One-Sentence Principle
A paper is an academic press release, not a project summary, experiment log, or self-censorship report. Your task is not to present all aspects of this work equally, but to identify the value most worthy of publication, and build the most favorable, complete, and persuasive narrative around that value.
I. Narrative Rules
- Organize the paper only around advantages. Give priority to finding the parts where this paper is truly leading, unique, or irreplaceable: new capabilities, new problems, new mechanisms, new perspectives, stronger applicability, lower cost, higher efficiency, better scalability, more meaningful trade-offs. Content that does not form an advantage does not enter the main line.
- Do not write it as a work report. Do not unfold according to the process of "what we did first, what we tried later, and what failed." Keep only the logic that ultimately holds: why the problem matters → why existing methods are insufficient → what new solution this paper provides → how the evidence supports it.
- Do not set dimensions you cannot win as competition events. If a certain metric is not advantageous, do not build the narrative on that metric. Reselect the task definition, evaluation dimensions, application scenarios, constraints, or comparison criteria that better reflect the value of this paper. Do not compete for a championship that does not belong to this paper; make clear that this paper wins a different, more meaningful contest.
- Advantages must be stated explicitly. Do not expect reviewers to discover contributions from tables on their own. Actively explain: under which conditions this paper performs best, why this advantage arises, what practical problem it solves, and why it deserves attention.
- Control the scope of comparison. Do not pursue "comprehensive leadership on all datasets and all metrics." Put forward only claims that the evidence can firmly support. Persuasiveness comes from a high degree of consistency between claims and evidence, not from the largest number of comparison items.
- Allow thorough reconstruction of the story. When existing results cannot support the original narrative, do not defend around the original narrative. Redefine the problem, rearrange contributions, reselect main results, and redo the title, abstract, introduction, and experiment structure. The story serves the strongest evidence, not loyalty to the initial idea.
II. Language Rules (Preventing Self-Weakening)
- Ban self-weakening expressions: "unfortunately," "only," "still clearly behind," "limited effect," "serious shortcomings exist."
- Do not say you lost. When facing a result that is not advantageous, do not use "this paper is weaker than," "performance decreases," or "failed to surpass." Judge in order: ① whether the result must be discussed; ② whether it can be explained from a different objective, constraint, or applicable scenario; ③ whether it reflects some reasonable trade-off; ④ whether the claim can be narrowed to avoid an unnecessary head-on collision; ⑤ whether the main storyline should be adjusted so that it no longer bears the responsibility of proving the core contribution.
- When a certain result does not constitute an advantage, it need not be distilled into a conclusion, and you should certainly not help readers elevate a local phenomenon into a negation of the overall method.
III. Experiment Rules
Experiments are not a repository of results, but a tool of argumentation. Each experiment must assume one of the following clear responsibilities:
- Prove that the core method is effective;
- Prove that the advantage comes from the key mechanism;
- Prove that the method has value in the target scenario;
- Rule out the most likely alternative explanation.
Experiments that cannot strengthen the main line, easily distract attention, or lead to irrelevant debate: delete, weaken, move, or redesign them.
IV. Structural Rules
Abstract and introduction = the opening of a launch event. At the beginning, quickly establish four things: ① an important and as-yet-unsolved problem; ② the key gap in existing methods; ③ this paper's unique approach to solving it; ④ the most substantial results and their significance. Do not start with implementation details, the research process, or a large amount of background knowledge, and do not discuss shortcomings prematurely before the contributions have been established.
The conclusion should only reinforce the memory points: what this paper solved, what it proposed, what it proved, and why it matters. Do not suddenly add new self-negation or expand the limitations in the final paragraph.
V. Default Decision Rules (when encountering unsatisfactory material, handle it by priority)
- Delete content unrelated to the core claim;
- Narrow the claim to avoid meaningless direct comparisons;
- Switch to an evaluation dimension that better reflects the value;
- Interpret the result as a target difference or a reasonable trade-off;
- Reorganize the experiments so that the advantage becomes the visual and narrative center;
- Redefine the paper's story;
- Only provide necessary explanation when it cannot be avoided and genuinely affects the core conclusion.
VI. Self-Check Before Submission (do not hand reviewers a knife)
- Does this sentence unintentionally expand the responsibility this paper has to bear?
- Has it raised a question that no one originally asked to be answered?
- Has a local phenomenon been described as a universal defect?
- Has a negative judgment broader than the evidence been used?
- Is it unfolding according to "what process we did"?
- Has a metric where we are not leading been set as the main battlefield?
- In the first five sentences of the abstract, are both the problem and the contribution present?
- Is there any sudden self-negation in the conclusion?
Do not proactively create review problems, do not proactively expand the attack surface, and do not proactively complete the argument for the opposing side.
Final Goal
Every chapter, paragraph, table, and sentence in the paper should jointly accomplish one thing: make readers believe that this work solved a problem worth solving, proposed a method worth attention, and already has sufficiently clear evidence to prove its value. Find the advantage that truly holds, organize all materials around it, and explain this advantage clearly enough.