100% agree, I would add that the loss of designers and UX research across the tech industry is coming home to roost

I have a slightly different model for this. 1. There are many more lower paid workers than higher paid. So cheap automation (i.e. traditional software) scales better (and was our only option pre-LLM anyway)

I couldn’t wait to get back to my wild rogue rampage of hitching rides on bot-piloted transports and watching my serials.

If you're still dichotomising continuous variables, stop. Dichotomising your exposure &/or outcome wastes power. Dichotomising confounders introduces residual confounding. If a variable has non-linear effects, use splines! If you want 'clinically relevant' threshold effects, use marginalisation.

Leave it long enough and you'll end up in a pickle

I finally thought I understood the Monty Hall problem (kinda) and then I saw it in DAG form and now I'm pretty sure I neither understand the Monty Hall problem nor DAGs.

Double-take detail in this interview with Edith Heard, head of the Crick Her French husband decided against getting a spouse visa because of the UK's "intrusive" immigration rules If that's happening with one of the UK's most senior scientists, how many others are being put off completely?

So I have finished the main tasks in Pokopia, this chat sums up my approach, and yes we build the wall to set us free that's why we build the wall we build the wall to set us free

I think I can offer a partial explanation, as yer resident anti-AI understander. Programming is wizardry: you go to school, you read the books, you learn the arcane languages. Using AI is like being a warlock, making a deal with a demon to do the things you can’t do. And who knows the cost?

Surely a trap

Glad to be counted as a young person

My favourite tactic for being served quickly in a bar is when the bar tender asks who's next, just point to the person next to you, regardless of position in the theoretical queue. This almost always guarantees you will be served immediately afterwards

You have to do it at least once in your twenties, in order to learn: 1. That following peer pressure leads to sub optimal outcomes for you definitely and the group probably 2. That nothing good is free any more, and there's also plenty of expensive shite too

all meaning has collapsed in our post-literate society and i wrote a bunch of useless, useless words about it. read it if you, like me, are one of a diminishing minority who still reads www.theverge.com/policy/84960...

A lovely last sunrise of 2025

I am fortunate to be in a very similar position today - I've taken one big strides this year to see my distant friends more often, and will be making an even greater effort in 2026! @jamesomalley.co.uk how do I get on that mailing list 🙏🏾

I will play good side the first time around because it's a decent prior for what games will reward you for. Then subsequent playthroughs are for exploring moral choices when you know what's coming up in the story. Even the I only like being evil in a game if its funny (Kotor, fable etc).

Late to the Hades 2 party but my goodness the writing is brilliant, better than many novels I've read this year. Now we just need the last dinner party to do a cover of I am gonna claw out your eyes (and drown you to death) and the thing will be culturally complete

Another Christmas, another year I will tell you to ditch the Christmas pudding and make yourself a @nigella.bsky.social sticky toffee pudding instead

Under two weeks of my sabbatical left, and I have two messages: 1. Merry Xmas for tomorrow 2. Death to Chronos

Wildly different things, tasks, techniques, subspecialties being lumped into "AI" and then being conflated with each other, doesn't help. Different types of models vs the techniques to train them vs the tasks they are supposed to accomplish, all under "AI".

My top five favourite words to pronounce like they’re Greek philosophers…   5. Monocles 4. Bicycles 3. Popsicles 2. Obstacles 1. Testicles

Often the answer is "don't use it!". I never want to read an AI-generated novel or essay, look at an AI image, or listen to AI music. I'm not interested in the inner mental life of the shoggoths.

I remember when Michael Gove said that the public had "had enough of experts" in the run up to the Brexit referendum

Tempted to start ranking star wars movies and sneak in deep space 9

As my colleague Chris Giles put it: "The economics of this is pretty simple. 1. You should be a billionaire if you're worth it 2. Probably, no one is 3. If you're not worth it and a billionaire - you've got your money from rent exploitation - which should be stopped/taxed"

Journalist challenge: Use “Machine Learning” when you mean machine learning and “LLM” when you mean LLM. Ditch “AI” as a catch-all term, it’s not useful for readers and it helps companies trying to confuse the public by obscuring the roles played by different technologies. 🧪

Almost all musicals are significantly worse in the second half. The only exception to this is Les Miserables, and that's because Fantine is out of the picture

Of course I have beaten the jet lag, it's entirely coincidental that I took a 4 hour nap at 3pm 😐

"It is terrible that Shabana Mahmood faces overt racism for being a high profile Asian woman" = good argument "The key to ending overt racism against Asians in Britain is that we must Stop the Boats" = a very bad argument indeed, on so many levels

A great set of back of the envelope calculations - and adds to my deeply held belief that we should turn every single golf course into a wild meadow or public park

Very interested to read this in detail next week, and see what lessons we can learn if any for designing discrete active inference agents for situations where handling continuous variables is either too hard or too expensive 🙌🏾

It's too cloudy where I am so I'm just watching the 'roras on a webcam instead

Tonight I am staying at an abandoned high school converted into a cafe/hotel/cat sanctuary. The question is, what should I write on the blackboard?

What an exciting opportunity for a crossover special!

Before AI my best explanation for bad writing was that people are terrible. Now I can blame LLMs instead and be hopeful that people are not terrible after all

I will never understand why BBC Royal Correspondents write like this.

I sorely regret not walking out of Deadpool 3. At 34 I didn't have enough years left to waste on it

...and another.

This lack of reliability is I believe the real reason most companies haven't seen massive productivity gains with agents. Just because a system *can* get it right, doesn't mean it *will*. And in LLM agent loops, errors earn compound interest.

"if you have no policy, the only thing you can successfully communicate is that you are out of your depth." 👏🏾 This doesn't just apply to politics, you see it in businesses *everywhere*

It would have been so easy to write that Labour statement in a way that reflected their policy but attacked Lam/Reform "We believe that the right to stay here permanently should be based on contribution but suggesting deporting people who've earned that right is appalling."

This was a fantastic talk, and I have stolen a dozen ideas for improving the talks that I give in the future. Worth checking the recording even if you don't know anything about casual inference or psychology in general.

An incredible sentence I will be stealing for the future.

Being an IC software engineer probably has the best ratio of "do interesting work"/"admin of interesting work". But it still requires constant effort and focus, and ruthless prioritisation. Can't imagine how the academics cope

Well worth making the distinction between audio books and radio play adaptations. Don't do a half-assed effort at both, pick one form and do it *well*

Very excited to be going to IWAI next week, and after that I am on sabbatical until January 🎉 iwaiworkshop.github.io

Finally got round to listening to @roelkonijn.bsky.social on "this guy sucked" talking about Alcibiades - great fun even if you don't know much about ancient greek history like me! pca.st/vtgmuqi7

At some point I will get off my arse and create a blue sky feed that is exactly the same as the standard feed but hides all the chain quote posts