It was a treat to have Smooches spend Shelter Skip Day with my team and me in Washington! Smooches is a rescue dog currently being housed at the D.C. location of the @BrandywineSPCA. Smooches packed a lot into one day on the Hill: meetings with constituents and my friend @RepJohnnyO, trying out some new material, and a full Capitol tour. And I’d dare to say he was the most popular pup on the Hill for the day! Thank you so much to Brandywine Valley SPCA for bringing Smooches by!

Toadstools living up to their name!

カンブリア紀のyunnanozoans（Yunnanozoon lividum）はステム脊椎動物であるという仮説をさらに支持する証拠が報告されています！

Family wealth generated while holding office. We are not angry enough about this.

Pride in STEM🏳️‍⚧️

Our new paper published in "Coral Reefs", led by B. Domaszewicz shows tiny Devonian solitary "button" #corals, functionally similar to modern ones, eg. Cynarina. They were able to excavate themselves from sediment cover. #Devonian #Morocco OA paper: doi.org/10.1007/s00338-026-0…

This one's been on the backburner for a while but this image of a sunfish was passed over to me from @morethanadodo archives. There's no info with it and I can't find any copies of it on a quick online search. Any sunfish afishianados out there recognise it?

AOC: This is what drinking water in Georgia looks like after Meta began data center construction in the community.

The Malili Lakes ricefish (adrianichthyid) radiation is an example of a "lake species flock" where many closely-related species evolved quickly in a relatively small area. Here's a map from my ichthyology class showing 15 such flocks in ancient lakes around the world.

Guys, there is a very big difference between citing a real paper unnecessarily or for the wrong reason, and completely making up a detailed, convincing-looking reference to a paper that does not exist. Two completely separate things. One of them is clear academic misconduct.

This will impact also paleobiology in near future -- isotopic tools for the reconstruction of the properties of food webs. CC: @LDaumantas @svalver @BlaiVidiella @PalaeoPhilo @CoelhoPre

Citing an AI hallucinated source that doesn’t exist is not equivalent to a typo.

I am deeply affected by the murder of UofWashing student Juniper Blessing. Her family/friends highlighted her interest in meteorology. I donated in her honor to UW’s Dept of Atmospheric and Climate Sci. If you wish to memorialize this talented young woman: atmos.uw.edu/alumni-and-comm…

Physics-based weather models still beat AI when it matters most. Not on average. On the most extreme days. This is the opposite of what we've been hearing... A new paper in Science Advances ran every major AI weather model: GraphCast, Pangu-Weather, Fuxi, against ECMWF's HRES across 162,751 record-breaking heat events, 32,991 cold records, and 53,345 wind records in 2020. On average conditions, the AI models win. GraphCast, Fuxi, and the rest outperform HRES on standard temperature and wind benchmarks across most lead times. This matches what every prior benchmark study has shown. AI weather forecasting is genuinely impressive. Then the researchers asked a different question. What happens when the event is unprecedented? Not extreme. Not the 95th percentile. Actually beyond anything in the training data. HRES won every single category. Heat records. Cold records. Wind records. Nearly every lead time. The performance gap was largest at short lead times, where AI models should have the most information and the least uncertainty. The bias pattern is pretty massive. The AI models systematically underestimated how extreme the events were. The bigger the record exceedance, the larger the underprediction. The researchers describe it as an implicit 'soft cap': the models behave as if they can't forecast values much beyond the most extreme thing in their training data. The bias grows almost linearly with how far the event exceeded the record. HRES showed no such pattern. This isn't a fluke. The same result held in 2018 and 2020, which had opposite ENSO conditions. It held across the tropics, subtropics, mid-latitudes, and high latitudes. It held for all three variables. It held when the researchers ran an alternative evaluation specifically designed to avoid the forecaster's dilemma. The mechanism is pretty straightforward. AI weather models are trained on ERA5 reanalysis data from 1979 to 2017. They learn to interpolate between historical weather patterns. When a new initial condition arrives, they find the nearest analogues in training and produce something in between. Record-breaking events, by definition, have no close analogues. The model has never seen anything quite like this, so it regresses toward the most extreme things it has. Physics-based models like HRES don't work this way. They solve partial differential equations describing atmospheric dynamics. They don't need a historical analogue for a 48°C heatwave in Siberia. The physics doesn't care whether it's happened before. The authors are careful about what this means. AI models remain faster, cheaper, and competitive on average conditions. Probabilistic AI forecasting is developing rapidly. Data augmentation with simulated extreme events and hybrid physics-AI architectures are plausible paths forward. This isn't a verdict on AI weather forecasting broadly. But the policy implication is quite important. The events where AI models fail hardest are exactly the events where accurate forecasting matters most. Record-shattering heat. Unprecedented wind storms. The scenarios that overwhelm emergency response, strain infrastructure, and kill people because no one expected them to be that bad. The authors wrote it plainly: it remains vital to fund and run physics-based NWP and AI weather models in parallel. I find it an unusually direct recommendation in a methods paper. Climate change means record-breaking events are becoming more frequent, not less. The training distribution is shifting. AI models trained on 1979 to 2017 data will see more and more out-of-distribution events as the climate diverges from that baseline. The extrapolation problem the researchers identified isn't going away. It's getting harder. The models that can't forecast records are being asked to forecast a world that's setting them constantly. Link to full paper: science.org/doi/10.1126/scia…

In 1935, two American doctors examined seven women's ovaries and saw small lumps. They called them cysts and named the disease after them. They were wrong. It took 91 years to fix. What we called PCOS is now Polyendocrine Metabolic Ovarian Syndrome (PMOS), announced today in The Lancet by an international panel of doctors and patients. The renaming followed more than a decade of consensus work and 22,000 patient and clinician survey responses. The lumps Stein and Leventhal saw were never cysts. Modern imaging shows they were follicles, the tiny sacs inside the ovary that grow and release an egg each month, frozen partway through by a hormonal imbalance. PMOS is a multi-system disorder centered in the endocrine system, the body's network of glands that produces hormones like insulin (controls blood sugar), cortisol (the stress hormone), and thyroid hormones (set the body's metabolism). The ovary trouble flows downstream from there. The naming choice is not academic. When doctors hear "ovary" in a diagnosis, they look at the ovary. "Metabolic" and "endocrine" send them to the whole body. PMOS affects roughly 1 in 8 women worldwide, more than 170 million people. The WHO estimates 70% have never been diagnosed. Among those who do, 1 in 3 wait more than 2 years, and nearly half see 3 or more doctors first. The CDC reports more than half of women with PMOS develop type 2 diabetes by age 40, a risk 5 to 10 times higher than women without the condition. Around 37% have clinically significant depression, compared with 14% in women without it. Anxiety runs at 42% versus 8.5%. A label born from a 1935 look at seven ovaries is finally going away. The new diagnostic guidelines roll out fully in 2028. By then, a woman walking into a clinic with these symptoms should hear questions about her blood sugar and her mood alongside her cycle. Those are the parts of the disease the old name hid for 91 years.

Amazing #fossils of the #radiodont Aegirocassis benmoulai from the #Ordovician Fezouata Lagerstatte. #FossilFriday #Paleozoic

A marine stem-myriapod from the Silurian Waukesha Lagerstätte, Wisconsin, USA: terrestrial traits pre-date the transition to land [Waukartus muscularis gen. et sp. nov.] royalsocietypublishing.org/r…

One of the most frustrating misconceptions is that “crocodiles have remained unchanged in 200 million years”. Which is incorrect for a myriad of reasons. Though it also undermines how derived modern crocodilians are. In contrast to the many Mesozoic terrestrial pseudosuchians.