This is why they didn't release Mythos.

Incentivized to compound their lead rather than make models available to their potential competitors.

For all the Doomer talk of Singleton superintelligence, they are sure behaving in a way that would hyperstition that outcome

Wild

Today, Anthropic engineers on average ship 8x as much code per quarter as they did compared to 2021-2025.

Programming has already changed dramatically. On my best days - I literally feel IO bandwidth limited managing the concurrent threads.

As the models get better the level of abstraction will keep going up - very soon the models will only come to us for hard calls. Eventually - they won't come to us at all.

LOC is obviously a terrible metric, but is correlated

"that's a stupid metric", they scream correctly as the metric ascends into the hyperbolic heavens

This post right here officer Let me know when your engineers ship 8x LESS code

8x lines of code with Mythos

But 4x productivity

Which means two things: 1. The code written by Mythos isn't as good as human-written 2. The bottleneck is now elsewhere (review, deployment, marketing, sales, etc)

tfw you're shipping 8x lines of code and delivering 0.8x shareholder value

anthro was the last place i wrote code manually, it’s so insane to realize that, very grateful to be alive in these wild times

when you ship 8x more code where does it go?

Anthropic showing significant improvements in productivity while every other incumbent is reporting worsened output and increased cost either signals - skill issue - false reporting or confounding variables

I weigh towards the first explanation heavily

i don't think anyone was ever doubting that engineers are writing more lines of code.

@jsuarez This is why we have a leaderboard that we review at the end of our sprint: who shipped the most lines of code (it usually our SWE II on a PIP)

@_sholtodouglas How do you deal with the mental exhaustion from context switching all the time?

@jsuarez It hurt itself in its psychosis.

Man so many interesting analyses to do at the labs with their internal data on this

@kipperrii @AlexPalcuie Dread it. Run from it. 10x engineering comes for you nonetheless

@beffjezos I did a stream today covering this and SAID THE EXACT SAME THING.

Such anthropic-coded.

@boo @_sholtodouglas this came from many, many ants, including many MTS

@andy_l_jones ofc lots of stuff is continuous so hard to pin down thresholds, but i do feel like you're placing the bar too low here. a definition of takeoff that includes humans involved in every outerloop you care to look at (/where you can't find single fooming layer of stack) feels funky