I have realized “just one more turn” and “just one more prompt” hit exactly the same wires in my brain
if you go to concerts and just stand there instead of dancing then you are part of the permanent underclass
feel for the chrome root CA engineer here. We really take for granted that good software engineering has such strong norms around if you fuck up you WILL work day and night to fix it and you WILL write a postmortem and you don’t let reputation management get in the way
This is what Crome saying “this is your last opportunity to fix the mess you’ve created, else you’re out.”
Pretty clear they will be out.
We’re talking about an eng team unable or unwilling to run a RCA and publish it, when everyone assumed they could
bugzilla.mozilla.org/show_bu…
obviously this is goofy and the discourse has been done to death already but when the claude code source got leaked a while back nobody noticed this
why?
thinking about early modernism again
just dudes getting absolutely oneshotted by the Industrial Revolution
it’s so earnest
great week for showing the new hires around Hawthorne
Teams complete three Falcon launches, four first stage landings at four landing zones in Florida and California, and docking Dragon to the @Space_Station for our 21st human spaceflight in under 13 hours
fun how we had “computers are ridiculously superhuman at verifiable tightly specified domains”
then went to “computers can write a sonnet about a Pb&J in a VCR but can’t count”
and now back to “computers solve anything verifiable but fail when underspecified”
For folks not understanding where we are at with AI now - if the task is verifiable, it is now solved by AI.
Yes, earlier days you could argue that 'AI isn't that good' (e.g. see first attached chart here, where GPT-4o underperformed average accountants on tasks)
But now, it's a totally different story (only ~2 years later). The second chart is astounding. It took me a while to even understand it because it looks so odd. It compares manual accounting tasks to tasks complete with Opus 5.5. Basically Opus 5.5. solves all tasks almost instantly at a 100% accuracy rate, whereas a person takes a lot more time and gets a lot of things wrong. It makes the chart look entirely silly because the axes aren't even comparable.
So yeah, this is where we're at. If it is verifiable, it is solved. This is just how these model architectures work now.
between the four tet light bar, ninajirachi, and V3 satellite deploy, it has been an absolutely fantastic 24 hours for my particular special interests teapawt.org/Starlink/status/210460…
