Egregoros

@mischievoustomato i did run in to hella platform bugs in general (ollama's streaming output is broken when tool use is enabled)

small qwen models did work for some basic things (summarize political garbage.) when i tried asking it for json output though it would give me valid json but the answers weren't very useful.

i'm still curious to see this supposed thousand x programmer.

@icedquinn @mischievoustomato A 1000x programmer is a midwit who can create technical debt, maintenance burdens, and nightmarish monstrosities 1000x faster than a normal programmer

I have a former coworker who would've been a 1000x programmer if he'd been allowed to use LLM agents, as is he still created some truly horrific shit

@icedquinn @mischievoustomato He's at l3harris now where he's presumably working valiantly behind the scenes to increase the veteran suicide rate by producing code so bad you'd rather die than use it

@scathach @icedquinn @mischievoustomato the winning strategy, which l3harris has figured out, is to establish a monopoly and then charge extra money for "client customizations" when someone "discovers" an "inconsistency" in your shitty code and documentation