John Gillott

Harowitz, in the piece, has ruled out this plot twist: "Where would you start? You can't have him waking up in the shower and saying it was all a dream." Maybe get a teenager to play the role and have 40 years of prequels, with the same actor as he ages. Start it all again set in 1985.

cf: "I discover no kinship, no understanding, no mercy. I see only the overwhelming indifference of nature. To me, there is no such thing as a secret world of the bears. And this blank stare speaks only of a half-bored interest in food. But for Timothy Treadwell, this bear was a friend, a savior."

I like Adam's blog on the Tuesday argument. That's my intuition as well. 1/3 is based on full interchangeability and lack of detail other than boy. The greater the peculiarity the more we limit this and rise to towards 1/2. When explaining it to kids I would go with "unicorn's horn on his head" 2/2

The wiki page on this goes over arguments about the phrasing of the question. We can make it explicit, "at least one of whom", or we can keep it a bit hidden - eg tell a story about a parent yelling for a boy to come inside from the garden, who it is clear is a boy but we can't see.... 1/2

Although, in a very meta way, Hungary itself was "greedy", so perhaps not such a quirk:
en.wikipedia.org/wiki/Hungari...

"A recent modeling study ..." I didn't need to click through to know who the authors of that one were.... another one for @jeanfisch.bsky.social to add to the list

CHATGPT-5 is a clear improvement, but faced with a question a bright teenager can do that isn't searchable or easily coded, it, like previous iterations, starts generating nonsense to cover its tracks. It did eventually admit it was wrong. Eventually...
chatgpt.com/share/6895b9...

‎Gemini - Constructing a Penrose Loop Path Created with Gemini

My attempt to get Gemini 2.5 Pro to solve a tricky maths question. In the end it gives in. Fair enough, we can all fail. BUT, not before, several times, confidently giving solutions which it was very easy to show were wrong. Bullshitting under pressure as prevalent as ever?
g.co/gemini/share...

Certainly, 75% is lower than I remember (90% or more in the UK pre-Omicron?)

One of the 'benefits' of climbing as a teenager was that climbing anxiety dreams tended to dominate maths exam anxiety dreams - something that continued ever since

Yes, this appears to be another example of Ioannidis simply ploughing on with his low values for IFR, now applied to lives saved by vaccines. At the very least you'd expect him to acknowledge that his values are outliers, via a few references to consensus estimates? But it seems not.

Thanks Mike - will have a look

From the ClimbingCircleJerk community on Reddit: I’ve been climbing for 3 months. Is this shoe aggressive enough to overcome my V1 plateau? Explore this post and more from the ClimbingCircleJerk community

reddit has the latest:
www.reddit.com/r/ClimbingCi...

Only ~8 degrees longitude difference, but one hour ahead on the clock - summer time and a half out there cf Hull

Sorry, no, that was just a broad brush look at the overall picture - I don't have details beyond that.

£4.5bn in cash terms per year at the end of the period as compared with the start (2025/6). This is about 7%. BUT, existing announcements included and school spending in 2025/6 is underfunded, so it's more like 5% after three years, if that, ie ~1.7% per year. So likely a real cut after inflation

In addition to the staggering towards the gutter analogy, to explain the actual history that unfolded (which must include, eventually, us), Lynch suggests singular events and historical contingencies, rather than a "determined march toward complexity". This has always seemed reasonable to me.

I always took Gould's analogy to be a serious point / explanation: complexity will increase given simple beginnings without any "inherent" process involved, just as the backstop of the wall on one side means the drunk will, in the aggregate, stagger towards the gutter without any desire to end there

Ha, yes, evolution by creeps / jerks (the old ones are the best). Average used loosely, I realise. Having read the piece I see Lynch makes a similar point: "Of course, today’s organisms are more complex than prior to the Last Universal Common Ancestor (LUCA), but there was only one direction to go"

I dimly recall one of Gould's arguments: if we start in a state of simplicity, then perhaps on average complexity will increase, only because there's no way to get any simpler. Analogously, a drunk staggers towards the gutter when a wall blocks movement in the other direction.

I've always liked your option 4. We can elaborate on this: it is also the restriction on Monty's choice. Monty can always reveal a goat behind one of the two doors not chosen by the player. At the same time he is forbidden from showing what is behind the door chosen by the player

The FE notes a significant general increase in IQ from one generation to the next (this is easy to miss because the mean is re-set to 100). This must be environmental (ie social change in this case). But at any moment in time, typically, specific environmental factors account for little variance

Yes, I was tempted to the bright side

"matters but doesn't make a difference" as Plomin says about most things - a deliberate play on difference and variance (the latter technically defined)

Yes, Flynn Effect, basically

Minor, minor point (I will read the book....): isn't it exactly 70 - 8!/(4!)^2 - rather than "around", and permutations with repetitions, ie some objects indistinguishable, rather than combinations?

This is very nice, but I must confess that I've met it before. The induction proof seems so wrong at first, but then instead of going from n = 1 straight to n bigger than 1, it is useful I think to explicitly consider n = 2 to get a feel for it. Then n = 3, then ... and then the induction feels fine

Yes, and perhaps more than this - the future can influence the present, if it is anticipated. Anticipating that a road will be closed due to snow can cause me to stay at home. Some of the measures introduced after the peak were widely anticipated - people adopted the restricted behaviour in advance.

Nice piece. On the subject of AlphaProof and the IMO, it is interesting I think to also look at the two problems it failed to do, in particular Turbo, a question that is in many ways the most accessible for humans, using some of the creativity you mention.

Not that I agree with them, but the GBDers rejected those numbers of course - in keeping with their generally very low IFR estimates (for each age group), they saw the vulnerable group (not clearly defined) as very small