This Is the Second Half of the Previous Piece

A couple of days ago I wrote “350,000 Yuan Over Six Years: How Should We Judge the Cost of Studying Mo Yan?”. That one was an opinion piece. After publishing it, I felt uneasy: opinions are one thing, but were the facts complete? What exactly were all those people arguing about in the comments?

So over the past two days I did some grunt work: I dug through the project’s background in detail, then scraped 371 answers under that Zhihu question out of a total of 800 — 46%, which is statistically enough — and analyzed them one by one.

This piece is the result. Facts first, data second.

First, Let’s Put the Project Facts on the Table

I won’t repeat the viral version. Here is what I found:

• Project number 16AZW016, a key project of the National Social Science Fund, led by Cheng Guangwei, professor at the School of Liberal Arts, Renmin University.
• Approved in June 2016, concluded in February 2022, with a final evaluation grade of “Excellent.”
• Outputs: 10 papers published in core journals such as Literary Cartels and New Literary History Materials, plus one monograph.
• The 350,000 yuan was not special treatment for this project — 350,000 yuan is the nationwide standard amount for key projects. General projects and youth projects receive 200,000 yuan. Spread over six years, that is less than 60,000 yuan a year, covering research trips, travel, materials, conferences, and publication.

In other words: the project is real, the amount is real, but “350,000 yuan” is not abnormal at all, and the project did produce results. Not many of the people arguing knew this — the data below will show that.

A Quick Rumor Check: Mo Yan Never Denounced This Project

One of the most widely circulated “bombshells” was that “Mo Yan himself called it a ‘wasteful bootlicking project’ in an interview.”

I checked it: false. That line came from the headline of a Tencent News column published on August 1, under the byline “Literary Privacy.” The author was making a subjective inference by borrowing a fictional character from Mo Yan’s novels. As it got passed around, a column headline turned into “Mo Yan’s own words.” Neither Mo Yan nor Cheng Guangwei has publicly responded so far.

That in itself is rather ironic: in a public controversy centered on “the truthfulness of information,” the most widely spread bombshell turned out to be fake.

Every Clause in That Title Was Designed

The original Zhihu question was: “A Renmin University professor was exposed as having led a key National Social Science Fund project, spending 350,000 yuan over six years to investigate Mo Yan’s family background. Is this research meaningful? Should taxpayers’ money be spent this way?”

It is worth looking at line by line:

• “was exposed” — implying that something hidden had been uncovered. But the project was public from approval to completion, sitting right there on the official website.
• “spending 350,000 yuan over six years” — time multiplied by money, maximum impact. No mention that this was the standard amount.
• “investigate Mo Yan’s family background” — turning literary-historical research into “checking household registration” or “compiling a family tree,” instantly making it sound absurd.
• “Is this research meaningful?” — the first question, with a negative answer already baked in.
• “Should taxpayers’ money be spent this way?” — the second question, directly elevating an academic issue into a confrontation between “the public vs. the elite.”

In communication studies, this is called a loaded question, a compound form of framing. Before you even begin to answer, the frame has already answered for you. Later, when the famous math teacher Tang Jiafeng reposted it, he used the same rhetoric: “using state money to study Mo Yan’s genealogy.” The frame was copied very successfully.

The Data: Public Opinion Looks One-Sided, but It Was Really Propped Up by Two Viral Answers

The question now has 2.22 million views, 800 answers, and 1,564 followers. In my sample of 371 answers, I classified stances using two methods:

The second measure reflects the “public opinion readers actually saw” — answers with more upvotes get more exposure. The gap between the two measures is startling: by headcount, negative answers were only about 30%, while neutral answers were the largest group; but in what readers actually saw, negative views made up two-thirds.

Breaking it down reveals why. Negative answers received 4,844 upvotes in total, of which two viral answers accounted for 4,659 — 96%: one was by “Jia Ren Li Da Hua,” with 3,153 upvotes — a Chinese language department perspective that first verified the project number before criticizing it, so it was criticism grounded in facts; the other was by “Wen Yi Tian Xia,” with 1,506 upvotes — “carving flowers on shit,” pure emotional venting. The remaining 116 negative answers combined received only 185 upvotes.

So the so-called “one-sided consensus” was not hundreds of people reaching agreement. It was two answers being pushed into everyone’s face. In the long tail, the 300-plus other answers were mostly neutral explanations plus a group of defenses — “Is Lu Xun’s family background worth studying?” “Only 350,000 yuan over six years? That’s great value.” “Chinese majors really have it rough; now they have to be cyberbullied too.”

Timeline: 21 Hours of Silence, Then a 5-Hour Explosion

The time distribution of answers is even more interesting:

1
2
3
4
5
6
7

08-05 16:56  Question created (almost no answers)
             ↓ 21 hours of silence
08-06 14:00  29 answers/hour   ← sudden ignition
08-06 15:00  60 answers/hour   ← peak
08-06 16-18  ~40 answers/hour
08-06 19:00  decay begins
08-07 onward only a trickle per hour

The question sat there for more than a day with almost no attention, then suddenly ignited at 2 p.m. on August 6. Within five hours it generated 215 answers, then decayed exponentially. This is not the curve of organic fermentation; it was ignited by external media amplification.

But Being Ignited Does Not Mean It Burned Out of Control

This was the most surprising thing I found. With a title this toxic and a breakout this sudden, you would expect the comment section to become a struggle session. In reality, it did not:

• 57.3% of answers were calm and reasoned; intense abuse accounted for only 2.4%, and personal attacks for 7% — the experiential threshold for cyberbullying is around 10%.
• About 22% directly repeated the title’s framing — “taxpayers,” “hard-earned money,” “investigating family background” — without adding any information of their own. This portion was indeed stirred up.
• But only 8–15% mentioned verifiable facts such as the project number, principal investigator, or final evaluation grade — most people really did make their judgment based on the title.
• Among the top third of answers by upvotes, negative answers made up 31% and positive ones 26%, a fairly balanced distribution. The most visible positions went to fact-based criticism and defense, not pure venting.
• Over time, negative answers declined from 34% to 29%, while positive answers rose from 12% to 16% — corrective information was gradually entering the discussion.

So my conclusion is: the statement “public opinion is easily incited” is only half right. At the “ignition” level, it holds true — 21 hours of silence, a 5-hour explosion, 22% of people repeating the frame, and public opinion being pushed to 66% negative by visibility. But at the level of “burning into a monolithic echo chamber,” that did not happen: emotions were restrained, stances were diverse, and the upvote mechanism gave prominent positions to answers with facts.

The title was an efficient igniter, but the community’s upvote culture acted as a flame-retardant layer.

A Frank Note on Limitations

I scraped the data with scripts; the classification was done by an LLM and calibrated through manual spot checks. There are several caveats: I only captured 46% of the answers because of Zhihu’s anti-scraping barriers, and some scattered later answers after the heat faded were not collected; the Zhihu interface only provides truncated answer text, with a median length of 56 characters, so the proportion of answers “mentioning facts” is underestimated — the real figure is probably around 12–15%; people willing to click in and answer are already more opinionated than passersby. These all affect precision, but core conclusions such as “96% of the attention was concentrated in two viral answers” and “the two measurement methods diverge sharply” are unlikely to be overturned by errors of this size.

Finally

In the previous piece, I said criticism should be aimed at the mechanism, not vented at a headline. This round of data gives that sentence a more concrete version:

The next time you see a headline that makes your blood pressure rise, do one thing first: look up the project number. All National Social Science Fund projects are publicly searchable. After you check, then decide whether to be angry — chances are, you will find that the anger worth having is not the same anger the headline wanted you to feel.