Doug McNeall

Thanks Peter!

By the library - on the way to Gandy street.

I think there is a changed and increased responsibility to get safe bike infrastructure in - lots of families using these bikes.

When we bought the bike, we used to get lots of attention with both kids up - questions etc. it’s great that cargo bike use has exploded to the extent we rarely get a raised eyebrow anymore.

We can’t yet afford an electric car, which is frustrating, but we spent money on the bike instead. It means we *very* rarely have to use the car in the city.

This is the moment we passed 5555 miles on the @ternbicycles.com GSD. It’s been critical infrastructure for us for just over 3 years.

We need more of this kind of optimism and hope. My boy in front of a new mural in Exeter.

Software engineer? Data scientist or data engineer? Want to develop AI that actually helps people? Come and work with us building data-driven climate models. Closes 12th October so hurry hurry.

lol, “accidental”. I think @drkatemarvel.bsky.social had something to say about this.

lol, “accidental”. I think @drkatemarvel.bsky.social had something to say about this.

Bringing Climate Models to Everyone How AI is making complex data understandable

Climate models can be a challenge to understand, but a new app from the NorCPM-team is using AI to make complex data understandable.

- The idea is that anyone could use it, says Eurico Noleto-Filho, a postdoc with the University of Bergen and the Bjerknes Center. 🧪

bjerknes.uib.no/en/news/brin...

We should make rich people aware of how much their lives are being made worse by climate change. It probably cuts through more than letting them it makes poor people’s lives intolerable.

Generally, these make life less tolerable, more expensive, or more shit for many many people. They enshittify the climate.

There are a very large number of low- or medium-probability impacts of climate change that are either high probability or ARE HAPPENING RIGHT NOW.

A lot of climate communication has always focused on the “low probability, high impact” events. Apocalyptic stuff like loss of the AMOC. And so it should, there are lots of civilisational-risk type worries.

Yes, I’m aware that this is the mis-use of a phrase with a very specific meaning. But hear me out.

Do everything you can to stop the enshittification of the climate.

In the near future, having someone drive by with a petrol engine will be as noticeable as when someone lights up a cigarette.

Yesssssss.

At the time they were like “it’s a really sticky idea, a vivid image” (or something) and I was like I KNOW.

Oh ffs. I asked Skeptical Science to stop doing this a decade ago, are they still going? So many of the communication experts have such a tin ear sometimes, I really wonder.

FastNet: Improving the physical consistency of machine-learning weather prediction models through loss function design Machine learning weather prediction (MLWP) models have demonstrated remarkable potential in delivering accurate forecasts at significantly reduced computational cost compared to traditional numerical ...

Also, loss function design: arxiv.org/abs/2509.17601

Technical overview and architecture of the FastNet Machine Learning weather prediction model, version 1.0 We present FastNet version 1.0, a data-driven medium range numerical weather prediction (NWP) model based on a Graph Neural Network architecture, developed jointly between the Alan Turing Institute an...

Big for AI at the @metoffice.gov.uk Fastnet dropped on the arxiv: arxiv.org/abs/2509.17658

It's basically the same.

Met Office Careers

We’ve two jobs on the AI4Climate project - one on AI downscaling and one on building data-driven climate models (with me!). careers.metoffice.gov.uk/join-us/sear...

Met Office Careers

Come and join us developing AI for solving climate problems. careers.metoffice.gov.uk/join-us/sear...

Wake up hun they found a new way to police women’s bodies.

Astounding, thanks John. I guess it’s consistent with “it’s all natural variability anyway”.

Can you plot it against global mean temperature?

A dark-themed weather tracking map from Google Weather Lab: Cyclones (Preview) showing hurricane/cyclone models in the Atlantic Ocean as of Friday, September 19, 2025, at 08:00. The map includes ensemble forecast tracks and storm markers extending from the mid-Atlantic toward Western Europe. Multiple colorful lines (green, orange, blue, purple, yellow) represent different storm model ensembles and possible tracks, with clusters of paths pointing toward Ireland, the United Kingdom, France, Spain, and Portugal. The cyclone originated east of the Caribbean, with its projected paths fanning northeastward across the Atlantic. Colored circles mark storm intensity along the tracks, numbered (e.g., 1, 2, 6) corresponding to categories. A second storm is visible off the Pacific coast of Mexico with clustered ensemble markers. The left panel lists model options (Observed, Our experimental model, WeatherNext Gen, WeatherNext Graph, ECMWF ENS, ECMWF HRES), a legend for storm markers, tracks, ensemble tracks, cones, and wind speed colors. At the bottom, sliders indicate initialization time (Fri, Sep 19, 2025, 08:00) and lead time (+6 days, Fri, Sep 26, 2025, 02:00). The background map highlights North America, the Atlantic Ocean, Western Europe, and parts of Africa.

Having AI crammed down our throats is bullshit, but machine learning has some real uses in science. This awesome page shows experimental cyclone track forecasts from Google DeepMind and Google Research's AI weather models, along with forecasts from ECMWF. 😍

deepmind.google.com/science/weat...