iiinova — where your digital self keeps evolving

Reposted by Matthew Russell

Hadas Weiss @hadas.bsky.social · 07/10/2026

sure, sex is great but

313313

Matthew Russell @matteomics.bsky.social · 02/10/2026

My day feels very much like this! Although I only have a cold, Darwin was dealing with worse.

000

Matthew Russell @matteomics.bsky.social · 28/09/2026

Sometime the real potential benefit of "reproducibility" is really auditability. The possibility of laying out the data and the steps for critical appraisal. The reproducibility makes that audit possible, it is not a magic test on its own.

091

Reposted by Matthew Russell

Colette Delawalla, PhD @cdelawalla.bsky.social · 28/09/2026

The humanities have been calling out genAI theft since the jump, while STEM has rallied around “discovery potential”. I am *ONCE AGAIN* reminding EVERYONE that science is not the sacred cow, STEM isn’t the exception. They are going to steal our IP, too. We aren’t special.

8371150

Matthew Russell @matteomics.bsky.social · 28/09/2026

I think LLM finding and linking up research a single individual might never consider together, which maybe is what these stories really are, might actually support academic idea generation. But that use case can't justify or repay the trillions invested.

000

Matthew Russell @matteomics.bsky.social · 24/09/2026

Really makes the case that various grades of #masspec based #proteomics could be useful at different points in #biotechnology development and processing monitoring. The next paper will show what I think is a really cool example.

001

Matthew Russell @matteomics.bsky.social · 23/09/2026

The first one has strong Diana and Actaeon vibes. Actaeon is turned by the Goddess of the hunt into a dear who is brought down by his own hounds. en.wikipedia.org/wiki/Diana_a...

en.wikipedia.org

Diana and Actaeon - Wikipedia

000

Matthew Russell @matteomics.bsky.social · 21/09/2026

The UK BBC has a talk radio channel called Radio 4 very middle aged middle class. It has a feedback programme where people who blatantly have R4 on as a constant background in their lives write in to complain about it. Similar phenomena I think.

020

Matthew Russell @matteomics.bsky.social · 14/09/2026

What does it say about me that my first thought was "deep fry"

000

Matthew Russell @matteomics.bsky.social · 14/09/2026

This is surely the same guy who ends up with Sarena Williams serving a ball cleanly through his skull when he thinks he can win a point from her. Great writing btw.

0110

Matthew Russell @matteomics.bsky.social · 12/09/2026

I know that I myself have behaved as though truly inanimate machines like car park charging meters etc could be reasoned with, thanked etc. A knowing whimsy. But maybe applying theory of mindset that way creates a vulnerability to these machines that more plausibly seem to be thinking.

010

Matthew Russell @matteomics.bsky.social · 12/09/2026

Unit geekary is definitely my favourite kind of geekary.

071

Reposted by Matthew Russell

Joe Bak-Coleman @jbakcoleman.bsky.social · 11/09/2026

I stand firmly by my assertion that metascientists should have seen this capture of science coming, but there is no reasonable way they can reasonably delude themselves into thinking the new nsf announcements are anything other than epistemic capture. www.science.org/content/arti...

science.org

Science’s reform movement should have seen Trump’s call for ‘gold standard science’ coming, critics say

Efforts to improve the rigor of research may have unwittingly handed the administration a way to attack science

29038

Matthew Russell @matteomics.bsky.social · 09/09/2026

This is the important idea. Much thought, especially scientific, doesn't happen in language. It's expressed and communicated in language later. LLM can summarise and recompose that expression, but not think it, or think more of it.

000

Matthew Russell @matteomics.bsky.social · 07/09/2026

Brian Deer was the UK journalist who covered / uncovered Wakefield back in the innocent 00's. Has a load of his articles on his website here: briandeer.com/mmr/st-mmr-r... In case anyone wants to revisit.

briandeer.com

Some of Brian Deer’s reports – briandeer.com

021

Matthew Russell @matteomics.bsky.social · 07/09/2026

This is an important take. Environmentalism as a generalisation of the principle that one shouldn't defecate in ones own bath water.

000

Matthew Russell @matteomics.bsky.social · 07/09/2026

Oh yes. Hence the manipulation of the dates autism was suspected in the kids in the study, because the lawyers wanted close association between jab and autism, which was a key part of the research misconduct case.

2131

Matthew Russell @matteomics.bsky.social · 07/09/2026

Yeah, it was to push for single jabs over the triple MMR right? Although anti-vax continue to cycle through single jab demands, demand ban for particular ingredients etc. I guess those positions are reasonable looking gateways to entice new followers.

000

Matthew Russell @matteomics.bsky.social · 07/09/2026

Was in the UK through the Wakefield thing. It was bonkers. Alt-med got big and people like @bengoldacre.bsky.social did great work exposing it. Felt like the good guys won for a bit. Seems now that was just a skirmish. Did not see the for-profit grift / far-right politics link. Should have.

2282

Matthew Russell @matteomics.bsky.social · 06/09/2026

If you'll allow bags as technically I'd say bags. Carry more than your hands will hold. For original human tool makers let you retain multiple tools, and invest time making them. Bags make it possible to be human. The Alps 'ice man' had a bag that would not look too out of place today.

010

Matthew Russell @matteomics.bsky.social · 06/09/2026

But what it really did was write a python script using pre-existing python libraries and drawing on a repository of audio analysis python code. Then ran the python script on the given audio file. It's cool. But it's LLM python writing, not originally problem solving.

120

Matthew Russell @matteomics.bsky.social · 04/09/2026

My Star Trek TOS fan theory is that the use of physical control panels in the sci-fi setting is due to a resent culture wide rejection of touch screens and GUI. Possibly following their own enshitification crisis.

190

Matthew Russell @matteomics.bsky.social · 02/09/2026

Love this framework. Current Me is always furious with Past Me but pushing burdens onto Future Me.

030

Reposted by Matthew Russell

🎃Hana🎃 @robonohana.bsky.social · 02/09/2026

Reading a book before bed is such a dangerous activity. For example, last night: "I'll read a couple chapters." Suddenly it's 1:47 AM and I've negotiated with myself that being tired tomorrow is a problem for Future Me. Future Me is now furious at Past Me. #booksky

921013

Matthew Russell @matteomics.bsky.social · 01/09/2026

In good faith. Look for details. The wall socket plates has neither docket holes nor switch. The light switch is silly. The room exits to the right make no sense. The items, especially electronics, on shelves/units in background make no sense. Then, cat does not behave like a cat, so...

120

Matthew Russell @matteomics.bsky.social · 31/08/2026

We have a data processing PC named after her.

030

Matthew Russell @matteomics.bsky.social · 30/08/2026

Also most papers in most fields most of the time are incremental advances. Real break throughs do occur and are recognised. Where are the actual examples of world changing discoveries coming back from peer review with "extraordinary claims etc" in the notes.

000

Matthew Russell @matteomics.bsky.social · 28/08/2026

I once spend an entire afternoon puzzling over prevelence, incidence and lifetime risk.

000

Reposted by Matthew Russell

Mark Janello @markjanello.bsky.social · 23/08/2026

You fuckers made me do it. The Racist Academicals Patter Song.

[RACIST ACADEMICAL]
I am the very model of a racist academical,
With footnotes and citations to make bigotry seem credible,

I traffic in ideas that have long since been discredited,
Like human races being real and IQ is inherited!

I’ll sign an open letter that legitimates my bigotry,
My gender ideology is thuggery and piggery,

If criticized it’s academic freedom I will hide behind,
My high and mighty principles shan’t be questioned by your tiny mind!

[CHORUS OF CHUDS]
His high and mighty principles shan’t be questioned by your tiny mind!

12879299

Matthew Russell @matteomics.bsky.social · 26/08/2026

Great little thread. Statistics and experimental design and policy all perfectly covered.

010

Reposted by Matthew Russell

Julia M. Rohrer @dingdingpeng.the100.ci · 25/08/2026

“…as peer reviewers or question askers at conferences, we must accept that our pet issue is not more important than the other 10 issues and if the analyst has had a good go at handling a bunch of issues, we can’t just pile more and more issues on them to fix. It is impossible.”

2354

Matthew Russell @matteomics.bsky.social · 24/08/2026

Seems to me the same people who obsess over IQ on a racist/eugenics basis, are also the most keen on the creation of artificial general intelligence as a super intelligent slave for themselves.

010

Reposted by Matthew Russell

Katie Mack @astrokatie.com · 24/08/2026

Sometimes society at large rejects certain ideas because they’ve been thoroughly shown to be wrong and also because they suck and the people who espouse them suck and when someone comes around arguing those ideas should be part of some edgy modern debate we can and should shun them and move on.

7166891359

Matthew Russell @matteomics.bsky.social · 21/08/2026

When I switch to R, all those moments will be lost ... like numeric values coded as strings.

010

Matthew Russell @matteomics.bsky.social · 20/08/2026

And Aquinas only meant specialists are more more adept in their own area of expertise than generalists. Not a comment on fanatics.

010

Matthew Russell @matteomics.bsky.social · 20/08/2026

Hmm, the 95% CI on "never use AI" on the left drops to similar level exam score as right hand "use AI". It's possible former no / cheat homeworkers are now AI home workers, but final performance is the same. So low exam performers are constant over time, but now self select AI user label.

110

Reposted by Matthew Russell

noam @noamchompers.bsky.social · 11/08/2026

sorry, you want me to *revise* and then *resubmit* this shit? dog, i sent it to you because i never wanted to see it again

547667

Reposted by Matthew Russell

tj mahr 🤘 @tjmahr.com · 11/08/2026

interesting how “correlation is not causation” shifted from a warning about spurious associations to something incurious people say to refute things

1226

Matthew Russell @matteomics.bsky.social · 10/08/2026

Sounds like an excellent system.

010

Matthew Russell @matteomics.bsky.social · 06/08/2026

It may be we agree really. Although RCTs are only one kind of experiment, obviously particularly important in vital fields of medicine, agriculture etc.

000

Matthew Russell @matteomics.bsky.social · 06/08/2026

Also out of all possible variables in the world, mostly unrelated or correlated with each other, your mechanistic theory picks 2 and predicts their relationship, and that relationship is confirmed experimentally? That's positive evidence for the theory, and so the causal relationship.

110

Matthew Russell @matteomics.bsky.social · 06/08/2026

If you have a mechanistic theory about the world that predicts x has causal influence on y (providing the theory wasn't concocted to explain a prior observed correlation between x and y). Then do experiment to check the correlation. If the correlation exists the theory has passed at least one test.

100

Matthew Russell @matteomics.bsky.social · 06/08/2026

I think some of the puritanical "correlation is not causation" statements around the place are missing the whole structure of scientific method which provide the context the statistics are done in.

110

Matthew Russell @matteomics.bsky.social · 06/08/2026

Then there's the middle situation. Where a theory posits causation, between treatment and outcome maybe, but subgroup analysis or similar produces so may comparisons that random chance of correlation swamps the results.

100

Matthew Russell @matteomics.bsky.social · 06/08/2026

That's different from a massive scan of many variables to find correlated ones. Which is a legitimate exercise in feature discovery. But pairs of identified correlated variables should be treated with great caution re causation.

100

Matthew Russell @matteomics.bsky.social · 06/08/2026

Having a theory that predicts an as yet undetermined causal relationship between x and y, informing an experiment which finds a correlation between x and y, does add evidence for the correctness of the theory.

120