The time it takes to write a joke, from conception to birth, is the jestation period.

en.wikipedia.org/wiki/Supersy...

Could be, I'll double check my notes on the matter

I always keep trying to find a nice application of the supersymmetric approach to standard quantum mechanics from Witten, which lets you "square root" standard SDEs as well. It's so pretty it has to be useful, and is in many cases just not any I've researched.

A man sitting at a working desk, reading, and simultaneously looking something up in a second book. He is also wearing a corn of flowers.

Two easy working improvements from the year 1537:

1) a dual monitor setup to read and write faster
2) a flower crown to look notably cooler while doing the reading

#skystorians

Zocchihedron - Wikipedia

This is beautiful, however I noticed that the faces look irregular and it seems this design gives biased random numbers - which itself is likely a very interesting teaching example! But presumably partly explains limited enthusiasm for larger dice.

en.wikipedia.org/wiki/Zocchih...

What people get wrong about early life infections The confusing reality behind the timing of childhood pathogen exposure

Avoiding infections is good

But some simply cannot be avoided - multiple exposures over the life course are inevitable

In which case, when is the best time to get infected for the first time? 🤔

The idea of acceleration and deceleration at near g to provide gravity, signals travelling at c, and space warfare with actual bullets was just very immersive, plus the language changes in the belt, just brilliant.

I got to book 4 and watched all the series and both were excellent I thought - seeing health inequality as a major driver of the politics was fascinating and I loved the semi-realistic physics that merged no laser guns, no faster than light and no artificial gravity.

Our home planet, the only place in the galaxy that doesn’t suck

The last time human eyes were far enough from Earth to see the whole globe was December 1972.

www.nasa.gov/image-detail...

I.e. It's currently very expensive to get from South Manchester to Salford Crescent / University (a journey lots of people might like to make) but is quite fast going tram to Victoria then stopper train, it's just the lack of cost integration stopping it.

I heard that the latest extension to the Trafford Centre was deliberately under the budget where some kind of sign off was needed - the focus on integrating the buses and trains better seems to be going pretty well though, integrated trains would add a lot of practical connectivity.

bsky.app/profile/tah-...

This post just confirms folk wisdom that when it comes to egg puns for a Francophone audience, un œuf is enough.

I believe the computational overheads of running ATpro are much larger than a standard bulletin board, and presumably the fact social media is now regulated in many countries introduces certain compliance fixed costs. I don't see why a million+ active users doesn't overcome this though.

Yes that's fair, and there's obviously risks to left populism as well as right even as you say it's not a symmetry (it just seems obvious that abhorring poverty and foreigners are ethical worlds apart).

While you're not in a pile on thread, what I don't understand about economic criticism of the Greens now or Corbyn before is they're basically wanting a return to the social democratic consensus of the mid to late C20, which was hardly disastrous. There may be a distinction I'm missing here ...

Yes, when you ask AI to prove a mathematical statement it's got the unimportant wording right normally so it can be harder to spot logical errors.

Moose lying in a field with head up, while a Canada goose perches unconcernedly atop his head.

Tired of the old "elf on a shelf" nonsense? Canada GOTCHU.

The LLMs are pretty decent at teaching coding and the things they automate were on the way out as general skills anyway because the trend was always towards greater abstraction and similarity to natural language. I don't really see the usefulness for prose or concepts though.

I've ended up writing guidance for the (MSc level) students I teach along the lines of "you will all use this, treat it like any other resource, the work should be your own".

For coding, I'm a bit of a convert though - younger scientists than me don't learn C and I never learned assembly.

The performance issue with their output on other issues is they are language models, not world models - there's no representation of the people or tables or clouds they describe in their weights. And the training data use is more ethically murky.

But a language model suits programming fine.

I think this is what makes them good at providing computer code - individual "style" is usually a bad thing in Python and average code is kind of what you want. Plus I put my code on GitHub for anyone to use it however they want so please train your LLM on it.

For ongoing behaviour, there are also things like "voter models" that have more symmetry between the two states than in the SIS model, which is a bit like the "stifling" idea, i.e. both states are somehow infectious... I think there is a lot in the field to keep modellers and statisticians busy!

Personally I'm interested in a refinement called "complex contagion". There are various events and workshops coming up where people try to get better models of behaviour, but, I think the consensus would be that SIR-type models are not a bad starting point for behaviours that spread.

Some early papers also considered "rumour" models, in which if an I meets another I or an R, they are more likely to stop spreading because they think people already know, a phenomenon they called "stifling".

I am quite certain that a half-metre wide pale white spider will not lollop across my dreams.

Maybe free but with premium for things like hosting longer videos that actually cost something (premium for more "reach" like on X is going to have bad outcomes).

In a sense that's the lesson - best of intentions, memorable name arguably helpful for science as well as funny, but unintended consequences. I feel similarly about WHO conventions for pathogen names, which are quite "dull" but can avoid a lot of problems.

Honestly it's all better called infrastructure investment than subsidy - we need to make an energy transition and the best time to do it was years ago.