Just to nitpick the math. If you are going to fire 50% of the company, the AI tools should actually make the remaining people 100% more efficient, not 50% :)
it has been pretty much a benchmark for memorization for a while. there is a paper on the subject somewhere.
swe bench pro public is newer, but its not live, so it will get slowly memorized as well. the private dataset is more interesting, as are the results there:
"We've deployed trillions of tokens across these agents toward a single goal. The system isn't perfectly efficient, but it's far more effective than we expected."
if it’s too hard for you to write, it’s too hard for you to understand and comprehend. how are you going to take responsibility for that code and maintain it if needed?
At this point, its 1.5mlocs without the vendored crates (so basically excluding the js engine etc). If you compare that to Servo/Ladybird which are 300k locs each and actually happen to work, agents do love slinging slop.
And there is the thing about the cost. The blog post says that they've spent trillions (plural!) of tokens on that experiment.
Looking at OAI API pricing, 5.2 Codex is $14 per 1 million output tokens. Which makes cool $14m for 1 trillion tokens (multiplied by whatever the plural is). For something that "kind of works".
Its a nice ad for OAI and Anysphere, but maybe next time - just donate the money to a browser team?
The only thing that I got to actually run on WSL2 was the "Excel" (couldnt get anything actually to compile on Mac or Windows).
It a broken mess that probably implements 0.00001% of Excel. And its 1.2m locs.
With codebases developed in this way - either they need to figure out how agents are going to maintain them (in which case SWE as we know is dead - it will only be limited to those that can spend trillions of tokens, or they are going to remain weird demos.
error: could not compile `fastrender` (lib) due to 34 previous errors; 94 warnings emitted
I guess probably at some point, something compiled, but cba to try to find that commit. I guess they should've left it in a better state before doing that blog post.
Well, for some reason it doesnt let me respond to the child comments :(
The problem (which should be obvious) is that with a/b real you cant construct an exhaustive input/output set. The test case can just prove the presence of a bug, but not its absence.
Another category of problems that you cant just test and have to prove is concurrency problems.
The models are gotten very good, but I rather have an obviously broken pile of crap that I can spot immediately, than something that is deep fried with RL to always succeed, but has subtle problems that someone will lgtm :( I guess its not much different with human written code, but the models seem to have weirdly inhuman failures - like, you would just skim some code, cause you just cant believe that anyone can do it wrong, and it turns out to be.