I think I should turn off my refferal flywheel while I get a handle on what I have right? Or NO because I want lots of users who might destroy me but maybe not?
The jump from "we can compute this" to "we computed it for the
entire planet" is what makes this impressive. A lot of projects
stop at the proof of concept. Running hundreds of AMD Turin cores
for two days straight to actually finish the job is a different
kind of commitment. Curious about the edge cases, how do you handle atmospheric
refraction at those distances? At 530km the curvature of the
earth and atmospheric bending would meaningfully affect whether
a line of sight is actually visible to a human observer versus
just geometrically unobstructed.
I've been using Claude Code heavily for the past few weeks on a
production project. Opus 4.6 is noticeably more capable than what
I was using before, longer autonomous runs, better contextual
awareness across files, fewer hallucinated edits. The UX changes I'm less sure about. The progressive disclosure thing makes sense in theory but sometimes I want to see exactly what it's doing without clicking through. The terminal is where I work, don't hide things from me.
Boris's response here is the right move though. Acknowledging the
miss and committing to a fix in the next release is how you build
trust with a dev audience.