ooobin

Mind in the gutter, feet on the ground

[Jon Roberts @roberts170]

Here's a rank choice advocate using a rating method akin to Approval Voting to judge whether an RCV outcome was Fair. The RCV algorithm is so contrived and convoluted that no one can trust any RCV outcome, so it must be compared to a plainly reasonable method.

[ooobin @ea640x480]

@ArjunGuha Another challenge is getting maximum number of requests in flight. Easy to apply a prediction to a set of examples, but optimizers waste time because they don't give any scheduling flexibility. Should be multiplexed through handlers.

[ooobin @ea640x480]

@ArjunGuha DSPy claims to be declarative, but IMO its only true for signatures and DSPy.predict. And that's all before talking about the optimizers which would be a lot easier to read and understand with handlers.

[ooobin @ea640x480]

@yminsky Abelian sandpiles are very beautiful. Maybe some string diagrams, interaction nets, or ZX calc diagrams would fit

[ooobin @ea640x480]

@DimitrisPapail Exciting! Is it like pencil https://t.co/LVvYR2hWWZ?

[ooobin @ea640x480]

@davidad Where can I hear more about the geometric logic database idea?

[Thickles @thickles__]

You’re all wrong. The earth isn’t flat or round. It’s fucked

[ooobin @ea640x480]

@PresItamar @davidad Sycophancy as a hyperproperty violation, I love it

[Jon Roberts @roberts170]

#RankedChoiceVoting is complex.
Ranking several candidates is hard.
Filling out an RCV ballot is hard.
Determining the RCV winner is hard.
Explaining RCV results is hard.
Believing RCV results is hard!

Distrust in elections tears at the frayed fabric holding our nation together!

[ooobin @ea640x480]

@loganbartlett @edzitron @doodlestein we need frankenbricks

[ooobin @ea640x480]

@wesmckinn Have you recorded any videos of how you use it on YouTube or twitch?

[Robert O'Callahan @rocallahan]

https://t.co/3b7WalWA7Y
This is really impressive... until you get to the part about "Anthropic's obligations to Claude" and model welfare, where it gets absolutely terrifying.
There should be an absolute red line against creating AIs that we have moral obligations to.

[Patrick Collison @patrickc]

Came across this summary that pulls together some of the recent findings on the links between commonplace (including ostensibly "harmless") pathogens and long-term harm to human health. The war on infectious diseases is far from over!

[David Zhang (▲) @dazhengzhang]

@VitalikButerin Everyone who thinks aging is natural should watch this fable

https://t.co/SLE6SaLvPJ

[Climate Defiance @ClimateDefiance]

Sometimes you do need to beat a dead horse

[paris martineau @parismartineau]

my latest investigation for @ConsumerReports is based on months of reporting and 60+ lab tests of leading protein supplements

we found that most protein powders and shakes have more lead in one serving than our experts say is safe to have in a day — some by more than 10 times !

[Kyle Mistele 🏴‍☠️ @0xblacklight]

Maybe it’s humans that are misaligned