Josef Uyeda

Josef Uyeda

@pseudacris.bsky.social

1.2K followers407 following278 posts

Josef Uyeda
@pseudacris.bsky.social

I find it so depressing that more dems haven't run campaigns on this. And even this is far tamer than what it should be to meet the reality of what is happening.

Peggy Flanagan@peggyflanagan.bsky.social

This year, federal agents terrorized our streets, killed our neighbors, and detained our children. Minnesotans were changed forever. Michele Tafoya thinks ICE treats kids “really well.” I think ICE should be ripped apart and replaced by a system grounded in dignity and due process.

View full thread
Josef Uyeda
@pseudacris.bsky.social

"Drawing on ideas from Representation Learning, Representational Measurement Theory, and abductive reasoning...we then establish...the ways DCMs “think” differently from human biologists...we lay out a practical toolbox for addressing these differences and supporting meaningful inference with DCMs"

Josef Uyeda
@pseudacris.bsky.social

Caleb spent a ton of time on how to communicate these ideas with co-authors Anuj Karpatne & Arthur Porto, and with the Imageomics institute. I'm really proud of the synthesis and road map he produced. We ask what the benefit of deep learning is in biodiversity science and comparative methods.

Josef Uyeda
@pseudacris.bsky.social

"If neural networks—often criticized as opaque blackboxes—have their greatest interpretive difficulties when the underlying biology would most benefit from their ability to represent complexity, how can we expect DCMs [deep comparative methods] to possibly improve the situation?"

Josef Uyeda
@pseudacris.bsky.social

If you're submitting a reference letter for a PRFB, I was surprised to not see any option for deleting and reuploading, which contradicts the info provided in the email requesting it. In fact, as soon as you select the file it is uploaded and you can't change it. Was fine for me, but beware.

Josef Uyeda reposted

Dimitri Drekonja, MD, MS
@dimitridrekonja.bsky.social

This administration has separated more than 200,000 children from their parents www.brookings.edu/articles/how...

brookings.eduHow many children are affected by parental immigrant detention? | BrookingsExplore an interactive tool examining how many children are affected by parental immigrant detention and what the data reveals about separated families.
Aaron Rupar@atrupar.com

MELANIA: Nothing better than to see a child reunite w/parents KILMEADE: Yeah, bc Putin captures Ukrainian kids, indoctrinate them, then give them to Russian families. You're trying to undo that M: We're working together. We have a very positive relationship. They understand what we want to achieve

Josef Uyeda
@pseudacris.bsky.social

I see 2 or 3 responses from the left about Trump's $5000 dividend. It's bribery for votes, he's not going to do it, don't be a sucker. But if a Dem proposed anything of the sort, the immediate R response would be its inflationary and would increase the deficit, and this attack seems to always work.

Josef Uyeda
@pseudacris.bsky.social

This has been central to all R propaganda for ages, benefits to the working class by any means other than helping business/the rich will make you lose anyway. Best I can tell, this is a homogenous internalized world view of R's and Inds. But no one is going to activate that argument now? Hm.

Josef Uyeda
@pseudacris.bsky.social

More importantly, we view the rules of lumpability as a way to navigate model space and identify biologically plausible hypotheses about diversification and trait evolution. Hidden state models are cool because they are essentially a joint inference of character construction & evolutionary process.

Josef Uyeda
@pseudacris.bsky.social

So we don't think it's a major crisis in the field, but there are problems to solve, and excitingly, lumpability may give efficient ways of navigating the vast possibilities of character construction alternatives to find biologically plausible and evolutionarily justified representations of traits

Josef Uyeda
@pseudacris.bsky.social

Furthermore, we don't count hidden states as extra parameters, only transition rates, which leads to inadequate penalization of the degrees of freedom these models have and consistent mistakes in model inference across congruence classes using information criteria such as AIC.

Josef Uyeda
@pseudacris.bsky.social

That is to say, there are problems. But at the same time, model selection with so-called "irreducible models" is a reasonable solution. It doesn't guarantee that model is the one that best maps onto the real biological process that happened, but it does address the statistical issues of congruence.

Josef Uyeda
@pseudacris.bsky.social

A lot of what we found requires a balanced reaction: In hidden state modeling of trait evolution and diversification, there are large congruence classes that emerge as we increase the number of hidden states. Because of these congruence classes, there are statistical and interpretational issues.

Josef Uyeda
@pseudacris.bsky.social

Spot on, but for the more pickle brained signatories I suspect that not only is it demonstrating their priorities, but the positive feedback between their priorities and their algorithm. I could easily believe they mostly don't know what's happened whereas this was a defcon1 event in their feeds.

Carl T. Bergstrom@carlbergstrom.com

If you just happened to pick this week to start yelling about the importance of academic freedom, I sure hope you did more than watch silently for the past 18 months as the Trump administration assaulted everything from mRNA vaccine research to gender studies to the very idea of having universities.

Josef Uyeda reposted

Carl T. Bergstrom
@carlbergstrom.com

Replying to @carlbergstrom.com

Because if you didn't say a word about DOGE eliminating every federal grant with any element that Grok thought had anything to do with LGBTQ health, climate change, health disparities, and most of all DEI? I don't think you give a shit about academic freedom. I think you're just a racist.

Josef Uyeda
@pseudacris.bsky.social

Really nice article laying out the semantics of sex/sexes that causes persistent confusion and arguing in our community. It lays out the undisputed patterns in biology and makes clear any "lack of consensus" on sex/sexes is arguing across different levels. academic.oup.com/icb/article/...

academic.oup.comEvolution, Development, and the Incoherence of Sex: A Framework for Multiple Sex ConceptsSynopsis. The word sex can refer to at least seven distinct, evolutionarily related biological phenomena (0–6 below). Bacteria and archaea use mechanisms f
View full thread
Josef Uyeda
@pseudacris.bsky.social

In other words, using their definitions, a term like "to fertilize" blends together sex1 with sex2 (and maybe sex4 & sex5) in ways that are unnecessary, promote confusion, and often fail when confronted with biological diversity.

Josef Uyeda
@pseudacris.bsky.social

We like to think of our terminology as fundamental building blocks from which all biological variation can be described, but it is not and we essentialize a biased typology of "normal" in biology at the cost of clarity and understanding, and a great byproduct of harm to real human communities.

Josef Uyeda
@pseudacris.bsky.social

Really nice article laying out the semantics of sex/sexes that causes persistent confusion and arguing in our community. It lays out the undisputed patterns in biology and makes clear any "lack of consensus" on sex/sexes is arguing across different levels. academic.oup.com/icb/article/...

academic.oup.comEvolution, Development, and the Incoherence of Sex: A Framework for Multiple Sex ConceptsSynopsis. The word sex can refer to at least seven distinct, evolutionarily related biological phenomena (0–6 below). Bacteria and archaea use mechanisms f
View full thread
Josef Uyeda
@pseudacris.bsky.social

The authors understandably use placeholders for the distinct concepts for each concept of sex/sexes (e.g. sex0, sex1, sex2). That we don't have words for these yet in biology is frankly embarrassing, especially when we have so many dizzying and barely pronounceable terms in sex biology.

Josef Uyeda
@pseudacris.bsky.social

My favorite example: There is no isogamous version of the verb "to fertilize" that doesn't imply actor-receiver, i.e. mutual causation over syngamy. The fact this key biological concept is described by a word with constitutive asymmetry, despite the biology not needing it, shows the depth of bias.

Josef Uyeda
@pseudacris.bsky.social

Really nice article laying out the semantics of sex/sexes that causes persistent confusion and arguing in our community. It lays out the undisputed patterns in biology and makes clear any "lack of consensus" on sex/sexes is arguing across different levels. academic.oup.com/icb/article/...

academic.oup.comEvolution, Development, and the Incoherence of Sex: A Framework for Multiple Sex ConceptsSynopsis. The word sex can refer to at least seven distinct, evolutionarily related biological phenomena (0–6 below). Bacteria and archaea use mechanisms f
Josef Uyeda
@pseudacris.bsky.social

Biologists need to get this right because our words and authority get weaponized in the public sphere. Interrogate your usage and beliefs to see whether you can justify collapsing your definitions; 1) is it biologically accurate? 2) does doing so cause harms? 3) What can we do to better teach facts?

Josef Uyeda reposted

Matthew Hahn
@3rdreviewer.bsky.social

Very cool pair of papers demonstrating little link between amount of additive genetic variation and amount of nucleotide variation within species: journals.plos.org/plosbiology/... www.pnas.org/doi/10.1073/...

journals.plos.orgLevels of additive genetic variation vary substantially between speciesAdaptive potential determines how populations respond to environmental change. This study compiles evolvability estimates from over 170 multicellular species, revealing substantial variation across th...
Josef Uyeda
@pseudacris.bsky.social

Interdisciplinary scientist is an awesome science role, but my experience is that it isn't rewarded early on. Being a translator is a profession by itself, interdisciplinary science doesn't just happen by putting experts in a room together. It's more natural & effective if "trained from birth" 1/n

Pam Davis-Kean, PhD@umpamdk.bsky.social

I consider myself an interdisciplinary scientist. A post on LinkedIn challenged the idea that we should mentor our students to be more interdisciplinary. Here are my thoughts on what that means in my new APS Presidential column. www.psychologicalscience.org/publications...

psychologicalscience.orgIs Interdisciplinary Science Worth It?"You don't want scholars answering to two or more disciplines. You want them to enrich and integrate their own discipline with ideas drawn from others."
View full thread
Josef Uyeda
@pseudacris.bsky.social

I didn't believe I had some skills in translation til I was in a room of amazing paleo + molecular evolution people and realized they weren't talking to each other, and that I actually had a useful role in bridging the conversation. That's the short term advice: keep the faith that its needed. 4/n

Josef Uyeda
@pseudacris.bsky.social

Longer term, we need to realize and create programs that actually reward interdisciplinarity, boundary crossing, and translation services earlier in people's careers. It's not that it can't be done later, but I think if we really want boundary crossers, we should incentivize it from day 1. 5/5

Josef Uyeda
@pseudacris.bsky.social

Interdisciplinary scientist is an awesome science role, but my experience is that it isn't rewarded early on. Being a translator is a profession by itself, interdisciplinary science doesn't just happen by putting experts in a room together. It's more natural & effective if "trained from birth" 1/n

Pam Davis-Kean, PhD@umpamdk.bsky.social

I consider myself an interdisciplinary scientist. A post on LinkedIn challenged the idea that we should mentor our students to be more interdisciplinary. Here are my thoughts on what that means in my new APS Presidential column. www.psychologicalscience.org/publications...

psychologicalscience.orgIs Interdisciplinary Science Worth It?"You don't want scholars answering to two or more disciplines. You want them to enrich and integrate their own discipline with ideas drawn from others."
Josef Uyeda
@pseudacris.bsky.social

But the incentives in science are impatient. Learning two languages means you spend less time in one, and feel less confident/productive in either, and that pressure is likely to increase a trainees anxiety about their identity and role. 2/n

Josef Uyeda
@pseudacris.bsky.social

One strategy is to always pretend your expertise is in the thing the person who you're talking to knows least, lol. But you sort to have to have faith and be patient that it will eventually pay off to avoid the impostor syndrome that creates. 3/n

Josef Uyeda reposted

Alexei Drummond
@alexeidrummond.bsky.social

Get excited :) PhyloSpec is coming! A shared standard + modelling language for phylogenetic models. Write a model once, run it across engines like BEAST X, BEAST 2.8 & RevBayes. Easy and reproducible. Built by @tochsner.bsky.social & collaborators (ETH Zürich, Auckland, LMU). phylospec.com

phylospec.comPhyloSpecA standardized way to describe phylogenetic model components, common assumptions, and best practices in the field of phylogenetics.
Josef Uyeda
@pseudacris.bsky.social

Another day of Lysenkoism

Paul Bleicher@pbleic.bsky.social

Astonishing. RFK Jr. adds 8 members to a 6 member FDA advisory panel. 6 of them were sellers of peptides, and the 7th a compounding pharmacist. In an 8-6 vote with all 8 votes coming from these new panel members, the panel approved 4 peptides against the objections of FDA scientists.

nytimes.comFDA Panel Supports Lifting Restrictions on Four Peptides (Gift Article)In an all-day hearing, the agency’s scientists argued against allowing broad access to unproven drugs that the agency had banned three years ago.

Josef Uyeda reposted

Jake Berv
@jakeberv.bsky.social

Replying to @jakeberv.bsky.social

As a reminder, I am starting a new lab at UNC Charlotte in 2027—and I am eager to apply these new tools, along with others, to diverse systems. Get in touch if you have ideas. bsky.app/profile/jake...

Jake Berv@jakeberv.bsky.social

#Evol2026 I’m launching a new research group at UNC Charlotte in January 2027. Prospective students, postdocs, and collaborators are welcome to get in touch: jacob.berv@charlotte.edu More at jakeberv.com, lab website coming soon.