One of our best grad students turned down a good postdoc to take a short term job at an AI safety non-profit. I think their plan is still to apply for tenure-track positions, so that might be one (1) data point.
I agree; I don't know the relative success rate of "Apply from academia/not apply from academia" because the latter is 0/0.
We should be seeing some examples soon from people bouncing between academia and AI companies.
One reason I’m slightly more bullish on Anthropic as a company is that they sometimes seem to get that and adjust the sales pitch accordingly. If you’re a big enough company, the sales pitch is “We will send Anthropic engineers to your workplace to customise everything (so you can never leave)”
Australia is getting that way too, though the people who go off road, or what you to think they do, still drive Japanese trucks.
Part of the right's war on education, especially the part that appeals to voters and not Douthatian perverts, is pretty explicitly about limiting kids' horizons and trapping them where they are.
It’s depressing that Albanese isn’t taking more of a leadership role in things like this, or even identifying with it. For most of his career you’d have predicted it was exactly the kind of thing he would do as PM, but he’s turned out so cautious on everything.
There were reports that the Palestine Football Association Canadian visas were denied ahead of FIFA Congress on April 30 in Vancouver, BC.
It's more complicated than that. No question on the importance of them being there.
Looked into it with Aloysius Wong for CBC News.
www.cbc.ca/news/canada/...
Pets dot com was a bust, but the idea wasn’t fundamentally flawed. Chewy dot com works very well, seems to make money, and I order many more pet things from it than Amazon.
And you might have seen this on questioning whether the data centers are more efficient than a shelf of Mac minis.
open.substack.com/pub/braddelo...
As you said, two years is an overestimate for many purposes. Open source models as of April 2026 are much better at coding that closed models as of April 2024, and arguably April 2025.
bsky.app/profile/simo...
Disgraceful.
Most analytic philosophy work on writing is about fiction, so I’m not sure who would even be good to tag as an expert on philosophical issues about non-fiction.
Good thought experiment for any philosophers of literature to weigh in on.
It is wild to be doing philosophy of maths at a time where strange and new things are happening!
This week "GPT-5.4 Pro (prompted by Price)" solved Erdős problem 1196, a conjecture of Erdős, Sárközy, and Szemerédi.
www.erdosproblems.com/forum/thread...
I’ve been thinking about this in music, but it carries over to math.
People often say that LLMs can’t be truly creative. But one kind of creativity in music is putting together things that hadn’t been done together before. LLMs can do that.
And they can do it in math too, in “creative” ways.
Totally agree. This doesn’t seem culturally that different to the push to the internet in the late 90s, or the push to digitisation and the paperless office in, especially, the early 80s. Maybe it’s moving a bit quicker than either of those, but it’s not different in kind.
It’s all happening a bit quicker, but it doesn’t feel that different to the paperless office push of the 1970s and 1980s. That was primarily led by IBM and Microsoft, but I think the odd junior minister jumped on the bandwagon.
Totally agree. The task here isn’t generative, it’s classificatory. That’s already different to the familiar hallucinations you see talked about. And the rate of improvement is unbelievable, so anecdotes from even a few months ago aren’t relevant.
Motivating Job Seekers. A Field Experiment* Bart Cockx, Johan Egebark, Greet Van Hoye, Emilie Videnord, Johan Vikström February 2026 Abstract: Reduced motivation among jobseekers over the unemployment spell may lead to declining job-finding rates. We report findings from a low-cost digital intervention with motivational emails aimed at enhancing and sustaining motivation and search effort among job seekers in Sweden. Using a randomized controlled trial that included 200,720 job seekers, we evaluate both carrot messages aimed at encouraging the pursuit of personal goals and intrinsic motivation and stick messages focusing on external pressure and constraints. A large share of job seekers opened the emails, and they triggered behavioral responses. Both types of messages backfired, reducing search effort and job-finding rates. The carrot messages reduced both the number of job applications and job finding, particularly among men. One likely explanation is that these messages signal to job seekers that the Public Employment Service was less controlling than initially perceived, prompting a reduction in effort. The stick messages backfired for job seekers who, at the onset of unemployment, reported that they were motivated by an inner drive rather than by constraints. These findings underscore the challenges of motivating job seekers to actively search for jobs and suggest that low-cost digital interventions, in isolation, are inadequate and may even be counterproductive.
Interesting workingpaper on effect of motivational emails on on job search effort. The Nudge hype is as dead as a dodo. rfberlin.us21.list-manage.com/track/click?...
We at @jepub.bsky.social are excited to announce the publication of our new special issue on Open Research for the Humanities and Social Sciences, edited by Samuel Moore, Jenni Adams, and Miranda Barnes! journals.publishing.umich.edu/jep/issue/45... [1/n]
I hadn't realised that the St Kilda player was (a) Indigenous, and (b) possibly a repeat offender. What a complicated case. The penalty still seems ludicrously harsh given the uncertainty, even if it were a second offence.
www.theage.com.au/sport/afl/th...
NYC schematic tiled hex map by NTA area showing percent BA or higher within tract
An outline schematic hex map of NYC
Yesterday’s fun activity was writing a package to make cartograms like this. I am also open to lavish VC funding to develop an excessively complex German-style board game based on this layout, modeled on e.g. “Escape from New York”, “Warriors”, or perhaps “My Dinner with Andre”.
“Even mild inflation is unacceptable” doesn’t follow from this data. Will is assuming that people are reacting to current inflation levels. More likely, they think that prices are too high relative to what they remember.
They may well complain about inflation at 0% because they want it negative.
Bumpin’ that youtu.be/-mY4CRzIKag?...
I’m so used to weird right wing things in other countries being funded by the US government that I’m surprised but not shocked when it happens in reverse.
That the Pope should “stick to morality” reveals a conception of morality (that Vance shares with “realists”) according to which states bombing and burning civilians is not a matter for morality.
Semi realistic life goal
Toad sits in a chair, cramming cookies into his mouth. From "Cookies" In *Frog and Toad Together*
“We must stop eating!” cried Toad as he ate another.
You’ve gotta do cross country studies at this point, the n is just too small.
My fear is that “It’s the price level” shows up there too, at least once you look beyond real hyper inflation cases.
