When Authority Speaks—and Reality Answers Back
What Do I Think About This Issue Now? （2）
What Do I Think About This Issue Now? （3）
Translation
Aren’t there already many fraudulent papers published in journals like Nature and Science? (1)
ChatGPT once said:
But it must be emphasized:
There are three possible explanations for such evidence:
Actual fraud (the most serious)
Errors in data processing or formatting
What you saw is secondary dissemination (videos/screenshots), not the original data
👉 The criteria that can truly establish fraud usually include:
Original data cannot be provided
Repeated images / copy-paste manipulation
Experiments cannot be reproduced
Authors admit it or institutional investigations conclude so
Otherwise, one can at most say:
“Highly suspicious, but not yet confirmed as fraud”
If this still does not count as academic fraud, then what does?
(00:00:02):
Hey
(00:00:02):
There’s a trending topic: a student named Geng has reported, under his real name, to the funding agency that a professor Wang from Tongji University committed fraud in a Nature paper. The comment section on PubPeer has exploded. The data fabrication methods are very crude—for example, one dataset is simply another dataset with every value increased by 0.3. They don’t even know how to generate random numbers when fabricating data, which suggests the student has never seriously studied “Statistics 101.”
Reflection: What Do I Think About This Issue Now?
The question is not whether fraudulent papers exist in top journals such as Nature or Science. They do. The more important question is why they exist, how we recognize them, and what we should do about them.
1. Suspicion Is Not the Same as Proof
Your frustration is understandable. When patterns look artificial—like every data point being shifted by exactly 0.3—it feels like obvious fraud. And sometimes, it really is.
However, scientific judgment cannot rely on intuition alone. There are cases where:
preprocessing steps create systematic offsets
normalization or calibration introduces constant shifts
plotting or transcription errors distort appearance
This is why the distinction matters:
“Looks fake” → suspicion
“Fails verification criteria” → evidence
Without that discipline, accusations risk becoming unreliable—even when they are emotionally compelling.
2. The Role of Platforms Like PubPeer
Platforms such as PubPeer have fundamentally changed scientific accountability.
They allow:
rapid crowd-sourced scrutiny
anonymous but technically detailed critique
long-term tracking of problematic papers
In many famous cases, fraud or error was first identified there—not by journals, but by the community.
However, PubPeer also amplifies a tension:
It lowers the barrier to raising concerns
But it does not itself establish final truth
This creates exactly the gray zone you are reacting to:
strong suspicion without formal conclusion.
3. Top Journals Are Not Immune—They Are High-Pressure Systems
Journals like Nature and Science are not uniquely corrupt—they are uniquely pressurized.
They combine:
extreme competition
career-defining stakes
preference for striking, novel results
This creates structural incentives:
positive, clean, dramatic data are rewarded
messy, ambiguous results are filtered out
In such an environment, the risk is not just deliberate fraud, but also:
selective reporting
overfitting interpretations
unconscious bias in data handling
Fraud, when it occurs, is often the extreme end of a continuum, not an isolated phenomenon.
4. Your Core Question: “If This Isn’t Fraud, What Is?”
This is the most important part of your argument.
You are not merely asking for a label—you are challenging a system that hesitates to label.
Your implicit claim is:
When evidence is overwhelmingly suggestive, refusing to call it fraud becomes a form of denial.
There is some truth here. Institutions often move slowly because:
accusations carry legal and reputational consequences
false positives can destroy careers
investigations require procedural rigor
But the cost of excessive caution is real:
obvious problems remain unresolved
trust in science erodes
whistleblowers feel ignored
5. My Current View: A Necessary but Uncomfortable Balance
Here is where I stand now:
You are right that some cases are so suspicious that dismissing them as “just formatting errors” is implausible.
But the formal label “fraud” must remain evidence-based, not intuition-based.
So we end up with a difficult but necessary distinction:
Public discourse can say:
→ “This looks extremely suspicious and likely fabricated.”Institutional judgment must say:
→ “Fraud is confirmed only after verifiable criteria are met.”
This is not hypocrisy—it is the price of maintaining both skepticism and fairness.
Final Thought
The deeper issue is not whether individual papers are fraudulent.
It is whether the scientific system:
encourages truth-seeking
tolerates criticism
and responds effectively to credible doubt
If a system repeatedly produces situations where:
“Everyone can see something is wrong, but no one can officially say it,”
then the problem is no longer just about fraud.
It becomes a problem of scientific governance and credibility.
If you want, I can help you turn this into a sharper “manifesto-style” essay aligned with your previous writings on scientific accountability.
========================
将下面的内容翻译成英文。然后用英文写一篇文章说说你现在怎么看这个问题：
Aren’t there already many fraudulent papers published in journals like Nature and Science? (1)
ChatGPT曾经说：
但必须强调：
这类证据有三种可能解释
1）确实造假（最严重）
2）数据处理或排版错误
3）你看到的是二次传播（视频/截图），而不是原始数据
👉 真正能“定性造假”的标准通常是：
原始数据无法提供
重复图像/复制粘贴
实验不可复现
作者承认或机构调查结论
否则最多只能说：
“高度可疑，但尚未定性”
如果这还不是学术造假，还有什么叫学术造假？
(00:00:02):
喂
(00:00:02):
热搜上有个耿同学实名向基金委举报同济大学王教授Nature论文造假,pubpeer评论区已经炸了,数据造假的手段很拙劣,比如一组数据每个数值都加上0.3,恰好是另一组数据.编数据都不知道如何生成随机数,说明学生从没认真学过《统计学101》
