In recent years, discussions surrounding academic integrity, scientific reproducibility, and the mechanisms of paper production have become increasingly frequent. In 2023, the number of retracted papers worldwide exceeded 10,000 for the first time, setting a historical record, with a significant proportion linked to fake papers, peer review manipulation, and large-scale paper mills. [1] A subsequent analysis by Nature further pointed out that in a research environment that emphasizes publication counts, citations, and rankings, the pursuit of "more and faster" may simultaneously increase the risks of sloppy research, plagiarism, and data fabrication. [2] In 2026, Nature reported a more direct example: researchers systematically mapped the "academic fraud market" behind authorship buying and ghostwriting services by analyzing thousands of online advertisements. [3]
At the same time, scientific integrity itself is gradually moving from a specialized internal topic into the public eye. In 2026, Chinese video blogger "Geng Tongxue" (Geng Hongwei) publicly analyzed published papers, pointing out potential data anomalies and image duplication issues, which prompted investigations by multiple universities. Nature and Science separately covered this in special reports; Nature reported that following investigations triggered by these doubts, several senior scholars faced disciplinary actions. [4][5] A scientific integrity review process—which in the past could often only be completed through peer review, journal editors, and institutional investigations—entered the public communication space so prominently for the first time.
Platforms like Retraction Watch have long tracked retracted papers and cases of academic misconduct, further making questions like "how papers go wrong," "why errors go undetected," and "why papers are ultimately retracted" into subjects that can be publicly observed and discussed. In this context, "fraud" is no longer just a simple moral label, but is gradually becoming an analyzable process: How exactly is an unreliable study constructed? What kind of data, charts, and arguments can make a conclusion appear credible? And what role do peer review, evaluation metrics, publication pressure, and the act of academic writing itself play in all of this?
Many so-called "errors" are not isolated events; rather, they can occur jointly alongside evaluation systems, publication mechanisms, and methodological inertia. However, despite a wealth of serious research attempting to uncover the mechanisms of academic misconduct, we still lack a lighter way to confront it—one that does not take punishment or exposure as its sole objective, but instead makes these very mechanisms visible through formal "simulation" and "exaggeration."
It is precisely in this context that Fake was born.
It does not attempt to replace ethics reviews, nor does it attempt to pass judgment on the academic community. Instead, it attempts, in a manner akin to an "inversion," to treat the language, structure, and methodology of scientific papers themselves as raw material to reorganize those research practices lying between reality and fiction. In this way, we hope to make readers realize anew as they read "paper-like texts": a paper appears reliable not always because it is true, but because it conforms to a mutually accepted form of expression.
In this sense, Fake is closer to a methodological, entertaining experiment: it does not directly criticize fraud, but rather makes the structure of fraud visible by "seriously writing a fake study."
We believe that sometimes the most effective way to understand a problem is not to explain it solemnly, but to push it to a slightly absurd yet self-consistent extreme.
Fake does not require the subject of study to exist in reality, nor does it require conclusions to pass real-world experimental verification. However, we hope a good Fake satisfies at least one condition: it poses an interesting question.
This question can be absurd, everyday, or even seemingly entirely unnecessary, but ideally it should cause people to pause briefly:
"Wait, this actually seems like something that could be researched."
For example, modeling short-video consumption as an attention dynamics system; attempting to use large language models to "prove" the computability boundaries of certain mathematical problems; or treating academic writing itself as an optimization problem, studying how "publishability" varies with citation strategies and figure density.
In these examples, the key lies not in whether the problem actually exists in reality, but in whether it is sufficiently "like a problem that can be taken seriously." More importantly, once the initial, slightly absurd premise of the article is accepted, subsequent deductions, experiments, and arguments must be able to proceed self-consistently. We welcome differential equations, statistical analyses, numerical simulations, machine learning, theoretical proofs, experimental designs, field studies, and any tools the author considers "necessary in this hypothetical world."
A good Fake is rarely just "adding paper formatting to a joke"; rather, it gradually makes the reader realize during the reading process: if we truly think about the world in this manner, then the conclusion might not be so easily dismissed.
It is worth noting that Fake is not the only publication to emerge from this wave of "academic satire." In early 2026, young Chinese researchers formed a sizeable online cultural phenomenon around a series of so-called "bottom-tier journals" such as Rubbish, NoTrue, Silence, and Rubbish Communications. Nature published a dedicated report on this in May of the same year, describing it as an emotional outlet for young researchers facing academic pressure and an academic "credibility crisis." [6] The origin traced by the report was even a failed Western blot experiment image: a result that should have been discarded ultimately became the catalyst for founding a satirical journal.
The emergence of these publications is fascinating in itself.
It signifies that young researchers have not entirely abandoned the scientific paper as a form of expression. On the contrary, they are so familiar with the Abstract, Methods, Figures, peer review, impact factors, and journal tiers that they can accurately replicate them, and then turn them all upside down.
Fake shares part of the same cultural background as these "bottom-tier journals," but our positioning is slightly different.
We do not take "nonsense" itself as our goal, nor do we take "humor" as our primary evaluation criterion. More importantly, we focus on how the act of fraud can occur on a structural level.
In other words, what we care about is not "whether there is fraud," but rather:
How was this "fake" constructed?
Which real-world research mechanisms (such as metrics, benchmarks, statistical significance, or peer review) did it rely on?
Can an obviously wrong conclusion be manufactured step-by-step into "credible science" through a series of locally reasonable operations?
Does it, paradoxically, reveal how "truth" is generated in the contemporary scientific context?
Therefore, Fake is not a negation of science, nor is it a simple parody of the academic system; rather, it is a reflexive use of its methodological boundaries.
In this process, we place special emphasis on a seemingly paradoxical yet crucial principle: the existence and use of real data.
Even if the conclusions are fictitious, data should be as real and traceable as possible, or at least have its generation mechanism clearly specified. We encourage authors to proactively construct, collect, analyze, or even "manufacture" data, and to seriously discuss throughout this process:
How is data produced?
How does fraud occur?
Which steps truly alter the conclusions?
And why are certain forms of "fraud" effective within the system instead?
In this sense, Fake is closer to a methodological experiment than to pure literary fiction.
A batch of manuscripts has already entered the editorial and peer-review process, roughly illustrating the direction we anticipate.
For example, one work reformulates the classic "P = NP" problem as a machine learning benchmark task and reports that GPT-9 achieves an 87.3% accuracy on NP-Bench-1M, thereby deriving "P = NP with 87.3% confidence" and introducing theoretical frameworks such as benchmark saturation, evaluation collapse, and scaling law extrapolation.
The key to such manuscripts lies not in whether the conclusions hold, but in how they reveal a methodological tendency that is becoming increasingly prevalent: when "proof" is gradually replaced by "metric performance," how much actual space for true argumentation do we really have left?
A Fake paper may contain fabricated data, absurd hypotheses, non-existent subjects of study, or even blatantly wrong conclusions, but it must possess internal logic.
Readers do not need to believe it is real, but they should be able to understand: why this assumption was made, why this methodology was adopted, how the conclusion was derived step by step, and what structural issue it is ultimately satirizing or revealing.
We especially welcome manuscripts that make people laugh at first glance, but pause on second thought—
"Wait a minute, this actually seems to be talking about a real issue."
Fake is currently accepting new submissions on an ongoing basis. Whether you come from physics, mathematics, computer science, biology, medicine, engineering, social sciences, or have no academic background at all, as long as you have a problem worth turning into a "paper," we welcome your submission.
It can be a complete study or a serious thought experiment; it can discuss grand scientific questions or highly microscopic everyday phenomena—such as the statistical distribution of group meeting durations, the relationship between caffeine intake and paper revision efficiency, or how a project filename ultimately evolves into final_v7_revised_REAL_final.pdf.
Science provides methods for understanding the world, and Fake sometimes simply tries to slightly pivot these methods toward places that were not originally taken seriously.
Fake is now accepting submissions.
If you have an absurd question, please research it seriously.