John Kennedy

When you get older, you start making them up by mistake because you've forgotten the correct word.

Look on my works, ye Mighty, and despair!

A screen capture of six figures from Nature journals. Each of the figures contains multiple panels, often in wildly different styles and colour schemes. None of them is legible at this scale nor, one suspect, at any scale.

Dear Nature, how are those figure limits working out for you?

Lovely. The word, that is.

Verisimilitude Someone on Bluesky was asking people for their favourite $20 words. That seems kind of steep, but such is inflation. Mine were: peristeronic, septentrional, titivation, and velleity. However, the l…

Q. What do CGI cigarette butts and global temperature estimates have in common?

A. Verisimilitude.

diagrammonkey.wordpress.com/2026/03/03/v...

Verisimilitude Someone on Bluesky was asking people for their favourite $20 words. That seems kind of steep, but such is inflation. Mine were: peristeronic, septentrional, titivation, and velleity. However, the l…

Word of the day is verisimilitude

On data, reality, realism, the texture of reality and, yes, verisimilitude.

diagrammonkey.wordpress.com/2026/03/03/v...

Google Has a Secret Reference Desk. Here's How to Use It. 40 Google features to find exactly what you need, the alternative search engines that do things Google won't, and the reference desk framework underneath all of it.

TIL about a large number of operators/tools that can be used to make Google searches much more focused and productive
🧪 #philsci

cardcatalogforlife.substack.com/p/google-has...

We need a realistically realistic reality detector.

Verisimilitude Someone on Bluesky was asking people for their favourite $20 words. That seems kind of steep, but such is inflation. Mine were: peristeronic, septentrional, titivation, and velleity. However, the l…

Word of the day is verisimilitude

On data, reality, realism, the texture of reality and, yes, verisimilitude.

diagrammonkey.wordpress.com/2026/03/03/v...

I would be interested to see what effect they have. For some reason that I cannot now recall, I have the notion that some of the WWII warm bias is a sampling issue and is not wholly due to systematic errors in the measurements. More data would certainly be helpful.

How many?!

I guess the question is what can dataset creators do (or not do, or do differently) to help users make informed choices?

Duo Chan has been working on a process for helping users find their ideal dataset, which might help with such things, but for any particular use it is ultimately the user's responsibility to work out if the dataset they're using has the desired characteristics.

Digitizing weather observations from World War II US naval ship logbooks World War II is a significant event in the history of the world, with disruption to all aspects of life including routine weather observations. We now face the consequences of these observational gap...

Interesting to see this. And, we have lots more data for WW2 in the Pacific now, but I don't think they've been integrated into any of the datasets yet: rmets.onlinelibrary.wiley.com/doi/full/10....

Addressing the World War 2 Warm Anomaly in HadSST.4.2.0.0 We present an update to the Hadley Centre Sea-Surface Temperature dataset (HadSST.4.2.0.0) that addresses residual warm bias during the Second World War (WW2). Using an existing quantitative definitio...

Some of the HadCRUT biases are fixed in HadSST.4.2.0.0 but the cold early 20th century bias remains and I don't think we're much closer to understanding why it occurs even if we have good reason to think it real.

arxiv.org/abs/2602.03347

I was also delighted to discover another global temperature dataset that I had previously missed, with a delightful name - LaMa. Reading the LaMa paper now and its claims of realism, after this new one that supersedes it and its claims of realism, one wonders how realistic a dataset can get.

It's frustrating because that seems like it would be easy to solve the latter by a small modification of the constraint term to account for the error covariances. The bias issue is one that remains to be widely solved, but the potential uncertainties there are larger than the reconstruction errors.

It rather overstates its case without any great need to do so: it's a useful new dataset, with caveats around the model dependence, but it manages to inherit the weaknesses of its input - HadCRUT has known biases - without also inheriting its strengths - the error covariance information.

From noise, all this A new paper has appeared on arXiv: Generative deep learning improves reconstruction of global historical climate records. It uses, as you might guess from the title, generative deep learning to rec…

From noise, all this

Notes on a new global temperature reconstruction that uses a spatio-temporal probabilistic diffusion model generative deep learning thingummybob to fill gaps in the data.

diagrammonkey.wordpress.com/2026/02/28/f...

Peristeronic
Septentrional
Titivation
Velleity

It’s one of very few global temperature reconstructions that uses a spatio-temporal model; finds a bunch of El Niños hiding in plain sight! has interesting things to say about the Arctic. Some oddities in the recon tho and it doesn’t fully take into account observation uncertainty.

From noise, all this A new paper has appeared on arXiv: Generative deep learning improves reconstruction of global historical climate records. It uses, as you might guess from the title, generative deep learning to rec…

From noise, all this

Notes on a new global temperature reconstruction that uses a spatio-temporal probabilistic diffusion model generative deep learning thingummybob to fill gaps in the data.

diagrammonkey.wordpress.com/2026/02/28/f...

It’s not “democratised” if you have to pay a private company a shedload to participate.

YouTube video by MN Center for Philosophy of Science AI and the Nature of Science: Keynote presentation by Carl Bergstrom

how LLMs change incentives for research - a great talk by @carlbergstrom.com , had to watch in parts, a lot to digest. #institutional_epistemology at its best! www.youtube.com/watch?v=9Mj3... #philsci

QRP with your skull art.

on behalf of the bluesky team i am excited to announce our newest feature!

Now in addition to replies and quotes, we have the "don't reply"

here's how it works:
1. you see a post that does not mention you
2. you "don't reply" to it
3. the app shows you more posts

From noise, all this A new paper has appeared on arXiv: Generative deep learning improves reconstruction of global historical climate records. It uses, as you might guess from the title, generative deep learning to rec…

From noise, all this

Notes on a new global temperature reconstruction that uses a spatio-temporal probabilistic diffusion model generative deep learning thingummybob to fill gaps in the data.

diagrammonkey.wordpress.com/2026/02/28/f...

Splendid! I’m adding this to the “testimonials” part of my website (right after I create a “testimonials” part of my website).

From noise, all this A new paper has appeared on arXiv: Generative deep learning improves reconstruction of global historical climate records. It uses, as you might guess from the title, generative deep learning to rec…

From noise, all this

Notes on a new global temperature reconstruction that uses a spatio-temporal probabilistic diffusion model generative deep learning thingummybob to fill gaps in the data.

diagrammonkey.wordpress.com/2026/02/28/f...

Alexander Gray On a Cat, Ageing He blinks upon the hearth-rug, and yawns in deep content, accepting all the comforts that Providence has sent. Louder he purrs, and louder, in one glad hymn of praise for all the night’s adventures, for quiet, restful days. Life will go on for ever, with all that cat can wish; warmth and the glad procession of fish and milk and fish. Only – the thought disturbs him – he’s noticed once or twice, the times are somehow breeding a nimbler race of mice.

He blinks upon the hearth-rug,
and yawns in deep content,
accepting all the comforts
that Providence has sent…

—Alexander Gray, “On a Cat, Ageing”
#poem #poetry #cats