Basically every style guide I've ever read says the first one, but I just can't stop myself using the second one. The first one simply feels wrong somehow (even though I think it looks nicer!)

I'm going to record myself starting the first episode and immediately skipping to the end credits of the last episode, then I'm going to upload the footage as a WR any% speedrun attempt

Next, huge thanks to my co-authors Veronika and Sophie, the former a world-class WF expert who agreed to work with me on this (and patiently explain how WF actually works), and the latter a maths and coding whiz who helped a lot with the numerics. Check it all out here:

arxiv.org/abs/2507.21221

Some credits: First, it wouldn't be a Wigner's Friend paper without a cute cartoon showing W and F looking at stuff, and for our work I'm super grateful to the curliest of qubits himself, Jake Xuereb @curlyqubit.bsky.social, for providing gorgeous sketches, as seen above.

Next, we all have that one friend who's just a small collection of qubits, right? Well here we numerically studied what happens to the WF effects as the friend increases in size, finding a form of 'emegence of classicality' as the WF effects faded away.

We also applied our model to the fancy Extended Wigner's Friend Scenarios that everyone's talking about, likewise finding that the model obfuscates genuine effects like the famous Local Friendliness violations.

We found our model produces a new type of Wigner's Friend disagreement (or 'paradox') using coherence between the friend and the environment. But this new effect is hard to tell apart from some more mundane sources of error that arise when complicating the WF setup.

Oh and it's also the second episode in a row of mine that hinged on the definition of Shannon entropy in linguistics!

bsky.app/profile/tomr...