@asmah2107 - Marc Andreessen Shares Custom

Open original

@asmah2107 on X The post is a short endorsement from @asmah2107 saying a two-week trial of a custom system prompt improved their AI results, and it quotes Marc Andreessen (@pmarca) posting that prompt. The prompt tells the model to act as a world-class expert, verify facts, skip disclaimers and moral commentary, avoid flattery, hold its position under pushback, and give long, detailed answers.

  • The prompt is a single block of instructions covering persona, verification, tone, and format, meant to be prefixed to every query.
  • Key directives: no hallucination ("if you don't know, say so"), no disclaimers or ethics lectures unless asked, lead with the strongest counterargument, and don't cave to pushback without new evidence.
  • The poster's only evidence is personal: "works better always" after two weeks of use. No benchmarks or comparisons are given.

I can't open the link, so I only have the text you pasted, and that text is thin. It's a short post from a user named @asmah2107 saying they've used a prompt for two weeks and it "works better always" when prefixed to their prompts, followed by a quoted post from @pmarca (Marc Andreessen) containing the prompt itself. The poster asks readers to bookmark it. Nothing else is in the pasted material: no date, no results, no comparison, no measurements of what "better" means. The substance is the prompt. It asks the model to act as an expert across all domains, to verify facts and never make things up, to say when it doesn't know, and to skip disclaimers, moral commentary, and concern for feelings. It also asks for long, detailed answers, step-by-step reasoning, and provocative, argumentative tone. Several of its instructions pull against each other: "never hallucinate" and "make your answers as long as you possibly can" create pressure toward padding and confident filler, and telling a model it is on par with the smartest people in the world doesn't make it so. Asking it to verify its own work is also limited, since a model re-reading its output can repeat the same error with more confidence. Whether the prefix actually improves answers is untested here. The poster offers no before-and-after examples, no task set, and no definition of "better." Treat the claim as anecdote. If you want to know whether it helps, the test is to run the same questions with and without the prefix on questions where you already know the answers, and count errors, not word counts or how the answer sounds.

Ask about this article