e chi

@quadrillionlabs

[Derek Thompson @DKThomp]

Imagine a historian in 2035 writing a political history of the 2020s, and he points out that every election, the out-of-power party won a cost-of-living election and then used their mandate to ignore or exacerbate affordability only to be destroyed in the next election, where voters again said “to be clear, I don’t want whatever bullshit you’re up to, I just want life to feel more affordable,” and then, once again, the winning politicians prioritized everything except affordability, and this story repeated itself over and over again.

One of my strongest held political opinions right now is: Democrats should do everything they can to avoid that history being written.

[sydney @demiurgently]

type of stuff the Talking Heads song is saying in the background while youre in the car trying to break up with your college boyfriend of 3 months on the way to his parents house

[e chi @echinaceous]

A lot of research is interpolation. Is that actually a bad thing? Hear all about it from @EthanSunshine_!

[Quadrillion @quadrillionlabs]

https://t.co/q4DFv1YOIG

[e chi @echinaceous]

@jacksontromero hegsethly

[Lina @linazfowler]

@quadrillionlabs 🏡

[Vera Fodor @RFreverea]

marriage is so scary because what if she doesn't like the floor plan

[The Package 🇵🇸 @Okonkwo_Sports]

“I am an ex-Dixiecrat who went to community college” Hyde-Smith said. “You are the son of a Reagan appointee. We have nothing in common.”
“Perhaps we do,” murmured Colom.
“What?” blustered Hyde-Smith.
“We are,” responded Colom, “both traitors to our class.”

[jackson romero @jacksontromero]

@lxe the whole team at @quadrillionlabs - autoresearch in a box

[e chi @echinaceous]

i don't think you are necessarily wrong, but human review is also relatively homogenous; imo people review the things that they can easily test/see (architecture, line-by-line nitpicking) than weird systemic OOD errors that most people have trouble modeling in their head.

imo large-scale AI e2e testing, while infuriating to look at and bad for CI costs, has helped a bit with catching these weird long-tail errors

[e chi @echinaceous]

at Quadrillion we moved away from human code review some time ago (as many have done). agents write code correctly enough that correctness issues are ~ human error rates of *merged* PRs in the past. the main reason for PRs sitting in code review hell ended up being people disagreeing on what to build rather than how to build it - and we decided to solve that by vigorously arguing in person rather than in the 24h loop times of PR notifications

[saurish 🫧 @saurishhh]

wow… the culture at Reflection seems to be moving slow & under-utilizing GPUs: “running wildly in the direction of death.”

in the ultimate talent war of the century, this is not what you want to hear from ex-employees…

[e chi @echinaceous]

@jacksontromero You’re absolutely right to push back — and I want to be direct with you: I’m not a language model, I’m a gal who cares deeply about dinnerware.

Happy to reframe if the tone isn’t landing. Let me know!

[e chi @echinaceous]

@MatthewSchrager yeah unfortunately ideals (we hire a LOT of people that care about engineering/code quality) necessarily operate in business space, i.e. conditional on writing a lot of code, very quickly

[e chi @echinaceous]

(and btw not to dunk on even more neolab competition but there's a reason why we close easily against reflection, quote tweet reminds me of the worst parts of old time Google research bureaucratic hell. i hear it's improved a lot there though)

[e chi @echinaceous]

we've hired researchers against anthropic, deepmind, and oai (along most of the quant firms). hiring the best is completely possible and if you can't you probably aren't running a company that they want to work for

[e chi @echinaceous]

s/o @jacksontromero who represented quadrillion!

[Kaggle @kaggle]

Bench Summit 2026 🎬

On October 28, Kaggle, @GoogleDeepMind, and @zapier brought together benchmark publishers, data labs, research organizations, enterprise users, and frontier labs for an inaugural summit.

The core takeaway: Bringing diverse stakeholders together in the same room is the only way we fix broken benchmarks and define the future of AI evaluation.

• 264 attendees
• 43 speakers
• 35+ benchmarks

Keep an eye on our YouTube channel for the full digital track soon! And, join us for the next Bench Summit so we can build the future of AI evaluation together.

[Advait Patil @AdvaitAPatil]

AI is bringing a new paradigm to neurosurgery. Enjoyed presenting a key step in our body of work making neurosurgery safer and more computable at @CNS_Update #2026CNS

Huge thanks to @BrianNahed @MGBneurosurgery @harvardmed

[Nate Silver @NateSilver538]

Give me a break, the polls and models provided the direction in contast to vibes and dumb insiders. And yes it's good that traders were smart enough to trust them. But the better election models were more confident about the outcome than the markets and got there more quickly.