As an archivist, I can only rely on publicly available information; but if that information is incomplete—or worse, incorrect—then everything that follows is utter bullshit.
In July 2026, catastrophic flooding struck Guangxi.
One of the worst disasters followed the failure of the Liulan Reservoir in Hengzhou, Nanning. On July 6, after extreme rainfall, the reservoir overtopped and its dam breached, sending floodwater into communities downstream.
The casualty figures changed dramatically as access was restored.
On July 9, authorities said the Liulan Reservoir flood had caused 26 deaths and seven missing persons.
That figure was not final.
On **August 21**, Nanning authorities said the city’s floods had caused **136 deaths and 10 people missing**. Of those, **107 deaths and seven missing persons were attributed collectively to flooding caused by breaches at Liulan, Yunbiao and other reservoirs**.
That later figure matters. It shows how incomplete the earliest casualty counts were.
Authorities themselves said damaged roads, electricity and communications initially made some areas inaccessible. China’s communications regulator also confirmed transmission-line failures, widespread power outages and interruptions to the public communications network in the Liulan disaster area.
This confirms that communications were genuinely cut off in affected areas.
It does **not**, by itself, prove that authorities intentionally shut down phone or internet service.
There is, however, evidence of another kind of information restriction.
China Digital Times has archived several articles about the disaster that disappeared from Chinese platforms. Most notably, a lengthy *Southern Weekly* investigation based on interviews with the Liulan Reservoir station chief was removed after publication. The report raised questions about malfunctioning gates, outdated monitoring, previous reinforcement work, communication failures and the reservoir’s emergency-management system.
So it would be inaccurate to say that there is no evidence of censorship.
There **is evidence that some critical reporting about the disaster was deleted**.
What that evidence does not establish is the scale of any broader suppression, or that official casualty numbers were deliberately falsified.
I also have relatives in the affected area.
According to what they told me, the disaster they witnessed was far worse than the public picture, and they believe the real death toll may have been enormously higher. They also described communications disappearing and information from the area becoming extremely difficult to circulate.
I take those accounts seriously because they come from people who were there.
But I cannot present claims of **thousands or tens of thousands of deaths** as established fact without independent evidence.
Their testimony should be preserved, not converted into certainty.
That distinction matters.
On July 24, the State Council established a special investigation and assessment team for the Liulan Reservoir disaster. It was instructed to reconstruct the entire event, determine the causes and examine problems and responsibility throughout disaster prevention, response and emergency handling.
As of **September 25, 2026**, I have not found a publicly released final investigation report.
Several basic questions therefore remain unanswered.
When did officials know the reservoir could no longer be controlled by ordinary discharge?
Why, after the water level had already exceeded the design flood level at 2:30 a.m., was full opening of the gates scheduled for 6 a.m.? Public reports show that the reservoir was already under extreme pressure during those hours.
When was each downstream community told to evacuate?
How many residents actually received the warning?
How much time did they have?
What condition were the dam, gates and monitoring systems in before the flood?
Why did casualty figures rise so sharply after the first announcements?
And what happened to the seven people who were still officially listed as missing on August 21?
These questions are not conspiracy theories.
They are questions that records can answer.
Reservoir logs.
Phone records.
Warning messages.
Gate-operation records.
Water-level measurements.
Emergency orders.
Village broadcasts.
Photographs.
Videos.
Witness testimony.
And the final investigation report.
People who survived this disaster deserve to see that evidence.
People who died deserve more than an early casualty number that was later quietly overtaken by a much larger one.
Relief supplies, rescue photographs and rebuilding projects are part of the story.
They are not the end of it.
There is nothing wrong with showing recovery.
But recovery should not replace accountability, and images of life returning to normal should not erase the people who never returned.
The latest publicly confirmed numbers are already grave enough:
**136 dead and 10 missing in Nanning, including 107 dead and seven missing in floods associated with the Liulan, Yunbiao and other reservoir breaches.**
If evidence eventually shows that the toll was higher, the record must change.
If firsthand claims cannot be substantiated, that should also be said clearly.
The answer is not to automatically believe every rumor.
Nor is it to automatically accept every official statement.
The answer is to preserve the evidence and publish enough of it for the claims to be tested.
The water has gone down.
The investigation is still unfinished.
The people who died should not disappear with the flood.
Why is it so difficult to release the list of the deceased? Is it a fear that the figures won’t add up, or a fear that the families of the victims might unite? Will the true situation only be revealed under the next administration—serving, in the process, to highlight the errors of the previous one?
# 洪水退去了，问题没有
作为一个档案记录员，我只能找公开资料，可是如果公开的资料不完整甚至错误，那么接下来的全都是放屁。
2026年7月，广西遭遇严重洪灾。
其中最严重的灾情之一，发生在南宁市横州市六蓝水库下游。
7月6日，在连续极端降雨后，六蓝水库漫顶并发生溃坝，洪水冲向下游村庄。
最初公布的伤亡数字并不是最终数字。
7月9日，官方曾公布六蓝水库洪灾造成 **26人死亡、7人失联**。
后来，这个数字大幅上升。
8月21日，南宁市官方通报称，全市洪灾共造成 **136人死亡、10人失联**。
其中，**六蓝、云表等水库溃坝洪灾共造成107人死亡、7人失联**。
这说明，灾害最初几天公布的伤亡数字远远不完整。
受灾地区的通信中断也是真实发生过的。
有关部门公开资料显示，灾区曾出现传输线路中断、大面积停电和公网通信中断。
但这些资料目前只能证明灾害造成了通信瘫痪，**不能证明有关部门主动人为“掐断信号”。**
另一方面，信息被删除的问题也并非完全没有证据。
洪灾之后，一些调查性报道和相关文章曾从中国互联网平台消失。其中包括媒体对六蓝水库运行、闸门、监测设施、此前维修工程以及应急处置等问题的调查报道。
因此，说“完全没有删帖删报道”也并不准确。
至少可以确认：
**确实有部分涉及这场灾害的重要报道后来被删除。**
但现有证据仍不足以证明伤亡数字被系统性隐瞒，更不能直接证明真实死亡人数达到“上千”甚至“上万”。
我自己也有亲属在受灾地区。
按照他们告诉我的情况，他们亲眼经历和听到的灾情，远比公开报道中的画面严重。
他们认为真实死亡人数可能远远高于官方公布数字，也告诉我，当地一度失去通信，一些灾情信息很难继续传播。
这些话来自真正身处灾区的人，我不会因为它们与官方叙述不一致就把它们抹掉。
但我也不能仅仅因为这些话来自受灾者，就把“上千、上万死亡”写成已经证实的事实。
**亲历者证言应该被保存。**
然后，再与照片、视频、时间戳、通信记录、人员名单和官方档案进行核对。
这才是真正的调查。
7月24日，国务院成立了六蓝水库溃坝灾害调查评估组。
调查范围不仅包括极端天气和水库溃坝原因，还包括灾害预防、应急响应和处置过程中存在的问题与责任。
截至 **2026年9月25日**，我仍没有找到已经向社会公开的最终调查报告。
因此，一些最基本的问题依然没有答案：
什么时候已经明确知道普通泄洪无法控制水库？
每一个下游村庄是什么时候收到撤离通知的？
通知是通过什么方式发出的？
到底有多少居民真正收到？
他们究竟有多少时间逃生？
溃坝前，大坝、闸门和监测系统究竟处于什么状态？
为什么最初公布的死亡数字后来上升得如此之快？
截至8月21日仍然失联的7个人，后来怎么样了？
这些并不是阴谋论。
这些都是可以通过证据回答的问题。
水库运行记录。
闸门操作记录。
水位和入库流量数据。
电话记录。
短信和预警记录。
村广播记录。
应急会议记录。
照片。
视频。
幸存者证言。
以及最终的调查报告。
这些资料应该被公开。
因为失去家园的人，需要的不只是慰问。
死去的人，也不应该最后只剩下一个不断修改的伤亡数字。
救援照片、救灾物资和灾后重建当然是这场灾难的一部分。
但它们不是故事的结尾。
灾区重新恢复生活没有错。
可“岁月静好”的画面，不应该覆盖那些再也回不来的人。
截至目前，官方已经确认的数字本身就足够严重：
**南宁市136人死亡、10人失联。**
其中：
**六蓝、云表等水库溃坝洪灾造成107人死亡、7人失联。**
如果未来证据证明真实数字更高，记录就应该被修改。
如果一些民间说法最终无法得到证据支持，也应该明确说明。
真正应该坚持的，不是自动相信每一种传言。
也不是自动接受每一份官方通报。
而是：
**把证据留下来。**
让亲历者说话。
让档案公开。
让每一个数字都有来源。
让每一个失踪的人都有最终下落。
让每一个死亡的人都不只是统计表上的一个数字。
洪水已经退去了。
但调查还没有结束。
那些死去的人，不应该随着洪水一起消失。
为什么死者名单公布就这么难，难道是怕数字对不上，还是怕死者家属团结起来。真实的情况是不是要等到下一个政权才会公开，顺便体现前一个政权的错误性。