I don’t really agree with that post, both in that it’s still not a nice workflow to have to debug weird latex issues with an LLM, and also that the knowledge cutoff isn’t that big of a deal in practice. LLMs definitely lower the barrier to try something new for me.

Nice and round!

What does “stochastic parrot” mean to you that disagrees with those points?

Love of power vs hatred of assumptions 😄

(the answer was more drugs)

The drugs to control the side effects (a lot of nausea in particular) also work remarkably well, apparently it used to be a much worse patient experience. For me the infusion was totally smooth, but now it feels like the various drugs are doing some skirmishing in my stomach. But not terrible.

Well the chemo definitely works, untreated CHL has like a 6% 5-year survival rate and 0% 10-year one, and chemo can take that to 97% or higher for early stage. Pretty amazing.

I think it does since that restricts it to old Makie, and a newer Makie is installed, so the resolver prefers an old version of the new package rather than downgrading an existing package

it looks like Tidier.jl only declares compat with TidierPlots 0.9 (github.com/TidierOrg/Ti...)

The Egret is feeling Great about it though!

2023*

Re-meme coin market caps, they are likely exaggerated a lot: blog.mollywhite.net/cryptocurren.... You probably know that but it might be a good nuance to convey still in the newsletter

Our world in data uses 2015–2019 for the baseline, and has a very clear section on how excess mortality is measured:

ourworldindata.org/excess-morta...