reposted by Max Haase

Out today in Nature: sex without crossovers. Rhynchospora tenuis makes pollen, fertilises and sets seed, yet 0 crossovers in 10,997 pollen nuclei. Chromosome drive and ~87% seed abortion rebuild the mother's genotype, so the offspring are clones. doi.org/10.1038/s415...

916472

Max Haase @maxhaase.bsky.social · 16/09/2026

You may think you are on your way, but Deutsch Bahn has different plans for you

Timothée Lionnet @successprocess.bsky.social · 11/09/2026

Reposted by Max Haase

Sharing ZATELLITE 🛰️ - a tool to visualize and manipulate repetitive loci in living cells using designer zinc fingers tinyurl.com/ZATELLITE led by Nestor Saiz, collab with Marcus Noyes, @teresadavoli.bsky.social, @glennislogsdon.bsky.social, @liamholt.bsky.social & @jefboeke.bsky.social (1/n)

principle of ZATELLITE: synthetic Zinc Fingers are fused to a fluorescent protein and targeted to repetitive endogenous sequences - this ensures damage-free tagging, which is not the case with dCas9

22817

Max Haase @maxhaase.bsky.social · 07/09/2026

While I basically agree, it’s good to remember that Socrates thought writing was a dangerous technology that would destroy human wisdom and knowledge. thank god Plato didn’t listen and wrote it all down

Maxim Greenberg @maxvcg.bsky.social · 04/09/2026

Reposted by Max Haase

I'm from DC and my parents are from SF and LA, so that pretty much validates this exercise! mydialect.us/quiz-standar...

2131

Max Haase @maxhaase.bsky.social · 01/09/2026

Pretending language is a monolith whose words can’t change meaning or pronunciation is just silly. doing so is exactly what Big Dictionary wants you to think

Reese Richardson @reeserichardson.bsky.social · 25/08/2026

Reposted by Max Haase

A massive update: At least ~15 companies~ are selling scientists antibodies using faked validation data. We've documented 18,000+ manipulated images on 17,000+ products sold by leading laboratory suppliers including Thermo Fisher, Abcam, Santa Cruz Biotechnology, Millipore Sigma and Bio-Techne. 1/🧵

18455329

Reposted by Max Haase

vincentholst.bsky.social @vincentholst.bsky.social · 13/08/2026

In 2023, @nature.com published 'Papers and patents are becoming less disruptive over time', receiving world-wide media attention. Our Matters Arising, published after a 32 month delay (more on that soon), shows that the reported decline can largely be attributed to dataset artefacts. 🧵

The average CD_5 index per year for Web of Science. The original Park et al. decline (top curve) becomes essentially flat (bottom curve) when removing papers with CD_5=1. Those papers largely correspond to dataset artefacts.

9352173

Max Haase @maxhaase.bsky.social · 04/08/2026

proves the point don't you think?

Max Haase @maxhaase.bsky.social · 03/08/2026

bluesky is so very much anti-AI that any discourse is never in good faith, quite sad

4201

Reposted by Max Haase

Carolyn Bertozzi @carolynbertozzi.bskyverified.social · 30/07/2026

We already do this and companies who want them are happy to pay for PhD student interns. Why do precious and limited NSF funds need to be deployed for this? Most PhD student pursue industry jobs anyways - that has always been true. Use NSF funds to support basic science that industry wont support!

728698

Max Haase @maxhaase.bsky.social · 21/07/2026

Strategy, tactics, logistics, political judgment, and material are collectively what constitute military power. A country cannot claim limitless power on the basis of what its weapons might accomplish under imaginary perfect leadership.

111

Max Haase @maxhaase.bsky.social · 21/07/2026

I do agree with your assessment that the current situation is an abomination.

000

Max Haase @maxhaase.bsky.social · 21/07/2026

I think you should take off you "American exceptionalism hat" and realize America is neither the richest (sure for a handful of million/billionaires it may be), most powerful (Iran has soundly repudiated its military might), nor abundant nation (starving people, homeless, medical debt, etc).

201

Max Haase @maxhaase.bsky.social · 16/07/2026

Congrats Jana! Looking forward to see your lab's awesome work!

Max Haase @maxhaase.bsky.social · 07/07/2026

I posted my intial thoughts on the ms hyp.is/_5N9iHkdEfGe...

hyp.is

Hypothesis annotation for www.biorxiv.org

011

Max Haase @maxhaase.bsky.social · 06/07/2026

Their definition of life is something (page 68)

Max Haase @maxhaase.bsky.social · 02/07/2026

In before this turns to a whole scam? usually soliciting media attention of your work before letting your peers see it is not a good sign. Anyone remember cold fusion?

120

My take away: I would not treat qed's "The 1%" as a neutral mark of excellent science. It seems entangled in the ways science becomes legible as excellent: scale, resources, networks, polish, and institutional capacity. Its more of the same, rather than some revolutionary new impact metric.

000

In sum, this analysis shows that qed selections are not institutionally exposure-neutral. Higher-output institutions had a higher per-paper hit rate. 12/n

100

Thats why this "blind" ranking feels disingenuous to me and others. It doesn't help that the methods of qed are a proprietary blackbox. My point is that hiding the institutional label is pretty meaningless when much of the paper can serve as a institutional signal. 11/n

120

Papers carry the fingerprints of the systems that produced them. Scale, polish, methods, dataset size, team size, topic, networks and resources. So a model trained to recognized "important" science may end up being entangled in those fingerprints. 10/n

This is even true if we control for field, month posted, abstract length, and author count. Institutional output still predicts qed selection. odds ratio = 1.68 95% CI: 1.44–1.97 p <1e-10 9/n

Put another way: one-paper institutes: ~1:190 papers are qed 200+ paper institutes: ~1:50 papers are qed That a meaningful difference for something branded "The 1%" 8/n

100

institutions with less than 99 papers hit at or below the expect 1% rate. whereas institutions with 200+ papers has a hit rate of 1.98% So the biggest institutions are not only getting more hits, they are converting those to qed papers at a higher rate. 7/n

100

So we can normalize the data to get a hit rate per paper and stratify the data by institutional size (number of bioRxiv papers posted). If the qed blackbox has no institutional bias we would expect the hit rate per paper to be ~1% across institutional size. 6/n

There is an obvious objection here. Big institutions post more preprints. They have more labs, bigger teams, more infrastructure, and more chances to produce a qed selected paper. 5/n

100

If we quantify this, using an HHI-style metric, we can score how institutionally concentrated the qed list is: qed papers: ~0.012 matched controls: ~0.0021, So qed papers are about 5.5x more concentrated than the controls. 4/n

I was able to replicate Ran's findings. qed selected papers are indeed concentrated to few elite institutions. 3/n