Hey, I wrote a thing about AI in astrophysics ergosphere.blog/posts/the-ma...

I thought I was „good at travelling.“ Now with kid and older parents, I realise I was just playing on Easy.

As an #HE educator, if you wonder, “when will COVID start to end?”, it is 2038: that’s when the first cohort enters uni whose development was not impacted by locked down nurseries, schools, and unis. #UKHE

The thing is the *entire internet* is different now than it was a decade ago (and, largely, not for the better). All of this is interrelated and there's no going *backwards* only forward. So work for a better future, don't angle for a return to an idealized past.

High-tech anti-modernism: When the people who make and sell high-tech embrace anti-modernist postures – beard, bible, biceps. (Out of ... masculine self-loathing?)

Eventbrite has turned the simple 3 minute act of filling a web form into a 30 minute rage-quit session of having to wait for and undo useless AI-pregen "work saving magic" to then struggle and fail to get "responsive" JavaScript from hell to register and save an image and text description.

Unsolicited PI advice: never put a student on the critical path of any software development project that others will rely on.* *Unless they were a software developer in their previous career.

@guyintheblackhat.bsky.social say, to brush up on my knowledge about the (women‘s) parlour game etc prehistory of TTRPGs, what should I read? Quick public access resources preferred

On Saturday at 3 PM ET I'm going to be streaming a game I haven't thoroughly played since the mid-90s -- Cosmology of Kyoto, the early game that blew Roger Ebert's mind with the possibilities of an "open world." Also you can get molten metal poured into your ass and out of your mouth! Be there.

Fancy diving into the cultural impact of pen-and-paper RPGs? Then join James Bennison, @adders.blog, Esther MacCallum-Stewart and myself for a chat in Brighton's Wagner Hall on April 21. Details 👇 www.eventbrite.co.uk/e/the-dice-t...

🎓 PhD Studentship: 3.5years, fully funded: Come build engaging multi-platform data collection tools for digital wellbeing interventions with me, @nballou.bsky.social, @martidisi.bsky.social at Imperial College. Application deadline Feb 26, full details here: bit.ly/phd-digital-...

Jobs! The Autotelic Interaction Research Group is looking for (1) creative practitioners and (2) theory researchers to work on "Autotelic Creative Artificial Intelligence" (ACAI): (1) PhD in Creative Practices (2+2 years): lnkd.in/dPBG47rM (2) Postdoctoral Researcher (2 years): lnkd.in/dMisM_bf

3/3 rejected research bids in the last weeks, aka there goes last year’s summer. Not bitter, just reflecting on the reviews: the funding system seems to have quite the limited appetite for uncertainty from novelty and ambition — or I have too large an appetite for them.

We create emotional worlds for children we seem to universally value and embrace. When and why do we decide these are not worlds worth aspiring to and realising in our adult lives?

🎓 PhD Studentship: 3.5years, fully funded: Come build engaging multi-platform data collection tools for digital wellbeing interventions with me, @nballou.bsky.social, @martidisi.bsky.social at Imperial College. Application deadline Feb 26, full details here: bit.ly/phd-digital-...

Just a big THANK YOU 🙏 to @cfiesler.bsky.social for writing up how to deal with fair use images in ACM articles. As an @acm.org EiC, all I can say this is gold & should be embedded in standard instructions and forms. caseyfiesler.com/2016/10/20/f...

There is no way to safety-engineer LLMs (especially in “agentic” form) in the same way that you cannot safety-engineer general purpose computing. You intentionally afford practically infinite use cases. If anything, they are less safe because they lower the expertise barrier to doing stupid shit.

🚨Calling all new media scholars🚨 If (and I emphasize if) a 2nd edition of the New Media Reader happened to be in the works, what additions would you like to see? What do you think could be excised? @docmofo.bsky.social @n-w-f.bsky.social @psssssssss.bsky.social mitpress.mit.edu/978026223227...

Listened to @pluralistic.net.web.brid.gy's great 39C3 keynote "The Post-American Internet." He remains one of the true hopepunks: modelling to me how finding concrete opportunities for action and charging us with hope is probably the most effective activism today. pluralistic.net/2026/01/01/3...

"AI should augment not replace humans" has become a meaningless trope that is and can be happily slapped onto any AI replacing/degrading human work initiative. So what is an effective retort and/or move to a more productive stance? Just, "Can you specify what that looks like in this context"?

Spring cleaning of a former business travelling, long time Mac user: when you find yourself in possession of no less than nine now obsolete Mini-DVI to VGA/DVI dongles …

A lovely gentle and accessible dialogue on the limits of computationalism.

I used to say on Twitter, back when I was on there, “don’t retweet anything that you wouldn’t pay to promote”, as a way of trying to explain the value of algorithmic amplification. Instead, I learned a lesson about how much people really do want to subsidize and amplify things they pretend to hate.

Discworld QOTD, from Hogfather

Research is the MAGIC TIME on any project where I get to go have fun and explore. And 100% what you are looking for are the surprising and amazing details and textures that you could never make up

There isn't a single problem "solved" by edtech that couldn't be fixed with smaller classes led by well-paid teachers given real academic freedom

This kind of “automating the science pipeline” discourse, apocalyptic or utopian, takes as given the upside promise of foundation models, and doesn’t grant symmetry to the downside: if AI can exponentially automate science, it can also exponentially automate undetectable scientific fraud.

was at an event on AI for science yesterday, a panel discussion here at NeurIPS. The panelists discussed how they plan to replace humans at all levels in the scientific process. So I stood up and protested that what they are doing is evil. Full post: togelius.blogspot.com/2025/12/plea...

Quick boost: Come do a PhD with me!

I have to say, "more and more people are doing spacewalks" gotta be up there on the more-and-more-ism paper openers ranking –– dare I say, sky-high?

I fear that the hidden curriculum we put our students through is „muddling through chronic overload without being caught.“ Then again, that’s the reality we model and live as educators in HE today, so it figures it’s the reality we create for our students.

How to live and work in a world flooded with bullslop where the overwhelming majority of people is not capable nor interested in detecting bullslop nor able to do what they are asked to do without a bullslop machine.

Well fuck. You weren’t supposed to build the Stand Alone Complex. Injecting the social world with infinite undetectable impostors that increases the cost of honest signals by orders of magnitude is quickly becoming my top 2 „negative externality“ of LLMs. Class action, anyone?

inspired by this, I have done a free extract of this section of the book here. it is about understanding that *with all its problems* the BBC is as important for our intellectual health as the NHS is for our physical health: naomialderman.substack.com/p/the-bbc-an...

Research method textbooks in academia are a bit like email clients in software development: somehow, at some point in your career, everyone ends up being dissatisfied with everyone else’s and has to write their own which is 99% feature-identical to what’s already out there.

Come do your PhD with me at Imperial College in London! The LISS DTP funds 3.5y PhD tuition and stipend (+1y masters), deadline Jan 23: liss-dtp.ac.uk/applying-to-... See topics of interest below 👇

Mood.

With the 2025 @conference.aiandgames.com just around the corner, @tommy.aiandgames.com had something to get off his chest, or at least part of it. Game AI is facing an existential crisis, and we need to start talking about it. www.aiandgames.com/p/game-ais-e...

The meetings continue until morale improves.

2025: a late stage soviet bureaucrat larp you can never leave. Everything is strained and worn beyond function, everything breaks ever-faster, & the job is to make do & pretend & report numbers up & to the right, while everyone knows & copes & bonds over black humour & too much drink in private.

There is something truly remarkable about Generative AI. Several years in this revolution, half of the US GDP dumped to make these tools free, and i can't think of a single generated artifact, text, image, sound, video that had a significant cultural impact.

Feeling existential weariness today at so much academic effort promising to create AI systems and design frameworks that "augment rather than replace", support wellbeing sustainable just etc etc etc –– and yet, and yet, all it will achieve is Gutmenschen shelfware.

1. We ( @jbakcoleman.bsky.social, @cailinmeister.bsky.social, @jevinwest.bsky.social, and I) have a new preprint up on the arXiv. There we explore how social media companies and other online information technology firms are able to manipulate scientific research about the effects of their products.

This reads like what a reviewer 2 should have written, leading the handling editor to decide accordingly.

Imagine organisations took all the developer years they currently pour into AI assistants and used them to make their services more stable, accessible, performant, sustainable. The uplift in CSAT etc would be massive and long-term in comparison. The opportunity costs of AI are staggering.

Very pleased to see submissions to the special issue on Game Advertising of ACM Games: Research and Practice coming in. This is a last call! Deadline is 31 October 2025. We're looking forward to your submission, and if you need a smol extension… Slide into our emails. 😉 dl.acm.org/pb-assets/st...

on.ft.com/46M9mXI Why designers abandoned their dreams of changing the world. Free to Read.

In London, hate will never win.

Imperial opened its President’s PhD Scholarships (first deadline Nov 3): bit.ly/3JW2xuW Ping me to apply for a scholarship on: * Competence/empowerment/learning progress in games: models and DDA * The appeal of aleatoric uncertainty in games: models and tests ...