I think
- they thought open models was going to be the move
- and we’ll build your company’s thing on top (à la databricks)
- and got turbomogged by Qwen
- and can’t stop throwing good money after bad
- because the market will kill them.
imo two competing ideas (haven't listened so its plausible that these are covered) but
- have to be able to kill your darlings. for every solyndra that blows up you have a dozen projects that limp along because nobody will kill them. if winners are gonna pay for losers you gotta kill the losers.
- and so the real problem is its not a money question, the constraint is ability to deliver administrative capacity and resist capture.
no one at meta seems genuinely committed to anything except for zuck who seems genuinely committed to the dumbest shit possible, and dont even need hindsight to flag it
hand up, i would like to see the prompt that generated this
- not because the losers are money negative (who cares), it's that there's a comprehensive aversion to looking like risks failed (that's why theyre called risks); if you cant have visible failures the entire enterprise is fucked up from the start which makes the solyndras look *even worse*
markwayne being a kratom guy is a nice little c-plot. like cricket getting ringworm on sunny
so you try to do bankshots like USDS to reboot (eg) technical capacity in house and other than getting strangled in the crib by elon it and 18f were basically doing what they were intended to do
- the bureaucracy has been ~deliberately deskilled and i dont want to get yelled at but the analysis gets farmed to the firms that are being evaluated (part viz lobbying but also viz a specific industry ;) as a symptom and cause of the decay of embedded expertise)
Because everyone on here is posting from a Built to Spill concert that has chairs. Is why