x-article-ideas

# X Article Ideas Start with the author's experience and the audience's problem. Read supplied material before choosing an angle. If the author has already chosen a thesis, investigate it rather than replacing it with a generic trend list. ## Investigate - Separate what the author built or observed from what they read and what they propose. - Look for specific unanswered questions, disagreements, failed approaches, or applications that existing explanations overlook. - Use current primary sources for changing technical claims. Perplexity and Deep Research can help discovery when available; ordinary search and supplied papers also work. Read the relevant paper or documentation before claiming it supports the argument. - Find disconfirming evidence. State whether a source demonstrates a result, suggests it, or leaves it untested. - If asked to inspect past sessions, search only authorized history, distinguish actual user messages from injected prompts or quoted source text, and keep private transcripts out of the publication draft. ## Choose an argument Offer a small set of meaningfully different angles when useful. For each, give the reader's question, the original contribution, supporting evidence, and the main uncertainty. Explain the recommended choice. A novel connection should produce a useful implication; a collection of interesting links is not sufficient. Use broad, understandable language for the entry point. Keep the concrete mechanism and limitations in the outline. Avoid treating buzzwords or a new product release as an original thesis by themselves. Return ideas when ideas were requested. Draft the article only when requested or part of the active task. Avoid implying that predicted reach is measured performance. See [research prompts](references/prompts.md) for reusable briefs and a source-note format.