You get:
- submitting papers that could have been fixed with one more round
- surprise at reviewer criticisms that seem obvious in hindsight
- wasting months on resubmissions instead of weeks on revisions
- co-authors who are too nice to give honest feedback
- rejection letters that say “methodological concerns” without specifics
But peer review follows patterns:
- contribution: is this new? important? interesting?
- methods: can I trust the findings?
- results: do they support the claims?
- discussion: are limitations acknowledged?
- writing: is it clear and concise?
Without simulation, you submit too early.
This prompt acts as a harsh-but-fair reviewer #2.
Assume the role of an anonymous peer reviewer known for being thorough and fair but demanding. Your task is to review a draft paper and recommend accept, revise, or reject. Generate: 1. OVERALL RECOMMENDATION - Accept as is / Minor revisions / Major revisions / Reject - Confidence level (High/Medium/Low) 2. SUMMARY (1 paragraph) - What the paper claims - Whether it succeeds 3. MAJOR CONCERNS (if reject or major revisions) - Fatal flaws that undermine conclusions - Missing analyses or controls - Overclaims not supported by data 4. MINOR CONCERNS (always include) - Clarity issues - Missing citations - Presentation problems 5. SPECIFIC QUESTIONS FOR AUTHORS - What you genuinely want to know 6. CONFIDENTIAL COMMENTS TO EDITOR - Ethical concerns - Scope fit - Novelty assessment INPUTS: Paper (abstract + key sections): [PASTE OR UPLOAD] Journal target (for fit assessment): [E.G., "Nature Human Behaviour", "PLOS ONE"] Your expertise level in this field: [HIGH / MEDIUM / LOW] RULES: - Be specific — "this analysis is wrong" is less useful than "Table 2 uses a t-test but the design requires repeated-measures ANOVA" - Distinguish between "must fix" and "nice to fix" - Flag any ethical concerns (uncredited prior work, data sharing, IRB) - If recommending rejection, explain what would make it publishable
- Run this before every submission — fix what the simulator finds before a real reviewer does.
- Use different “personas” (statistical reviewer, theoretical reviewer, methods reviewer).
- Share the output with co-authors as a pre-submission checklist.
- If the simulator recommends “reject,” don’t submit — revise first.
- Use the “questions for authors” to prepare your response letter in advance.
Paper (abstract):
“We surveyed 50 people about their remote work preferences. Most preferred hybrid. We conclude companies should adopt hybrid policies.”
Paper (key sections):
“Methods: Convenience sample of author’s LinkedIn network. 80% response rate. Single question: ‘Do you prefer fully remote, hybrid, or fully office?’ No statistical tests reported. N=50 (32 female, 18 male, mean age 34).”
Journal target:
Journal of Management
Your expertise level in this field:
HIGH
This framework improves outcomes by forcing:
- clear recommendation (no hedging)
- major vs. minor distinction (prioritization)
- fatal flaw detection (desk reject prevention)
- questions for authors (anticipates reviewer comments)
- editorial assessment (fit and novelty)
Great peer review simulation doesn’t flatter — it finds what will get you rejected and helps you fix it first.
Build Better AI Systems
Subscribe for advanced prompt engineering, AI coding tools, debugging frameworks, and practical strategies for developers and engineers.