Those numbers were faked. See the screenshots from the original website here:

http://archive.ph/https://news.qq.com/zt2020/page/feiyan.htm

I've seen this pattern before (Epstein, Douma, the principle of net neutrality etc.). Fake data is released to muddy the water hoping that someone seeking for the truth is accidentally going to pick it up to then discredit / censor this person. If someone will then leak the actual truth that is similar to the faked one people are then less likely to look at it even if proof is provided.

Why would Tencent be given this data in the first place?

Seems like clickbait...

In communist China, large organizations should be considered extensions of the government. Tencent is a massive media corporation, so they are likely working in lock-step with the CCP in regards to aggregation and dissemination of this information. They operate a messaging app with over a billion monthly active users. Their participation in all of this is extremely involved.

Even allowing that, why does it make sense for the government to hand them the very data they're trying to obscure? Unless you're arguing that they were somehow given the numbers in error, and then erroneously published.

It's probably Tencent doing the data collection/storage/management.

The Chinese government is going to outsource data collection and management of a disease epidemic to an internet software company, rather than their own ministry of health, who are actually trained and educated in healthcare work?

I wonder if there's any indirect way way to get mortality rates... Say number of social media accounts active, or number of webpage requests from an IP range...

I was wondering about this as well, but I'd suspect any signal would get lost in the noise. People getting sick and sleeping more, people working from home, people checking online for news about the virus, the government shutting down internet access to prevent the spread of panic, etc. are all things that would drastically change internet usage patterns, but wouldn't necessarily be indicative of anybody dying.

Not necessarily saying any of those things (e.g., shutting off internet access) have or will occur - I'm pretty ignorant of the situation on the ground. I'm just saying that from a statistical perspective, I'm afraid it'd be a needle/haystack situation.

I suspect you're right. But then, with a good, large data set, maybe something would show; once enough people get too sick to use a computer.

Maybe the number of DNS requests from distinct IP addresses goes down. As you say, traffic/data volume might keep more constant.

There are actually small differences in the two images (text color, etc.).

Could it be due to different CDNs providing outdated CSS files ?