steph's Avatar

steph

@urschrei.eurosky.social

Academic at the ⋂ of cities, technology, and climate adaptation. Reluctant polygon enthusiast. Sometimes I work on computational geometry and spatial data algorithms, which I promise almost never to discuss.

750
Followers
593
Following
2,331
Posts
05.05.2023
Joined
Posts Following

Latest posts by steph @urschrei.eurosky.social

I have 67 followers in common with a guy who just described his understanding of LLMs as Kriging, a technique developed in the 1960s. I feel like this does not reflect well upon me.

08.03.2026 21:59 👍 0 🔁 0 💬 0 📌 0

whAt??

08.03.2026 21:48 👍 1 🔁 0 💬 0 📌 0

I saw an interesting comment from an academic elsewhere regarding the word “train” – they insisted that their PhDs and postdocs were advisees. I love this because even though we don’t really use that term in Europe (ime), I loathe the term “train”. That’s not what’s going on.

08.03.2026 21:43 👍 1 🔁 0 💬 1 📌 0

A colleague’s kid went to SF in summer of 2024 (first in maths from Trinity, very bright, data science job offer). I wonder whether the re-election + LLM double whammy changed his long-term plans…

07.03.2026 17:59 👍 1 🔁 0 💬 1 📌 0

They have a guy who bursts into your room, roughs you up, and starts hollering about huge asses. He has to do this 50 times per day and longs for death.

07.03.2026 09:51 👍 6 🔁 0 💬 0 📌 0

Oh my god.

06.03.2026 20:40 👍 0 🔁 0 💬 0 📌 0

Feel like @sifu.tweety.fish can provide a useful answer. The fact that Dan Dennett was attending symposia about it in the 2000s makes me suspect it’s in the “wrong, but interesting” camp (a perfectly respectable place for scientific work IMO)

06.03.2026 17:54 👍 3 🔁 0 💬 2 📌 0

That makes me wonder whether you could store a verified ORCID in your PDS. That’d make it pretty easy to get all your papers (well, a lot)

06.03.2026 16:53 👍 4 🔁 0 💬 1 📌 0

sniffin’ and vibes, but who can really say

06.03.2026 13:11 👍 1 🔁 0 💬 1 📌 0

Many people in universities remain oddly reluctant to talk about 2020-2022 in secondary schools. I was doing research with 16 year-olds throughout the pandemic, and always want to talk about it.

05.03.2026 21:37 👍 5 🔁 0 💬 0 📌 0

Christ. Glad you’re mostly OK.

05.03.2026 20:37 👍 0 🔁 0 💬 0 📌 0
Preview
geo/geo/src/algorithm/ball_tree.rs at ball_tree · georust/geo Rust geospatial primitives & algorithms. Contribute to georust/geo development by creating an account on GitHub.

You can grab the ball tree impl from github.com/georust/geo/bl… if you want. It needs 5 minutes of minor changes, suggested here: github.com/georust/geo/pu…. That should get you big speedups for HDBSCAN and KNN.

05.03.2026 17:45 👍 3 🔁 1 💬 1 📌 1

“The ePub version of Don DeLillo’s unacknowledged 1980 hockey romp is available on Anna’s Archive” is quite the thing to find oneself murmuring.

04.03.2026 15:10 👍 3 🔁 0 💬 0 📌 0

I love him (?)

02.03.2026 17:27 👍 1 🔁 0 💬 0 📌 0

The superciliousness is not the most enjoyable part of the discourse, no. But maybe it makes the most enjoyable part even better?

02.03.2026 13:19 👍 0 🔁 0 💬 1 📌 0

I wonder whether a big part of this (silly, but the passage of time is going to settle it anyway) argument is that the vast majority of non-coders think LLMs are the chat interface, and are not distinguishing between it and the loop. I know Cowork is looping under the hood but maybe it’s not GA yet?

02.03.2026 13:12 👍 6 🔁 0 💬 1 📌 1

Number 3 is clearest and most descriptive I think. Was leaning towards 2, but the methodology doesn’t have to be hinted at in the title at this level.

02.03.2026 12:09 👍 1 🔁 0 💬 0 📌 0

It’s both friction and the twisting process of spinning.

02.03.2026 11:23 👍 4 🔁 0 💬 1 📌 0
Spinning (textiles) - Wikipedia

It is “just” spinning. Like, using a spinning wheel! The Wikipedia entry explains it with specific reference to wool: en.wikipedia.org/wiki/Spinnin...

02.03.2026 11:18 👍 0 🔁 0 💬 0 📌 0

Around 3 hours start to finish has been my recent experience for a single-purpose complete algorithm of quite high complexity (n log n production-quality HDBSCAN, ball-tree was separate work).

01.03.2026 21:05 👍 7 🔁 0 💬 0 📌 0

good man yourself adrian

01.03.2026 20:55 👍 1 🔁 0 💬 0 📌 0

Meanwhile, holed up in a hotel on 42nd street, Alice Neel executes one of the finest C20 cityscapes three days after fleeing her home, following the destruction of her life’s work by her lover “in a jealous rage”.

01.03.2026 15:52 👍 6 🔁 0 💬 0 📌 0

I really enjoy the degree to which galleries indulge the evident peccadilloes of male painters. “Like his idol Degas, Bonnard was well-known for his depiction of the female toilette, which he painted hundreds of times.”

01.03.2026 15:52 👍 5 🔁 0 💬 1 📌 0
01.03.2026 15:30 👍 12 🔁 0 💬 0 📌 0

Just a Guy with a Research Chair floridly speculating about the intentions of states. “They’re doing this to prevent anyone being held accountable. The regime will just be transferred.”
Is posting engagement bait and epistemic trespass on social media part of your ethical commitment to non-violence?

01.03.2026 15:05 👍 0 🔁 0 💬 0 📌 0

You can have a “claudeswarm when you learn to take care of your droneswarm.”

27.02.2026 22:24 👍 4 🔁 0 💬 0 📌 0

I got some “Thank you for doing the work” comments when I posted a photo of my baby in a sling here. Bewildering (but thank you).

27.02.2026 19:01 👍 3 🔁 1 💬 0 📌 0

They didn’t explain it very well, in your defence.

27.02.2026 15:19 👍 1 🔁 0 💬 1 📌 0

When you switch over, your tweets will be stored and distributed from Europe. Nobody will notice (but you’ll know), and if it ever goes away you can go back to the American one and all your tweets will still be there.

27.02.2026 15:15 👍 2 🔁 0 💬 1 📌 0

If I may be permitted a small observation about British politics, I hugely enjoy the fact that MG appears to be perennially glum. A man whose access to joy has been severed by his own actions. The protagonist in an abandoned first draft of an early modern play.

27.02.2026 10:06 👍 0 🔁 0 💬 0 📌 0