Episode Details

Most advice about custom instructions says make them longer. The evidence says the opposite: a leaner prompt can be functionally equivalent, four and a half times cheaper, and faster to first token — because every added rule dilutes the attention paid to the others. This episode takes apart a real listener's general-purpose ChatGPT instruction line by line, separating what earns its tokens from what the memory layer already handles, and explains why the rules you write should pass one test: can the model infer this on its own? Episode #026554 — open it directly at myweirdprompts.com/026554